View FW82439TX_389099.PDF datasheet online --- IC-ON-LINE

Datasheet File OCR Text:

PRELIMINARY
om .c 4U et n n he anS at n .D n w w w
Intel Extended Temperature 430TX PCISET: 82439TX System Controller (MTXC) Datasheet
n
Fully Synchronous, Minimum Latency 30/33-MHz PCI Bus Interface Five PCI Bus Masters (including PIIX4) 10 DWord PCI-to-DRAM Read Prefetch Buffer 18 DWord PCI-DRAM Post Buffer Multi-Transaction Timer to Support Multiple Short PCI Transactions Power Management Features PCI CLKRUN# Support Dynamic Stop Clock Support Suspend to RAM (STR) Suspend to Disk (STD) Power On Suspend (POS) Internal Clock Control SDRAM and EDO Self Refresh During Suspend ACPI Support Compatible SMRAM (C_SMRAM) and Extended SMRAM (E_SMRAM) SMM Writeback Cacheable in E_SMRAM Mode up to 1 MB 3.3/5V DRAM, 3.3/5V PCI 3.3/5V Tag and 3.3/2.5 SRAM Support Test Features NAND Tree Support for all Pins Supports the Universal Serial Bus (USB)
n
The Intel 430TX PCIset (430TX) consists of the 82439TX System Controller (MTXC) and the 82371AB PCI ISA IDE Xcelerator (PIIX4). The 430TX supports both mobile and desktop architectures. The 430TX forms a Host-to-PCI bridge and provides the second level cache control and a full function 64-bit data path to main memory. The MTXC integrates the cache and main memory DRAM control functions and provides bus control to transfers between the CPU, cache, main memory, and the PCI Bus. The second level (L2) cache controller supports a writeback cache policy for cache sizes of 256 Kbytes and 512 Kbytes. Cacheless designs are also supported. The cache memory can be implemented with pipelined burst SRAMs or DRAM cache SRAMs. An external Tag RAM is used for the address tag and an internal Tag RAM for the cache line status bits. For the MTXC DRAM controller, six rows are supported for up to 256 Mbytes of main memory. The MTXC is highly integrated by including the Data Path into the same BGA chip. Using the snoop ahead feature, the MTXC allows PCI masters to achieve full PCI bandwidth. For increased system performance, the MTXC integrates posted write and read prefetch buffers. The 430TX integrates many Power Management features that enable the system to save power when the system resources become idle.
Information in this document is provided in connection with Intel products. No license, express or implied, by estoppel or otherwise, to any intellectual property rights is granted by this document or by the sale of Intel products. Except as provided in Intel's Terms and Conditions of Sale for such products, Intel assumes no liability whatsoever, and Intel disclaims any express or implied warranty, relating to sale and/or use of Intel products including liability or warranties relating to fitness for a particular purpose, merchantability, or infringement of any patent, copyright or other intellectual property right. Intel products are not intended for use in medical, life saving, or life sustaining applications. Intel retains the right to make changes to specifications and product descriptions at any time, without notice. The Intel 430TX PCIset may contain design defects or errors known as errata. Current characterized errata are available on request. Third-party brands and names are the property of their respective owners.
m o .c U t4 e e h S ta a .D w w w
n n n
(c) INTEL CORPORATION 1999 February 1999
Supports Mobile and Desktop (R) Supports the Pentium Processor Family Host Bus at 66 MHz and 60 MHz at 3.3V and 2.5V PCI 2.1 Compliant Integrated Data Path Integrated DRAM Controller 4 Mbytes to 256 MBytes main memory 64-Mbit DRAM/SDRAM Technology Support FPM (Fast Page Mode), EDO and SDRAM DRAM Support 6 RAS Lines Available Integrated Programmable Strength for DRAM Interface CAS-Before-RAS Refresh, Extended Refresh and Self Refresh for EDO CAS-Before-RAS and Self Refresh for SDRAM Integrated L2 Cache Controller 64-MB DRAM Cacheability Direct Mapped Organization--Write Back Only Supports 256K and 512K Pipelined Burst SRAM and DRAM Cache SRAM Cache Hit Read/Write Cycle Timings at 3-1-1-1 Back-to-Back Read/Write Cycles at 3-1-1-1-1-1-1-1 64K x 32 SRAM also supported
n
324-Pin MBGA 430TX PCIset Xcelerated Controller (MTXC) with integrated Data Paths
om .c 4U et he aS at .D w w w
Order Number: 273234-001
Extended Temperature 82439TX (MTXC) Datasheet
H D [6 3 :0 ] A [3 1 :0 ] B E [7 :0 ]# ADS# D /C # M /IO # W /R # BRDY# EADS# H IT M # BO FF# AHO LD NA# K E N # /IN V CACHE# HLOC K# S M IA C T # CCS# TW E# CO E# GW E# CADS# CADV# T IO [7 :0 ] K R Q A K /C S 4_ 64# BW E#
H ost In te rfa c e
PCI In te rfa c e
C LK R U N # A D [3 1 :0] C /B E [3:0]# FRAM E# TRDY# IR D Y # STOP# PLOCK# DEVSEL# PAR R E Q [3 :0 ]# G N T [3 :0 ]# PH LD A# PH LD # M D [6 3 :0 ] S R A S [A ,B ]# R A S [5 :0 ]# /C S [5 :0 ]# C A S [7 :0 ]# /D Q M [7 :0 ]# M A [1 1 :0 ] S C A S [A ,B ]# MW E# MW EB# C K E /M A A 0 , C K E B /M A A 1
DRAM In te rfa c e C ache In te rfa ce
C lo c k s , R e s e t, T e s t, and Power M gnt
H C L K IN P C L K IN RST# TEST# SU SCLK SU SSTAT1#
mtx_blk
MTXC Simplified Block Diagram
PRELIMINARY
2
Extended Temperature 82439TX (MTXC) Datasheet
CONTENTS
PAGE 1. 0. ARCHITECTURE OVERVIEW ..............................................................................................................6 2. 0. SIGNAL DESCRIPTION .......................................................................................................................9 2.1. MTXC Signals ..................................................................................................................................9 2.1.1. HOST INTERFACE........................................................................................................................9 2.1.2. DRAM Interface ...........................................................................................................................11 2.1.3. SECONDARY CACHE INTERFACE............................................................................................13 2.1.4. PCI INTERFACE .........................................................................................................................14 2.1.5. TEST AND CLOCK......................................................................................................................15 2.1.6. POWER MANAGEMENT.............................................................................................................15 2.1.7. POWER AND GROUND PINS.....................................................................................................15 2.2. MTXC Strapping Options................................................................................................................16 2.3. Power Planes .................................................................................................................................16 2.4. Power Sequencing Requirements ..................................................................................................17 2.5. Improving Signal Integrity in Lightly Loaded Systems.....................................................................18 2.6. Signal States During And After A Hard Reset.................................................................................18 3. 0. REGISTER DESCRIPTION.................................................................................................................20 3.1. I/O Mapped Registers.....................................................................................................................20 3.1.1. PM2_CNTRLPM2 REGISTER BLOCK.....................................................................................21 3.1.2. CONFADDCONFIGURATION ADDRESS REGISTER.............................................................21 3.1.3. CONFDATACONFIGURATION DATA REGISTER ..................................................................22 PCI Configuration Space Mapped Registers ......................................................................................22 3.1.4. VIDVENDOR IDENTIFICATION REGISTER............................................................................25 3.1.5. DIDDEVICE IDENTIFICATION REGISTER .............................................................................25 3.1.6. PCICMDPCI COMMAND REGISTER ......................................................................................25 3.1.7. PCISTSPCI STATUS REGISTER ............................................................................................26 3.1.8. RIDREVISION IDENTIFICATION REGISTER..........................................................................27 3.1.9. CLASSCCLASS CODE REGISTER.........................................................................................27 3.1.10. MLTMASTER LATENCY TIMER REGISTER.........................................................................27 3.1.11. HEDTHEADER TYPE REGISTER .........................................................................................28 3.1.12. BISTBIST REGISTER ............................................................................................................28 3.1.13. ACONARBITRATION CONTROL REGISTER........................................................................28 3.1.14. PCONPCI CONTROL REGISTER..........................................................................................29 3.1.15. CCCACHE CONTROL REGISTER ........................................................................................29 3.1.16. CECEXTENDED CACHE CONTROL REGISTER..................................................................31 3.1.17. SDRAMCSDRAM CONTROL REGISTER..............................................................................32 3.1.18. DRAMECDRAM EXTENDED CONTROL REGISTER............................................................34 3.1.19. DRAMCDRAM CONTROL REGISTER ..................................................................................35 3.1.20. DRAMTDRAM TIMING REGISTER........................................................................................36
PRELIMINARY
3
Extended Temperature 82439TX (MTXC) Datasheet
3.1.21. PAMPROGRAMMABLE ATTRIBUTE MAP REGISTERS (PAM[6:0]) ....................................37 DOS Application Area (00000h-9FFFh).............................................................................................39 Video Buffer Area (A0000h-BFFFFh) ................................................................................................39 Expansion Area (C0000h-DFFFFh) ...................................................................................................39 Extended System BIOS Area (E0000h-EFFFFh)...............................................................................39 System BIOS Area (F0000h-FFFFFh)...............................................................................................39 Extended Memory Area (100000h-FFFFFFFFh) ...............................................................................39 3.1.22. DRBDRAM ROW BOUNDARY REGISTERS .........................................................................40 3.1.23. DRTHDRAM ROW TYPE REGISTER HIGH ..........................................................................42 3.1.24. DRTL--DRAM ROW TYPE REGISTER LOW............................................................................43 3.1.25. MTTMULTI-TRANSACTION TIMER REGISTER (RESERVED TEST MODE REGISTER) ....43 3.1.26. ESMRAMCEXTENDED SYSTEM MANAGEMENT RAM CONTROL REGISTER..................44 3.1.27. SMRAMCSYSTEM MANAGEMENT RAM CONTROL REGISTER ........................................45 3.1.28. MCTLMISCELLANEOUS CONTROL REGISTER ..................................................................47 4. 0. FUNCTIONAL DESCRIPTION............................................................................................................48 4.1. Host Interface.................................................................................................................................48 4.2. Secondary Cache Interface ............................................................................................................48 4.2.1. CLOCK LATENCIES....................................................................................................................51 4.2.2. SNOOP CYCLES ........................................................................................................................51 4.2.3. DRAM CACHE SECOND LEVEL CACHE MODE........................................................................52 4.3. DRAM Interface..............................................................................................................................52 4.3.1. DRAM ORGANIZATION ..............................................................................................................53 4.3.2. CONFIGURATION REQUIREMENTS .........................................................................................55 4.3.3. DRAM ADDRESS TRANSLATION ..............................................................................................59 4.3.4. DRAM PAGING ...........................................................................................................................59 4.3.5. DRAM TYPES .............................................................................................................................59 4.3.5.1. FPM Mode ............................................................................................................................59 4.3.5.2. EDO Mode ............................................................................................................................59 4.3.5.3. SDRAM Mode .......................................................................................................................59 4.3.6. AUTO DETECTION .....................................................................................................................61 4.3.7. DRAM PERFORMANCE..............................................................................................................61 4.3.8. DRAM REFRESH ........................................................................................................................64 4.4. PCI CLK Control (CLKRUN#) .........................................................................................................64 4.4.1. CLOCKING STATES ...................................................................................................................64 4.4.2. OPERATION................................................................................................................................64 4.5. SMRAM Memory Space .................................................................................................................64 4.5.1. COMPATIBLE SMRAM (C_SMRAM) ..........................................................................................64 4.5.2. EXTENDED SMRAM (E_SMRAM) ..............................................................................................64 4.5.3. SMRAM PROGRAMMING CONSIDERATIONS ..........................................................................66 4.6. Low Power States ..........................................................................................................................66 4.6.1. CHIP STANDBY ..........................................................................................................................67 4.6.2. SUSPEND/RESUME ...................................................................................................................67
PRELIMINARY
4
Extended Temperature 82439TX (MTXC) Datasheet
4.6.2.1. Power Transition Changes ....................................................................................................68 4.7. PCI Interface ..................................................................................................................................69 4.8. System Arbitration ..........................................................................................................................70 4.8.1. Priority Scheme and Bus Grant....................................................................................................70 4.8.2. CPU Policies................................................................................................................................72 5. 0. CLOCKS AND RESET........................................................................................................................72 5.1. Clock Generation and distribution...................................................................................................72 5.2. RESET Sequencing .......................................................................................................................72 6. 0. ELECTRICAL TIMING SPECIFICATIONS...........................................................................................72 6.1. Absolute Maximum Ratings ..............................................................................................................72 6.2. Thermal Characteristics....................................................................................................................73 6.3. MTXC DC Characteristics.................................................................................................................74 6.4. MTXC AC Characteristics.................................................................................................................79 7. 0. MTXC Timing Diagrams .....................................................................................................................85 8. 0. PINOUT INFORMATION.....................................................................................................................89 9. 0. MTXC PACKAGE INFORMATION .....................................................................................................94 10. 0. TESTABILITY .....................................................................................................................................97 10.1. NAND Tree Mode ...........................................................................................................................97 10.2. NAND Chain Mode .........................................................................................................................97 11. 0. ERRATA ...........................................................................................................................................102 11.1. SDRAM Speculative Read Enable (SSRE)...................................................................................102 11.2. Fast Back-to-Back, PCI Peer-to-Peer Cycles ...............................................................................102
PRELIMINARY
5
Extended Temperature 82439TX (MTXC) Datasheet
1.0.
ARCHITECTURE OVERVIEW
The MTXC host bridge provides a completely integrated solution for the system controller and datapath components in a Pentium processor system. The MTXC Supports all Pentium family processors since P54C, it has 64-bit Host and DRAM Bus Interface, 32-bit PCI Bus Interface, Second level Cache Interface, and it integrates the PCI arbiter. The MTXC interfaces with the Pentium processor host bus, a dedicated memory data bus, and the PCI bus (see Figure 1). The MTXC bus interfaces are designed to interface with 2.5V, 3.3V and 5V busses. The MTXC implements 2.5V and 3.3V drivers and 5V tolerant receivers. The MTXC connects directly to the Pentium processor 3.3V or 2.5V host bus, directly to 5V or 3.3V DRAMs, and directly to the 5V or 3.3V PCI bus. The 430TX also interfaces directly to the 3.3V or 5.0V TAGRAM and 3.3V Cache. The MTXC works with the PCI IDE/ISA Accelerator 4 (PIIX4). The PIIX4 provides the PCI-to-ISA/EIO bridge functions along with other features such as a fast IDE interface (PIO mode 4 and Ultra DMA/33), Plug-n-Play port, APIC interface, PCI 2.1 Compliance, SMBUS interface, and Universal Serial Bus Host Controller functions. DRAM Interface The DRAM interface is a 64-bit data path that supports Standard (or Fast) Page Mode (FPM), Extended Data Out (EDO) and Synchronous DRAM (SDRAM) memory. The DRAM controller inside the MTXC is capable of generating 3-1-1-1 for posted writes for any type of DRAM that is used. While read performance is 6-1-1-1 for SDRAM, 5-2-2-2 for EDO, and 6-3-3-3 for FPM. The DRAM interface supports 4 Mbytes to 256 Mbytes with six RAS lines. The MTXC supports 4-Mbit, 16Mbit, and 64-Mbit DRAM and SDRAM technology, both symmetrical and asymmetrical. Parity is not supported, and for loading reasons, x32 and x64 SIMMs/DIMMs/SO-DIMMs should be used. Second Level Cache The second level cache is direct mapped and supports both 256-Kbyte and 512-Kbyte SRAM configuration using Pipeline Burst SRAM or DRAM Cache SRAM. The Cache performance is 3-1-1-1 for line read/write and 3-1-1-1-1-1-1-1 for back to back reads that are pipelined. Cacheless configuration is also supported. PCI Interface The PCI interface is 2.1 compliant and supports up to four PCI bus masters in addition to the PIIX4 bus master requests. Datapath and Buffers The MTXC contains three sets of data buffers for optimizing data flow. A five QWord deep DRAM write buffer is provided for CPU-to-DRAM writes, second level cache write backs, and PCI-to-DRAM transfers. This buffer is used to achieve 3-1-1-1 posted writes to DRAM and also provides DWord merging and burst merging for CPU-to-DRAM write cycles. In addition, an extra line of buffering is provided that is combined with the DRAM Write Buffer to supply an 18 DWord deep buffer for PCI to main memory writes. A five DWord buffer is provided for CPU-to-PCI writes to help maximize the bandwidth for graphic writes to the PCI bus. Also, five QWords of prefetch buffering has been added to the PCI-to-DRAM read path that allows up to two lines of data to be prefetched at an x-2-2-2 rate. The MTXC interfaces directly to the Host and DRAM data bus.
PRELIMINARY
6
Extended Temperature 82439TX (MTXC) Datasheet
Power Management Features The MTXC implements extensive power management features. The CLKRUN# feature enables controlling of the PCI clock (on/off). The MTXC supports POS, STR, STD, and Soft-off suspend states. SUSCLK and SUSSTAT1# signals are used for implementing Suspend Logic. The MTXC supports two SMRAM modes; Compatible SMRAM (C_SMRAM) and Extended SMRAM (E_SMRAM). The C_SMRAM is the traditional SMRAM feature implemented in Intel PCIsets. The E_SMRAM is a new feature that supports writeback cacheable SMRAM space up to 1 Mbytes. In order to minimize the idle power, the internal clock in MTXC is turned off (gated off) when there is no activity on the Host and PCI Bus.
PRELIMINARY
7
Extended Temperature 82439TX (MTXC) Datasheet
(R) Pentium Processor
Host Bus (3.3V or 2.5V I/O; 60-66 MHz)
Second Level Cache Cache (PBSRAM) Cntl 82349TX (MTXC) DRAM Interface (3.3V or 5V) Main Memory (DRAM)
Tag Cntl Tag TIO[7:0]
For Mobile Docking Stations Only PCI Slots PCI Slots PCI Bus (5V)
PCI Bus (3.3V or 5V, 30/33 MHz)
82380FB (MPCI2) PCI Docking Connector
CD ROM
Hard Disk BMI IDE Ultra DMA/33 Hard Disk 82371AB (PIIX4)
USB 1
USB 2
82380AB (MISA) ISA Bus (5V)
Universal Serial Bus GP[I,O] (30+) SMB (I2C)
ISA Slots ISA/EIO Bus (3.3V; 5V Tolerant)
Audio
KBD
SP, PP, FDC, IR
BIOS
mtx_sys
Figure 1. MTXC System Block Diagram
PRELIMINARY
8
Extended Temperature 82439TX (MTXC) Datasheet
2.0.
SIGNAL DESCRIPTION
This section provides a detailed description of each signal. The signals are arranged in functional groups according to their associated interface. The "#" symbol at the end of a signal name indicates that the active, or asserted state occurs when the signal is at a low voltage level. When "#" is not present after the signal name, the signal is asserted when at the high voltage level. The terms assertion and negation are used exclusively. This is done to avoid confusion when working with a mixture of "active low" and "active high" signal. The term assert, or assertion indicates that a signal is active, independent of whether that level is represented by a high or low voltage. The term negate, or negation indicates that a signal is inactive. The I/O buffer types are shown below: Buffer Type I O I/O s/t/s od 3.3V/2.5V 3.3V/5V 5V Description input only signal totem pole output bi-direction, tri-state input/output pin sustained tri-state open drain Indicates the buffer is 3.3V or 2.5V only, depending on the voltage (3.3V or 2.5V) connected to VccX pins. Indicates that the output is 3.3V and input is 3.3V receiver with 5V tolerance. Indicates 3.3V receiver with 5V tolerance.
2.1.
2.1.1.
MTXC Signals
HOST INTERFACE Name Type Description
A[31:3]
I/O Address Bus. A[31:3] connects to the address bus of the CPU. During CPU 3.3V/2.5V cycles A[31:3] are inputs. The MTXC drives A[31:3] during inquire cycles on behalf of PCI initiators. Bits A[31:26] act as inputs when RST# is active I Byte Enables. The CPU byte enables indicate which byte lane the current 3.3V/2.5V CPU cycle is accessing. All eight byte lanes must be provided to the CPU if the cycle is a cacheable read regardless of the state of BE[7:0]#. I Address Status. CPU asserts ADS# in T1 of the CPU bus cycle. 3.3V/2.5V O Bus Ready. The MTXC asserts BRDY# to indicate to the CPU that data is 3.3V/2.5V available on reads or has been received on writes. O Next Address. This signal is asserted by the MTXC to indicate to the 3.3V/2.5V Processor that it is ready to process a second cycle.
BE[7:0]#
ADS# BRDY# NA#
PRELIMINARY
9
Extended Temperature 82439TX (MTXC) Datasheet
Name AHOLD
Type
Description
O Address Hold. The MTXC asserts AHOLD when a PCI initiator is performing 3.3V/2.5V a cycle to DRAM. AHOLD is held for the duration of the PCI burst transfer. The MTXC will negate AHOLD when the completion of the PCI to DRAM read or write cycles complete and during PCI peer transfers. AHOLD is kept asserted while PHLDA# is asserted (i.e., duration of PIIX4 granting). O External Address Strobe. Asserted by the MTXC to inquire the first level 3.3V/2.5V cache when servicing PCI master references of DRAM. O Back Off. Asserted by the MTXC when required to terminate a CPU cycle 3.3V/2.5V that was in progress. I Hit Modified. Asserted by the CPU to indicate that the address presented 3.3V/2.5V with the last assertion of EADS# is modified in the first level cache and needs to be written back. I Memory/IO; Data/Control; Write/Read. Asserted by the CPU with ADS# to 3.3V/2.5V indicate the type of cycle that the system needs to perform. I Host Lock. All CPU cycles sampled with the assertion of HLOCK# and 3.3V/2.5V ADS#, until the negation of HLOCK# must be atomic, i.e. no PCI activity to DRAM is allowed. I Cache. Asserted by the CPU during a read cycle to indicate the CPU will 3.3V/2.5V perform a burst line fill. Asserted by the CPU during a write cycle to indicate the CPU will perform a burst writeback cycle. If CACHE# is asserted to indicate cacheability, the MTXC will assert KEN# either with the first BRDY#, or with NA# if NA# is asserted before the first BRDY#. O Ken/Invalidate. KEN#/INV functions as both the KEN# signal during CPU 3.3V/2.5V read cycles, and the INV signal during L1 snoop cycles. During CPU cycles, KEN#/INV is normally low. KEN#/INV is driven high during the 1st BRDY# or NA# assertion of a non-L1-cacheable CPU read cycle. KEN#/INV is driven high(low) during the EADS# assertion of a PCI master DRAM write(read) snoop cycle. Note that KEN#/INV operation during snoop cycles is independent of the FLCE bit programming.
EADS# BOFF# HITM#
M/IO#, D/C#, W/R# HLOCK#
CACHE#
KEN#/INV
SMIACT#
I System Management Interrupt Active. This is asserted by the CPU when it 3.3V/2.5V is in system management mode as a result of an SMI. This signal must be sampled active with ADS# for the processor to access the SMM space of DRAM, located at A0000h, after SMM space has been loaded and locked by BIOS at system boot. I/O Host Data. These signals are connected to the CPU data bus. These signals 3.3V/2.5V have internal pull-down resistors.
HD[63:0]
NOTES: All of the signals in the host interface are described in the Pentium Processor data sheet. The preceding table highlights MTXC specific uses of these signals.
PRELIMINARY
10
Extended Temperature 82439TX (MTXC) Datasheet
2.1.2. Name RAS[3:0]# or CS[3:0]#, RAS4#/CS4#/ BA1, RAS5#/CS5#/ MA13 CAS[7:0]# or DQM[7:0] O 3.3 V DRAM INTERFACE Type O 3.3 V Description Row Address Strobe--RASx# (EDO/FPM). These pins select the DRAM row. Chip Select--CSx# (SDRAM). These pins activate the SDRAMs. SDRAM accepts any command when its CS# pin is active low. Note: For 64Mbit SDRAM support, BA1/MA12 and MA13 are muxed with the RAS4# and RAS5# signals, respectively. When SDRAMC[bit 1]=1, BA1 and MA13 are driven out on these lines. Column Address Strobe (EDO/FPM). These pins select the DRAM column. Input/Output Data Mask SDRAM). These pins act as synchronized output enables during a read cycle and a byte mask during a write cycle. The read cycles require Tdqz clock latency before the functions are actually performed. In case of a write cycle, word mask functions are performed in the same cycle (0 cycle latency). Memory Address (EDO/FPM/SDRAM). This is the row and column address for DRAM. These buffers now include programmable size selection, as controlled by the DRAMEC[MAD] bit. For 64-Mbit SDRAM support BA1/MA12 and MA13 are muxed with the RAS4# and RAS5# signals, respectively. Memory Write Enable (second copy) (EDO/FPM/SDRAM). MWE# should be used as the write enable for the memory data bus. This signal has programmable buffer size selection. Memory Write Enable (EDO/FPM/SDRAM). MWE# should be used as the write enable for the memory data bus. This signal has programmable buffer size selection. SDRAM Row Address Strobe (SDRAM). When asserted, this signal latches Row Address on the positive edge of the clock. This signal also allows Row access and precharge. Two copies are provided for loading purpose. These signals have programmable buffer size selection. SDRAM Column Address Strobe (SDRAM). When asserted, this signal latches Column Address on the positive edge of the clock. This signal also allows Column access. Two copies provided for loading purpose. These signals have programmable buffer size selection. SDRAM Clock Enable (SDRAM). SDRAM clock enable pin. When this signal is negated, SDRAM enters power down mode. This signal is also muxed to provide a second copy of memory address MA0 (MAA0). The MA function is selected via DRT[bit2] (offset 67h). MTXC negates CKE (and CKEB) when SUSSTAT1# is asserted. Note that MTXC asserts CKE (and CKEB) for all rows (i.e., CKE and CKEB cannot be selectively asserted for certain rows and negated for other rows).
MA[11:0]
O 3.3 V
MWEB#
O 3.3 V O 3.3 V O 3.3 V
MWE#
SRAS[A,B]#
SCAS[A,B]#
O 3.3 V
CKE/MAA0
O 3.3 V
PRELIMINARY
11
Extended Temperature 82439TX (MTXC) Datasheet
Name CKEB/MAA1
Type O 3.3 V
Description SDRAM Clock Enable (SDRAM) (second copy). SDRAM clock enable pin. When this signal is negated, SDRAM enters into power down mode. Note that this signal is not implemented in the "Suspend Well" and should not be used if suspend to RAM (STR) is implemented. This signal is also muxed to provide a second copy of memory address MA1 (MAA1). The MA function is selected via DRT[bit2] (offset 67h). MTXC negates CKE (and CKEB) when SUSSTAT1# is asserted. Note that MTXC asserts CKE (and CKEB) for all rows (i.e., CKE and CKEB cannot be selectively asserted for certain rows and negated for other rows).
MD[63:0]
I/O 3.3V/5V
Memory Data. These signals are connected to the DRAM data bus. These signals have internal pull-down resistors
PRELIMINARY
12
Extended Temperature 82439TX (MTXC) Datasheet
2.1.3. SECONDARY CACHE INTERFACE Name CADV# CADS# CCS# Type O 3.3V O 3.3V O 3.3V Description Cache Advance. Assertion causes the PBSRAM in the secondary cache to advance to the next QWord in the cache line. Cache Address Strobe. Assertion causes the PBSRAM in the secondary cache to load the PBSRAM address register from the PBSRAM address pins. Cache Chip Select (CCS#). The second level cache will power up, if necessary, and perform an access if this signal is asserted when CADS# is asserted. The second level cache will power down if this signal is negated when CADS# is asserted. When CCS# is negated the second level cache will ignore ADS#. If CCS# is asserted when ADS# is asserted, the second level cache will power up, if necessary, and perform an access. Cache Output Enable. The secondary cache data RAMs drive the CPUs data bus when COE# is asserted. Global Write Enable. GWE# assertion causes all the byte lanes to be written into the secondary cache data RAMs, if they are powered up. Byte Write Enable. Asserted low with GWE#=HIGH to enable using host's BE[7:0]# to be used to control byte lanes to pipeline burst SRAM cache. Tag Address. These are inputs during CPU accesses and outputs during second level cache line fills and second level cache line invalidates due to inquire cycles. These signals have internal pull-down resistors. Tag Write Enable. When asserted, new state and tag addresses are written into the external tag. KRQAK/Chip Select 4 (for 64-Mb Technology). This pin is a dual-function signal. KRQAK is used in a DRAM Cache L2 implementation and is a bi-directional refresh request/acknowledge. The CS4_64# function is used to generate the fifth chip select line in a SDRAM L2 Cache implementation that supports five rows of 64-Mbit SDRAM. During a hard reset, this signal is sampled to determine if DRAM cache is in the system (see MTXC Strapping options). This signal has a weak internal pull-down. If SDRAMC[bit 1]=1 and DRAM cache is not present in the system (indicated by CEC[bit 5]=0, offset 53h), the CS4_64# function is selected. If DRAM cache is in the system or SDRAMC[bit 1] (offset 54h)=0, then KRQAK is used to drive the KRQAK function.
COE# GWE# BWE# TIO[7:0]
O 3.3V O 3.3V O 3.3V I/O 3.3V/5V O 3.3V I/O 3.3V
TWE# KRQAK/ CS4_64#
PRELIMINARY
13
Extended Temperature 82439TX (MTXC) Datasheet
2.1.4. Name AD[31:0] C/BE[3:0]# FRAME# DEVSEL# IRDY# TRDY# STOP# LOCK# REQ[3:0]# GNT[3:0]# PHLD# PCI INTERFACE Type I/O 3.3/5V I/O 3.3/5V I/O 3.3/5V I/O 3.3/5V I/O 3.3/5V I/O 3.3/5V I/O 3.3/5V I/O 3.3/5V I 3.3/5V O 3.3V I 3.3/5V O 3.3V I/O 3.3/5V I/O 3.3/5V Description Address/Data. The standard PCI address and data lines. Address is driven with FRAME# assertion, data is driven or received in following clocks. Command/Byte Enable. The command is driven with FRAME# assertion, byte enables corresponding to supplied or requested data is driven on following clocks. Frame. Assertion indicates the address phase of a PCI transfer. Negation indicates that one more data transfer is desired by the cycle initiator. Device Select. This signal is driven by the MTXC when a PCI initiator is attempting to access DRAM. DEVSEL# is asserted at medium decode time. Initiator Ready. Asserted when the initiator is ready for a data transfer. Target Ready. Asserted when the target is ready for a data transfer. Stop. Asserted by the target to request the master to stop the current transaction. Lock. Used to establish, maintain, and release resource locks on PCI. PCI Request. PCI master requests for PCI bus. PCI Grant. Permission is given to the master to use PCI. PCI Hold. This signal comes from the expansion bridge. It is the bridge request for PCI. The MTXC will drain the DRAM write buffers, drain the CPU-to-PCI posting buffers, and acquire the host bus before granting via PHLDA#. PCI Hold Acknowledge. This signal is driven by the MTXC to grant PCI to the expansion bridge. PHLDA# protocol has been modified to include support for passive release. Parity. A single parity bit is provided over AD[31:0] and C/BE[3:0]. This signal should be pulled high through a weak external pull-up resistor. CLOCK RUN. An open drain output and also an input. MTXC requests the central resource (PIIX4) to start, or maintain the PCI clock by the assertion of CLKRUN#. MTXC will tri-state CLKRUN# upon negation of reset (since CLK is running upon negation of reset). External pull-up is required. Note: This signal should be connected to the PIIX4 CLKRUN# pin. However, if it is left as a no connect on the MTXC, it must be pulled low through a 100 (pull-down resistor. Reset. When asserted this signal asynchronously resets the MTXC. The PCI signals also tri-state compliant to PCI Rev 2.0 and 2.1 specifications.
PHLDA#
PAR CLKRUN#
RST#
I 3.3/5V
PRELIMINARY
14
Extended Temperature 82439TX (MTXC) Datasheet
2.1.5. Name TEST# TEST AND CLOCK Type I 3.3/5V I 3.3/2.5 V I 3.3/5V Description Test In. NAND tree mode is activated by driving this pin low. The test mode selected depends on the state of REQ[3:0]#. This pin should be pulled high with an external pull-up during normal operation. Host Clock In. This pin receives a buffered host clock. This clock is used by all of the MTXC logic that is in the Host clock domain. PCI Clock In. This pin receives a buffered divide-by-2 host clock. This clock is used by all of the MTXC logic that is in the PCI clock domain.
HCLKIN
PCLKIN
2.1.6. Name SUSCLK
POWER MANAGEMENT Type I 3.3V Description Suspend Clock. The signal is a 32 KHz input for DRAM refresh circuitry and clocking events in suspend state. The DRAM refresh during suspend and non-suspend states is performed based on this clock. This signal has an internal pull-down resistor. Suspend Status. SUSSTAT1# indicates MTXC's power plane status during suspend mode. SUSSTAT1#, along with SUSCLK and RST#, define the suspend protocol between MTXC and PIIX4. This signal has an internal pull-up resistor.
SUSSTAT1#
I 3.3V
2.1.7. Name VCC
POWER AND GROUND PINS Type 3.3V 3.3V or 2.5V 3.3V Description Main voltage supply. These pins are the primary voltage supply for the MTXC core and I/O periphery and must be connected to 3.3V. CPU Interface Voltage Supply. These pins are the primary voltage supply for the MTXC Host periphery and must be connected to either 2.5V or 3.3V, depending on the voltage level of the CPU interface. Refer to the Power sequencing requirements section for additional details. Suspend Well Voltage Supply. These pins are the primary voltage supply for the MTXC suspend logic and I/O. If suspend to RAM is supported, these pins should be on an isolated power plane; otherwise, they can be connected to the same 3.3V source used for the VCC pins. Voltage Reference. This pin is tied to 5V through a small external power sequencing circuit, if MTXC signals are required to be 5V Tolerant. In a non 5V tolerant system (i.e. 3.3V only system), this signal can be tied directly to VCC. Refer to the Power sequencing requirements section for additional details. Ground. These pins are the ground for the MTXC.
VCC (CPU)
VCC (SUS)
VCC5REF
3.3V or 5V
VSS
0V
PRELIMINARY
15
Extended Temperature 82439TX (MTXC) Datasheet
2.2.
Name SCS
MTXC Strapping Options
Type A[31:30] A[29:28] KRQAK Description Secondary Cache Size. Described in the Cache Control Register bits 7:6. Initial L2 RAM Type. Described in the Cache Control Register bits 5:4. DRAM Cache L2 Present Upon Reset Negation. This bit is sampled to detect DRAM L2 cache. If sampled high, a DRAM Cache is present. A weak pulldown is provided internally. A DRAM cache module should implement a pull-up on this pin that overrides the weak pulldown. BIOS does not have to be aware of this, this information is used by the MTXC to maintain optimal Pburst timings. 2.5V Voltage Detection. This bit is used to determine the voltage level (3.3V or 2.5V) of the host clock connected to the host clock pin and the voltage on the VCC(CPU) pins. An external pull-down or pull-up resistor is required on this pin (pulled down for 2.5V and pulled up for 3.3V). Frequency Detection. BIOS can use this bit to determine if the system is 60 MHz (external pull-up) or 66 MHz (no strapping is present) as described in the DRTH Register, bit 7. DRTH[bit 7] register is initialized with the inverted value of pin A27 upon reset negation. The A27 input buffer includes a weak pulldown resistor which will force DRTH[bit 7] to default to 1 if no strapping is present.
L2RAMT DRAM Cache
25VD
A26
HFD
A27
2.3.
Power Planes
The MTXC has three primary internal power planes. These power planes permit parts of the MTXC to power down to conserve battery life. Table 1 shows the internal planes and their uses. Table 1. MTXC Internal Power Planes Power Plane SUSPEND Description Signals Powered VCC Pins VCC (SUS) GND Pins VSS
MWE#, MWEB#, Contains the logic needed to resume from the CKE, RAS[5:0]# 1, Suspend-to-RAM state. This power supply should be capable of providing a "trickle" current. CAS[7:0]#, SUSCLK, SUSSTAT1# The input signals attached to the SUSPEND power plane Do Not Support 5V Input Levels. These signals must not exceed VCC (SUS).
CPU
CPU Interface signals have a separate supply so A[31:3], BE[7:0]#, VCC (CPU) that the CPU interface can be 3.3V for existing ADS#, BRDY#, NA#, CPUs and can be 2.5V on future CPUs. AHOLD, EADS#, BOFF#, HITM#, M/IO#, D/C#, W/R#, HLOCK#, CACHE#, KEN#/INV, SMIACT#, HD[63:0], HCLKIN The VCC5REF signal provides protection for the 5V tolerant 3.3V signals. PCI Bus Input and I/O, MD[63:0], TIO[7:0], PCLKIN, TEST# VCC5REF
VSS
VCC5REF
VSS
PRELIMINARY
16
Extended Temperature 82439TX (MTXC) Datasheet
Table 1. MTXC Internal Power Planes Power Plane MAIN Description Contains all the rest of the MTXC logic. This plane is powered by the main system power supply. Signals Powered All Other Signal Pins VCC Pins VCC GND Pins VSS
NOTES: 1. KRQAK is not part of the suspend well. When this pin is used as the 5th RAS line (CS4_64), special considerations must be taken.
2.4.
Power Sequencing Requirements
The VCC5REF signal must be tied to 5V in a system requiring 5V tolerance. In a 5V tolerant system, VCC5REF must power up before or simultaneous to VCC. It must power down after or simultaneous to VCC. At any time, VCC5REF should not be more than 0.6 volts below VCC. In a non-5V tolerant system (3.3V only), this signal can be tied directly to VCC. In this case, there are no sequencing requirements. Refer to Figure 2 for an example circuit schematic which may be used to ensure the proper VCC5REF sequencing.
Vcc Supply (3.3V)
5V Supply 1 k
1 uF
To System
VREF
To System
pwr_seq
Figure 2. Example VCC5REF Sequencing Circuit (FIX SYMBOL- See hardcopy)
The VCC(CPU) power plane is tied to either 2.5 volts or 3.3 volts, depending on the voltage level of the CPU interface. In a system that ties this power plane to 2.5 volts, the VCC(CPU) pins must power up after or simultaneous to VCC. It must power down before or simultaneous to VCC. At any time, VCC should not be more than 1.2 volts below the VCC(CPU) plane. VCC(SUS) can power up and power down independent of all other power planes.
PRELIMINARY
17
Extended Temperature 82439TX (MTXC) Datasheet
2.5.
Improving Signal Integrity in Lightly Loaded Systems
To insure that DRAM interface signal integrity is maintained for lightly loaded systems, series termination and/or diodes (Gnd and Vcc diodes) are recommended on the following signals: CAS#/DQMx, MWEx, SCASx, SRASx, CKEx, and all MA lines (note: RAS4# and RAS5# are also used as MA lines, depending on the configuration, and should be terminated when used as MA lines). This will insure that the overshoot, undershoot, and most importantly, ring-back does not cause any problems. If series termination is used, use 10 ohms. This value provides the best signal integrity and flight time results. Place as close to the driver as possible. If diodes are used, the diodes should have a forward current of at least 200ma at 1 volt. A MMBD1203 diode or equivalent meets this requirement. The diodes should be placed at the end of the trace. Diodes improve signal integrity without increasing the flight time. A 10 ohm series resistor will increase the flight time by approximately 300ps. Both provide similar signal integrity results.
2.6.
Signal States During And After A Hard Reset
Table 2 shows the state of all the MTXC output and bi-directional signals when RST# is asserted. An undefined state means that the signal is driven either high or low, but not tri-stated. Table 2. Signal States During/After Reset Name A[31:3] BRDY# NA# AHOLD EADS# BOFF# KEN#/INV HD[63:0] RAS[5:0]# or CS[5:0]# CAS[7:0]# or DQM[7:0] MA[11:0], BA1,MA13 MWE#, MWEB# SRAS[A,B]# State during RST# Low High High High High High Low Tri-State Undefined Undefined Undefined High High State After RST# Tri-State High High Low High High Low Tri-State High Undefined Undefined High High Table 2. Signal States During/After Reset Name SCAS[A,B]# CKE,CKEB MD[63:0] CADV# CADS# CCS# COE# GWE# BWE# TIO[7:0] TWE# KRQAK AD[31:0] C/BE[3:0]# FRAME# DEVSEL# State during RST# High Undefined Tri-State High High Low High High High Low Low Input Low Low Tri-State Tri-State State After RST# High High Tri-State High High Low High High High Tri-State High Input Tri-State Tri-State Tri-State Tri-State
PRELIMINARY
18
Extended Temperature 82439TX (MTXC) Datasheet
Table 2. Signal States During/After Reset Name IRDY# TRDY# STOP# LOCK# State during RST# Tri-State Tri-State Tri-State Tri-State State After RST# Tri-State Tri-State Tri-State Tri-State
Table 2. Signal States During/After Reset Name GNT[3:0]# PHLDA# PAR CLKRUN# State during RST# Tri-State High Low Tri-State State After RST# High High Undefined Tri-State
PRELIMINARY
19
Extended Temperature 82439TX (MTXC) Datasheet
3.0.
REGISTER DESCRIPTION
The MTXC contains two sets of software accessible registers (I/O Mapped and PCI configuration registers), accessed via the Host CPU I/O address space. The I/O mapped registers control access to PCI configuration space. Configuration registers residing in PCI configuration space used to specify PCI configuration, DRAM configuration, cache configuration, operating parameters and optional system features. The MTXC internal registers (both I/O Mapped and PCI Configuration registers) are only accessible by the Host CPU and cannot be accessed by PCI masters. The registers can be accessed as Byte, Word (16-bit), or DWord (32-bit) quantities, with the exception of CONFADD, which can only be accessed as a DWord. All multi-byte numeric fields use "little-endian" ordering (i.e., lower addresses contain the least significant parts of the field). The following nomenclature is used for access attributes: RO R/W R/WC READ ONLY. If a register is read only, writes to this register have no effect. READ/WRITE. A register with this attribute can be read and written. READ/WRITE CLEAR. A register bit with this attribute can be read and written. However, a write of 1 clears (sets to 0) the corresponding bit and a write of 0 has no effect.
Some of the MTXC registers described in this section contain reserved bits. Software must deal correctly with fields that are reserved. On reads, software must use appropriate masks to extract the defined bits and not rely on reserved bits being any particular value. On writes, software must ensure that the values of reserved bit positions are preserved. That is, the values of reserved bit positions must first be read, merged with the new values for other bit positions and then written back. In addition to reserved bits within a register, the MTXC contains address locations in the PCI configuration space that are marked "Reserved" (Table 3). The MTXC responds to accesses to these address locations by completing the Host cycle and returning a value of zero. The registers marked as "Undefined" will return a non-zero value and are defined as read only. Software should not write to reserved or undefined MTXC configuration locations in the device-specific region (above address 3Fh). Upon RESET, the MTXC sets its internal configuration registers to predetermined default states. The default state represents the minimum functionality feature set required to successfully bring up the system. Hence, it does not represent the optimal system configuration. It is the responsibility of the system initialization software (usually BIOS) to properly determine the DRAM configurations, cache configuration, operating parameters and optional system features that are applicable, and to program the MTXC registers accordingly.
3.1.
I/O Mapped Registers
The MTXC contains three registers that reside in the CPU I/O address space--the Configuration Address (CONFADD) Register, the Configuration Data (CONFDATA) Register, and the PM2 Register Block. The Configuration Address Register enables/disables the configuration space and determines what portion of configuration space is visible through the Configuration Data window.
PRELIMINARY
20
Extended Temperature 82439TX (MTXC) Datasheet
3.1.1. PM2_CNTRLPM2 REGISTER BLOCK 0022h 00h Read/Write Descriptions Reserved. Arbiter Disable (ARB_DIS). When ARB_DIS=1, the MTXC does not respond to any REQ# signals (including PHOLD#) going active until this bit is set back to 0. This bit is used to disable bus master accesses prior to placing the CPU in a stop clock state. This bit maintains cache coherency by preventing PCI masters from gaining access to the PCI bus and causing snoop cycle activity. MCTL[Bit 6] (offset 79h) must be set to 1 before this register is accessible.
I/O Address: Default Value: Access: Bit 7:1 0
3.1.2.
CONFADDCONFIGURATION ADDRESS REGISTER 0CF8h (Accessed as a DWord) 00000000h Read/Write
I/O Address: Default Value: Access:
CONFADD is a 32-bit register accessed only when referenced as a DWord. A Byte or Word reference will "pass through" the Configuration Address Register onto the PCI bus. The CONFADD register contains the Bus Number, Device Number, Function Number, and Register Number for which a subsequent configuration access is intended. Bit 31 30:24 23:16 Descriptions Configuration Enable (CONE). 1=Enable. 0=Disable. Reserved. Bus Number. When the Bus Number is programmed to 00h the target of the Configuration Cycle is either the MTXC or the PCI Local Bus that is directly connected to the MTXC, depending on the Device Number field. A type 0 Configuration Cycle is generated on PCI if the Bus Number is programmed to 00h and the MTXC is not the target. If the Bus Number is non-zero a type 1 configuration cycle is generated on PCI with the Bus Number mapped to AD[23:16] during the address phase. Device Number. This field selects one agent on the PCI bus selected by the Bus Number. During a Type 1 Configuration cycle this field is mapped to AD[15:11]. During a Type 0 Configuration Cycle this field is decoded and one of AD[31:11] is driven to a 1. The MTXC is always Device Number 0. Function Number. This field is mapped to AD[10:8] during PCI configuration cycles. This allows the configuration registers of a particular function in a multi-function device to be accessed. The MTXC responds to configuration cycles with a function number of 000b; all other function number values attempting access to the MTXC (Device Number=0, Bus Number=0) will generate a type 0 configuration cycle onto the PCI bus with no IDSEL asserted, which will result in a master abort. Register Number. This field selects one register within a particular Bus, Device, and Function as specified by the other fields in the Configuration Address Register. This field is mapped to AD[7:2] during PCI configuration cycles. Reserved.
15:11
10:8
7:2
1:0
PRELIMINARY
21
Extended Temperature 82439TX (MTXC) Datasheet
3.1.3. CONFDATACONFIGURATION DATA REGISTER 0CFCh 00000000h Read/Write
I/O Address: Default Value: Access:
CONFDATA is a 32-bit read/write window into configuration space. The portion of configuration space that is referenced by CONFDATA is determined by the contents of CONFADD. Bit 31:0 Descriptions Configuration Data Window (CDW). If bit 31 of CONFADD is 1, any I/O reference that falls in the CONFDATA I/O space is mapped to configuration space using the contents of CONFADD.
PCI Configuration Space Mapped Registers The PCI Bus defines a slot based "configuration space" that allows each device to contain up to 256 8-bit configuration registers. The PCI specification defines two bus cycles to access the PCI configuration space Configuration Read and Configuration Write. While memory and I/O spaces are supported by the Pentium microprocessor, configuration space is not supported. The PCI specification defines two mechanisms to access configuration space, Mechanism #1 and Mechanism #2. The MTXC supports only Mechanism #1. The bus cycles used to access MTXC internal configuration registers are described later in the PCI cycle timings section. The configuration access mechanism makes use of the CONFADD Register and CONFDATA Register. To reference a configuration register, a DWord I/O write cycle is used to place a value into CONFADD that specifies the PCI bus, the device on that bus, the function within the device, and a specific configuration register of the device function being accessed. CONFADD[31] must be 1 to enable a configuration cycle. CONFDATA then becomes a window onto four bytes of configuration space specified by the contents of CONFADD. Any read or write to CONFDATA will result in the MTXC translating CONFADD into a PCI configuration cycle. Type 0 Access If the Bus Number field of CONFADD is 0 a Type 0 Configuration cycle is performed on PCI. CONFADD[10:2] is mapped directly to AD[10:2]. The Device Number field of CONFADD is decoded onto AD[31:11]. The MTXC is Device #0 and does not pass its configuration cycles to PCI so AD11 will never be asserted. Device #1 will assert AD12, Device #2 will assert AD13, and so forth up to Device #20 which will assert AD31. Only one AD line is asserted at a time. All device numbers higher than 20 cause a type 0 configuration access with no IDSEL asserted, which will result in a Master Abort. Type 1 Access If the Bus Number field of CONFADD is non-zero a Type 1 Configuration cycle is performed on PCI. CONFADD[23:2] is mapped directly to AD[23:2]. AD[1:0] are driven to 01 to indicate a Type 1 Configuration cycle. All other lines are driven to 0.
PRELIMINARY
22
Extended Temperature 82439TX (MTXC) Datasheet
Table 3. MTXC Configuration Space Address Offset Register Symbol Register Name PCI Specific Registers 00-01h 02-03h 04-05h 06-07h 08 09-0Bh 0Ch 0Dh 0Eh 0Fh 10-3Fh VID DID PCICMD PCISTS RID CLASSC MLT HEDT BIST Vendor Identification Device Identification PCI Command Register PCI Status Register Revision Identification Class Code Reserved Master Latency Timer Header Type BIST Register Reserved MTXC Specific Registers 40-4Eh 4Fh 50h 51h 52h 53 54-55h 56h 57h 58h 59-5Fh 60-65h 66h 67h 68h 69-6Ah ACON PCON CC CEC SDRAMC DRAMEC DRAMC DRAMT PAM[6:0] DRB[5:0] DRTH DRTL Reserved Arbitration Control PCI Control Reserved Cache Control Extended Cache Control SDRAM Control DRAM Extended Control DRAM Control DRAM Timing Programmable Attribute Map (7 registers) DRAM Row Boundary (6 registers) Reserved DRAM Row Type High DRAM Row Type Low Undefined R/W R/W R/W R/W RW R/W R/W R/W R/W R/W R/W R/W RO RO RO R/W RO, R/WC RO RO R/W R/W Access
PRELIMINARY
23
Extended Temperature 82439TX (MTXC) Datasheet
Table 3. MTXC Configuration Space Address Offset 6B-6Fh 70h 71h 72h 73h 74h 76-78h 78h 79 7A-FCh FDh FE-FFh MTT ESMRAMC SMRAMC MCTL Register Symbol Reserved Multi-Transaction Timer Extended System Management RAM Control System Management RAM Control Reserved Undefined Reserved Undefined Miscellaneous Control Register Reserved Undefined Reserved Register Name Access R/W R/W R/W RO RO R/W RO
PRELIMINARY
24
Extended Temperature 82439TX (MTXC) Datasheet
3.1.4. VIDVENDOR IDENTIFICATION REGISTER 00-01h 8086h Read Only
Address Offset: Default Value: Attribute:
The VID Register contains the vendor identification number. This 16-bit register combined with the Device Identification Register uniquely identify any PCI device. Writes to this register have no effect. Bit 15:0 Description Vendor Identification Number. This is a 16-bit value assigned to Intel.
3.1.5.
DIDDEVICE IDENTIFICATION REGISTER 02-03h 7100h Read Only
Address Offset: Default Value: Attribute:
This 16-bit register combined with the Vendor Identification register uniquely identifies any PCI device. Writes to this register have no effect. Bit 15:0 Description Device Identification Number. This is a 16 bit value assigned to the MTXC.
3.1.6.
PCICMDPCI COMMAND REGISTER 04-05h 06h Read/Write
Address Offset: Default: Access:
This 16-bit register provides basic control over the MTXC's ability to respond to PCI cycles. The PCICMD Register in the MTXC enables and disables the assertion of SERR# and PCI master accesses to main memory. Bit 15:10 9 8 7 6 5 4 Reserved. Fast Back-to-Back (FB2B). (Not implemented) This bit is hardwired to 0. SERR# Enable (SERRE). (Not implemented) This bit is hardwired to 0. Address/Data Stepping. (Not implemented) This bit is hardwired to 0. Parity Error Enable (PERRE). (Not implemented) This bit is hardwired to 0. Video Pallet Snooping (VPS). (Not implemented) This bit is hardwired to 0. Memory Write and Invalidate Enable (MWIE). (Not implemented) This bit is hardwired to 0. The MTXC will never use the Memory Write and Invalidate PCI command. Description
PRELIMINARY
25
Extended Temperature 82439TX (MTXC) Datasheet
Bit 3 2 1
Description Special Cycle Enable (SCE). (Not implemented) This bit is hardwired to 0, as the MTXC does not respond to PCI special cycles. Bus Master Enable (BME). (Not implemented) This bit is hardwired to 1. The MTXC does not support disabling of its bus master capability on the PCI Bus. Memory Access Enable (MAE). When MAE=1, the MTXC permits PCI masters to access main memory if the PCI address selects enabled DRAM space. When MAE=0, the MTXC does not respond to main memory accesses. I/O Access Enable (IOAE). (Not implemented) The MTXC does not respond to PCI I/O cycles. This bit is hardwired to 0.
0
3.1.7.
PCISTSPCI STATUS REGISTER 06-07h 0200h Read Only, Read/Write Clear
Address Offset: Default Value: Access:
PCISTS is a 16-bit status register that reports the occurrence of a PCI master abort and PCI target abort. PCISTS also indicates the DEVSEL# timing that has been set by the MTXC hardware. Bit 15 14 13 Description Detected Parity Error (DPE). This bit is hardwired to 0, as PCI received parity checking is not implemented by the MTXC. Signaled System Error (SSE). This bit is hardwired to 0 as MTXC does not support SERR#. Received Master Abort Status (RMAS). When the MTXC terminates a Host-to-PCI transaction (MTXC is a PCI master) with an unexpected master abort, this bit is set to 1. Note that master abort is the normal and expected termination of PCI special cycles. Software resets this bit to 0 by writing a 1 to it. Received Target Abort Status (RTAS). When a MTXC-initiated PCI transaction is terminated with a target abort, RTAS is set to 1. Software resets RTAS to 0 by writing a 1 to it. Signaled Target Abort Status (STAS). This bit is hardwired to 0, as the MTXC never terminates a PCI cycle with a target abort. DEVSEL# Timing (DEVT). This 2-bit field indicates the timing of the DEVSEL# signal when the MTXC responds as a target, and is hardwired to the value 01b (medium) to indicate the slowest time that DEVSEL# is generated. Data Parity Detected (DPD). This bit is hardwired to 0, as PERR# is not implemented. Fast Back-to-Back (FB2B). This bit is hardwired to 0, as fast back to back cycle generation is not implemented. User Defined Format (UDF). This bit is hardwired to 0. This is because the MTXC does not contain any configurations that depend on the environment, such as network frequencies. 66-MHz PCI Capable (66C). This bit is hardwired to 0. The MTXC does not interface to 66-MHz PCI. Reserved.
12 11 10:9
8 7 6 5 4:0
PRELIMINARY
26
Extended Temperature 82439TX (MTXC) Datasheet
3.1.8. RIDREVISION IDENTIFICATION REGISTER 08h 01h Read Only
Address Offset: Default Value: Access:
This register contains the revision number of the MTXC. These bits are read only and writes to this register have no effect. Bit 7:0 Description Revision Identification Number. This is an 8-bit value that indicates the revision identification number for the MTXC.
3.1.9.
CLASSCCLASS CODE REGISTER 09-0Bh 00h Read Only
Address Offset: Default Value: Access:
This register contains the device programming interface information related to the Sub-Class Code and Base Class Code definition for the MTXC. This register also contains the Base Class Code and the function sub-class in relation to the Base Class Code. Bit 23:16 15:8 7:0 Description Base Class Code (BASEC). 06=Bridge device. Sub-Class Code (SCC). 00h=Host Bridge. Programming Interface (PI). 00h=Hardwired as a Host-to-PCI Bridge.
3.1.10.
MLTMASTER LATENCY TIMER REGISTER 0Dh 00h Read/Write
Address Offset: Default Value: Access:
MLT is an 8-bit register that controls the amount of time the MTXC, as a bus master, can burst data on the PCI Bus. The Count Value is an 8-bit quantity. However MLT[2:0] are reserved and assumed to 0 when determining the Count Value. MLT is used to guarantee the host CPU a minimum amount of the system resources. The number of clocks programmed in the MLT represents the guaranteed time slice (measured in PCI clocks) allotted to the MTXC, after which it must surrender the bus as soon as other PCI masters request the bus. The default value of MLT is 00h or 0 PCI clocks. Bit 7:3 2:0 Master Latency Timer Count Value Reserved. Read as 0s Description
PRELIMINARY
27
Extended Temperature 82439TX (MTXC) Datasheet
3.1.11. HEDTHEADER TYPE REGISTER 0Eh 00h Read Only
Address Offset: Default Value: Access:
This register contains the Header Type of the MTXC. This code is 00h indicating that the MTXC's configuration space map follows the basic format. Bit 7:0 3.1.12. Description Device Type (DEVICET). 00h=Indicates a basic configuration space format. BISTBIST REGISTER 0Fh 00h Read/Write
Address Offset: Default Value: Access:
The Built In Self Test (BIST) function is not supported by the MTXC. Writes to this register have no effect. Bit 7 6 5:4 3:0 3.1.13. Description BIST Supported. This read only bit is always set to 0, disabling the BIST function. Writes to this bit position have no effect. Start BIST. This function is not supported and writes have no effect. Reserved. Completion Code. This read only field always returns 0 when read and writes have no effect. ACONARBITRATION CONTROL REGISTER 4Fh 00h Read/Write
Address Offset: Default Value: Access:
The ACON Register enables and disables features related to PCI arbitration and PCI 2.1 compliance. Bit 7 Description Extended CPU-to-PIIX4 PHLDA# Signaling Enable (XPLDE). When XPLDE=1, the MTXC adds the following additional signaling to signal PHLDA# (i.e., in addition to the normal CPU/PIIX4 PHOLD/PHLDA# protocol): 1. Whenever the North bridge begins a PCI read/write transaction, it will assert PHLDA# for 1 PCLK within the address phase of the transaction. 2. If the CPU is attempting a LOCKed cycle AND LOCK has been established (i.e. PLOCK# was seen negated in address phase), the PHLDA# remains asserted for one additional clock following the address phase. This bit should be set to 1 anytime both Passive Release and Delayed Transaction are enabled in the PIIX4. Passive release and delayed transaction are enabled via bits 1 and 0 in PIIX4 register 82h (function 0). When bit 7 in this register is set to 1 (enabled), Bit 7 in PIIX4 Register, 6A (Function 0) must also be set to 1. When enabling these two bits, enable Bit 7 in the PIIX4 first, followed by bit 7 in this register. When disabling these two bits, disable Bit 7 in this register first, followed by bit 7 in the PIIX4. Reserved.
6:0
PRELIMINARY
28
Extended Temperature 82439TX (MTXC) Datasheet
3.1.14. PCONPCI CONTROL REGISTER 50h 00h Read/Write
Address Offset: Default Value: Access:
The PCON Register enables and disables features related to the PCI bus that are not already covered in the required PCI space. Bit 7:4 3 Reserved. PCI Concurrency Enable (PCE). 1=CPU can access DRAM and L2 while a non-PIIX4 PCI master is targeting Peer PCI devices. 0 (default)=CPU is held off of the bus during all PCI master cycles. This bit should be set to 1 by BIOS during normal operation. Reserved. Description
2:0
3.1.15.
CCCACHE CONTROL REGISTER 52h SSSS0010 (S=Strapping option) Read/Write
Address Offset: Default Value: Access:
This 8-bit register defines the secondary cache operations. The CC register enables and disables the second level cache, adjusts cache size, selects the cache write policy, selects the caching policy when CACHE# is negated on reads, informs the MTXC how the SRAMs are connected, and defines the cache SRAM type. After a hard reset, CC[7:4] reflect the signal levels on the Host address lines A[31:28]. Bit 7:6 Description Secondary Cache Size (SCS). This field reflects the inverted signal level on the A[31:30] pins at the rising edge of the RESET signal. The options are: Bits[7:6] 00 01 10 11 Secondary Cache Size Cache not populated 256 Kbytes 512 Kbytes Reserved
The RESET values can be overwritten with subsequent writes to the CC Register. NOTE 1. When bits[7:6]=00, the secondary cache is disabled. 2. When bits[7:6]00, the FLCE bit must also be set to 1 (L2 cache cannot be enabled unless the L1 cache is also enabled).
PRELIMINARY
29
Extended Temperature 82439TX (MTXC) Datasheet
Bit 5:4
Description L2 SRAM Type (L2SRAMT). This field reflects the inverted signal level on the A[29:28] pins at the rising edge of the RESET signal. The RESET values can be overwritten with subsequent writes to the CC Register. The options are: Bits[5:4] 00 01 10 11 SRAM Type Pipelined Burst SRAM Reserved Reserved Two banks of Pipelined Burst NOTE When 512k, Two Banks of Pipelined Burst mode is selected (SCS = 10 and L2SRAMT = 11), NA# will be delayed by one HCLK during burst reads from L2 to ensure that the active bank is not de-selected too early by pipelining a cycle to the opposite bank. Pipelined Burst SRAM (bits 5:4 = 00) also applies to a 512k L2 size when two 64kx32 SRAM devices are used.
3
NA Disable (NAD). 1=Disable. 0=Enable. When disabled, MTXC never asserts the NA# pin. When enabled, NA# assertion is dependent on the cache type and size selected (via SRAMT, SCS). Note that NAD must be set to 1 if the NA# pin of the MTXC is not connected to the processor. This bit should be set to 0 for normal operation in systems that connect NA# to the processor. Reserved. Secondary Cache Force Miss or Invalidate (SCFMI). When set to a 1, the L2 hit/miss detection is disabled, and all tag lookups result in a miss. If the L2 is enabled, then the cycle is processed as a miss (as described in Chapter 4.2). If the L2 is populated but disabled (FLCE=0), then when SCFMI is set to a 1, any CPU read cycle will invalidate the selected tag entry. When SCFMI is set to a 0, normal L2 cache hit/miss detection and cycle processing occurs. Software can flush the cache (cause all modified lines to be written back to DRAM) by setting SCFMI to a 1 with the L2 enabled (non-zero SCS, FLCE=1), and reading all L2 cache tag address locations. See FLCE bit description for FLCE/SCFMI interaction.
2 1
0
First Level Cache Enable (FLCE). 1=Enable. 0=Disable. When FLCE=1, the MTXC responds to CPU cycles with KEN# asserted for cacheable memory cycles. When FLCE=0, KEN# is always negated. This prevents new cache line fills to either the first level or second level cache. The FLCE/SCFMI interaction is summarized below. Note that "Normal L2 operation" is further dependent on the SCS field programming. FLCE 0 0 1 1 SCFMI 0 1 0 1 L2 Result L2 disabled L2 disabled, MTXC tag invalidate on reads Normal L2 operation (dependent on SCS) L2 enabled, MTXC miss forced on reads/writes (Note that writes to the cache are also forced as misses, making it possible to create incoherent DRAM/L2 data.)
PRELIMINARY
30
Extended Temperature 82439TX (MTXC) Datasheet
3.1.16. CECEXTENDED CACHE CONTROL REGISTER 53h 14h Read/Write, Read Only
Address Offset: Default Value: Access:
This 8-bit register defines the refresh rate (in HCLKs) for a DRAM CACHE L2 cache implementation, if enabled. Bit 7:6 5 4:0 Reserved DRAM CACHE L2 Present (ML2)RO. When ML2=1, an L2 DRAM CACHE is present. DRAM Cache L2 Refresh Timer (MCRT)R/W. These bits determine the time the MTXC remains idle during a DRAM cache refresh sequence. The smallest value for the MRCT must be 04h; otherwise, the MTXC will not function properly. The default value sets the timer refresh to 20 HCLKs. During normal operation, the value programmed into register offset 53h should not be changed from its default value of 14h. Description
PRELIMINARY
31
Extended Temperature 82439TX (MTXC) Datasheet
3.1.17. SDRAMCSDRAM CONTROL REGISTER 54-55h 0000h Read/Write Description Reserved. Special SDRAM Mode Select (SSMS). These bits select 1 of 4 special SDRAM modes for testing and initialization. Note that the NOP command must be programmed first before any other command can be issued. After the DRAM detection process has completed, bits[7:5] must remain at "000" during normal DRAM operation. Bits[8:6] 000 001 010 Mode Normal SDRAM mode (default). NOP Command Enable (NOPCE). This mode forces all CPU cycles to DRAM to generate a SDRAM NOP command on the memory interface. All Banks Precharge Command Enable (ABPCE). This setting enables a mode where all CPU cycles to DRAM are converted to an all banks precharge command on the memory interface. Used for BIOS Detection algorithm. Mode Register Command Enable (MRCE). This setting enables a mode where all CPU cycles to DRAM are converted into MRS commands to the memory interface. The command is driven on the MA[11:0] lines. MA[2:0] needs to be always driven to 010 for burst of 4 mode. MA3 needs to be always driven to 1 for interleave wrap type mode. MA4 needs to be driven to the value in the CAS# Latency bit. MA[6:5] needs to be always driven to 01. MA[11:7] needs to be always driven to 00000. The BIOS will select an appropriate host address for each Row of memory such that the right commands are generated on the Memory Address MA[11:0] lines. The BIOS needs to be cognizant of the mapping of the Host addresses to Memory addresses. e.g. A Host address of 1D0h will set up the Mode registers in Row 0 of SDRAM with Burst length of 4, Wrap type of interleaved, and CAS latency of 3. CBR Cycle Enable (CBRC). This setting enables a mode where all CPU cycles to DRAM are converted to SDRAM CBR refresh cycles on the memory interface. Reserved Reserved
Address Offset: Default Value: Access: Bit 15:9 8:6
011
100 101 11X 5
RAS# to CAS# Override (RCO). When set to 1, and the CL bit (CAS Latency) is 0 (CAS Latency=3), then a RAS# to CAS# delay of 2 HCLKs is provided for SDRAM. When set to 0, a RAS# to CAS# delay for SDRAM is determined by the CL bit. CAS# Latency (CL). When set to 1, a CAS# latency of 2 is used for all SDRAM cycles. When reset to 0, CAS# latency of 3 is used for all SDRAM cycles. RAS# Timing (RT). This bit controls RAS# precharge, RAS# active to precharge time and Refresh to RAS# active delay (in HCLKs): Bit 3 0 1 RAS# Precharge 3 3 RAS# act. to Precharge 5 4 Refresh to RAS# act. 8 7
4 3
2
Reserved.
PRELIMINARY
32
Extended Temperature 82439TX (MTXC) Datasheet
Bit 1
Description 64-Mbit Technology Enable (64MTEN). 1=Enable. 0=Disable. When set to 0, the MTXC does not support 64-Mbit SDRAM devices. In this mode, the MTXC supports 4-Mbit, 16-Mbit, and 64Mbit technology for EDO/FPM systems and 4 Mbit and 16 Mbit for SDRAM systems (i.e., 64 Mbit not supported in SDRAM systems). When set to 1, the MTXC supports 4 Mbit, 16 Mbit, and 64 Mbit for both SDRAM and EDO/FPM devices. In this mode, the RAS5#/CS5# signal becomes RAS5#/CS5#/MA13, RAS4#/CS4# becomes RAS4#/CS4#/BA1, and KRQAK/CS4_64# becomes CS4_64#. CS4_64# (fifth row) function is provided if this signal is set to 1 and DRAM Cache is not present in the system (indicated by a 0 in bit 5, register 53h). Reserved.
0
Table 4 lists the CAS# Latency, RAS# to CAS#, RAS# Precharge, RAS# Active to Precharge and Refresh to RAS# active programmable timings. Table 4. Programming Timings Operating Frequency 60/66 MHz 60/66 MHz 60/66 MHz CAS Latency (CL) 3 HCLKs 3 HCLKs 2 HCLKs RAS# to CAS# (Trcd) 3 HCLKs 2 HCLKs 2 HCLKs RAS# Precharge (Trp) 3 HCLKs 3 HCLKs 3 HCLKs RAS# active to Precharge (Tras) 5 HCLKs 5 HCLKs 4 HCLKs Refresh to RAS# (Trc) 8 HCLKs 8 HCLKs 7 HCLKs Register 54h Bits[5:3] 000 100 011
PRELIMINARY
33
Extended Temperature 82439TX (MTXC) Datasheet
3.1.18. DRAMECDRAM EXTENDED CONTROL REGISTER 56h 52h Read/Write
Address Offset: Default Value: Access:
This 8-bit register contains additional controls for main memory DRAM operating modes and features. Bit 7 6 5 Reserved. Refresh RAS# Assertion(RRA). 1=5 clocks (RAS# asserted for Refresh cycles). 0=4 clocks. Fast EDO Lead Off (FELO). 1=Enables fast timing EDO read cycles. 0=Disable. This is valid for EDO DRAMs only (in both a synchronous cache and a Cacheless system). This result is a 1 HCLK pull-in for all read leadoff latencies for EDO DRAMs. (i.e., Page hits, Page misses, and Row Misses). This bit must be 0 if any of the DRAM rows is populated with FPM DRAMs. Speculative Lead Off (SLD). If set to 0, the DRAM Controller read request is presented before the final memory target (Cache/DRAM/PCI) has been decoded by the MTXC. This results in a 1 HCLK pull-in for all read leadoff latencies. Note that if the cycle does not actually target DRAM, the DRAM cycle is later terminated. The SLD bit applies to EDO/FPM and SDRAM. This bit should be set to 1 in systems with a L2 cache and to 0 for systems without a L2 cache Reserved. Memory Address Drive Strength (MAD). This field controls the strength of the output buffers driving the MA, SRASx#, SCASx#, MWEx# and CKEx pins. It is recommended that series termination or undershoot and overshoot diodes be used on these lines. Bit[2:1] 00 01 10 11 MA[13,11:0], BA1 10 mA 10 mA 16 mA 16 mA SRAS[A,B],SCAS[A,B], MWEx#, CKEx 10 mA 16 mA 10 mA 16 mA Description
4
3 2:1
Setting Memory Address Drive Strength: 1 Row 2 Row 3 Row 00 00 11 * Assuming All Rows are buffered 0 Reserved. 4 Row 11 5 Row 11 6 Row 01*
PRELIMINARY
34
Extended Temperature 82439TX (MTXC) Datasheet
3.1.19. DRAMCDRAM CONTROL REGISTER 57h 01h Read/Write
Address Offset: Default Value: Access:
This 8-bit register controls main memory DRAM operating modes and features. Bit 7:6 Description Hole Enable (HEN). This field enables a memory hole in DRAM space. CPU cycles matching an enabled hole are passed on to PCI. PCI cycles matching an enabled hole will be ignored by the MTXC (no DEVSEL#). Note that a selected hole is not remapped. Bits[7:6] 00 01 10 11 5 4 Reserved. Enhanced Paging Disable (EPD). 1=MTXC keeps page open until a page/row miss. When EPD=0, the MTXC uses additional information to keep the DRAM page open when the host may be "right back". See DRAM section for additional information. This bit should be set to 0 for normal operation. EDO Detect Mode Enable (EDME). 1=Enables a special timing mode for BIOS to detect EDO DRAM type on a bank-by-bank basis. Once all DRAM row banks have been tested for EDO, the EDME bit should be set to 0. Otherwise, performance will be seriously impacted. DRAM Refresh Rate (DRR). The DRAM refresh rate for "FPM/EDO only" DRAM subsystem is adjusted according to the value selected by this field. DRAM refresh is implemented using SUSCLK. Bits[2:0] 000 001 010 011 100 101 110 111 DRAM Refresh Rate Refresh Disabled (results in the eventual loss of DRAM data) 15.6 s 31.2 s (for EDO/FPM only memory subsystem) 64.4 s (for EDO/FPM only memory subsystem) 125 s (for EDO/FPM only memory subsystem) 256 s (for EDO/FPM only memory subsystem) Reserved Reserved NOTES 1. If any of the row is populated with SDRAMs, this field must be set to 15.6 s refresh rate. 2. Selecting refresh rate of 125 s or 256 s may violate the max RAS# active time DRAM specification. It is up to the system designer to make sure this does not happen. Hole Enabled None 512 KB-640 KB (128 Kbytes) 15 MB-16 MB (1 Mbyte) 14 MB-16 MB (2 Mbytes)
3
2:0
PRELIMINARY
35
Extended Temperature 82439TX (MTXC) Datasheet
3.1.20. DRAMTDRAM TIMING REGISTER 58h 00h Read/Write
Address Offset: Default Value: Access:
This 8-bit register controls main memory DRAM timings. For SDRAM specific timing control, see the SDRAMC timing register definition. Bit 7 6:5 Reserved. DRAM Read Burst Timing (DRBT). The DRAM read burst timings are controlled by the DRBT field. Slower rates may be required in certain system designs to support loose layouts or slower memories. Most system designs will be able to use one of the faster burst mode timings. The timing used depends on the type of DRAM on a per-bank basis, as indicated by the DRT register. DRBT 00 01 10 11 4:3 EDO Burst Rate x444 x333 x222 Reserved FPM Burst Rate x444 x444 x333 Reserved Description
DRAM Write Burst Timing (DWBT). The DRAM write burst timings are controlled by the DWBT field. Slower rates may be required in certain system designs to support loose layouts or slower memories. Most system designs will be able to use one of the faster burst mode timings. DWBT 00 01 EDO/FPM Burst Rate x444 x333 DWBT 10 11 EDO/FPM Burst Rate x222 Reserved
2 1:0
Reserved. DRAM Leadoff Timing (DLT). The DRAM leadoff timings are controlled by the DLT bits. Slower leadoffs may be required in certain system designs to support loose layouts or slower memories. The Row Miss leadoff timings are summarized below for EDO/FPM reads and writes. Changing DLT affects the Row Miss and Page Miss timings only (e.g., DLT=01 is one clock faster than DLT=00 on Row Miss and Page Miss timings). These bit control MA setup to CAS# assertion. DLT does not affect page hit timings. Thus, DLT=00 or DLT=01 has same page hit timings for reads and writes (e.g., for reads, it would be 10-3=7 clocks for DLT=00 or DLT=01) DLT 00 01 10 11 Read Leadoff 11 10 11 10 Write Leadoff 7 6 7 6 RAS# Precharge 3 3 4 4 RAS-to-CAS Delay 4 3 4 3
SLD and FELO bits have cumulative effect on the leadoff timings. The above leadoff represent timings with SLD=1 and FELO=0.
PRELIMINARY
36
Extended Temperature 82439TX (MTXC) Datasheet
3.1.21.
PAMPROGRAMMABLE ATTRIBUTE MAP REGISTERS (PAM[6:0]) 59h (PAM0) (5Fh (PAM6) 00h Read/Write 8 bits (each register)
Address Offset: Default Value: Attribute: Size:
The MTXC allows programmable memory and cacheability attributes on 14 memory segments of various sizes in the 640-Kbytes to 1-Mbyte address range. Seven Programmable Attribute Map (PAM) Registers are used to support these features. Three bits are used to specify L1 cacheability and memory attributes for each memory segment. These attributes are: RE WE CE Read Enable. When RE=1, the CPU read accesses to the corresponding memory segment are directed to main memory. Conversely, when RE=0, the CPU read accesses are directed to PCI. Write Enable. When WE=1, the CPU write accesses to the corresponding memory segment are directed to main memory. Conversely, when WE=0, the CPU write accesses are directed to PCI. Cache Enable. When CE=1, the corresponding memory segment is L1 cacheable. CE must not be set to 1 when RE is reset to 0 for any particular memory segment. When CE=1 and WE=0, the corresponding memory segment is cached in the first level cache only on CPU code read cycles.
The RE and WE attributes permit a memory segment to be Read Only, Write Only, Read/Write, or disabled. For example, if a memory segment has RE=1 and WE=0, the segment is Read Only. The characteristics for memory segments with these read/write attributes are described in Table 5. Table 5. Attribute Definition Read/Write Attribute Read Only Definition Read cycles: CPU cycles are serviced by the DRAM in a normal manner. Write cycles: CPU initiated write cycles are ignored by the DRAM interface as well as the cache. Instead, the cycles are passed to PCI for termination. Areas marked as Read Only are L1 cacheable for Code accesses only. These regions are not cached in the second level cache. Write Only Read cycles: All read cycles are ignored by the DRAM interface as well as the second level cache. CPU-initiated read cycles are passed onto PCI for termination. The write only state can be used while copying the contents of a ROM, accessible on PCI, to main memory for shadowing, as in the case of BIOS shadowing. Write cycles: CPU write cycles are serviced by the DRAM and L2 cache in a normal manner. Read/Write Disabled This is the normal operating mode of main memory. Both read and write cycles from the CPU and PCI are serviced by the DRAM and L2 cache interface. All read and write cycles to this area are ignored by the DRAM and cache interface. These cycles are forwarded to PCI for termination.
PRELIMINARY
37
Extended Temperature 82439TX (MTXC) Datasheet
Each PAM Register controls two regions, typically 16 Kbytes in size. Each of these regions has a 4-bit field. The four bits that control each region have the same encoding and are defined in Table 6. PCI master access to DRAM space is also controlled by the PAM Registers. If the PAM programming indicates a region is writeable, then PCI master writes will be accepted (DEVSEL# generated). If the PAM programming indicates a region is readable, PCI master reads will be accepted. If a PCI write to a nonwriteable DRAM region, or a PCI read to a non-readable DRAM region is seen, the MTXC will not accept the cycle (DEVSEL# will not be asserted). PCI master accesses to enable memory hole regions will not be accepted. Table 6. Attribute Bit Assignment Bits [7, 3] Reserved x x x x x x Bits [6, 2] Cache Enable x 0 1 0 0 1 Bits [5, 1] Write Enable 0 0 0 1 1 1 Bits [4, 0] Read Enable 0 1 1 0 1 1 Description DRAM disabled, accesses directed to PCI read only, DRAM write protected, non-cacheable read only, DRAM write protected, L1 cacheable for code accesses only write only read/write, non-cacheable read/write, cacheable
As an example, consider a BIOS that is implemented on the expansion bus. During the initialization process the BIOS can be shadowed in main memory to increase the system performance. When a BIOS is shadowed in main memory, it should be copied to the same address location. To shadow the BIOS, the attributes for that address range should be set to write only. The BIOS is shadowed by first doing a read of that address. This read is forwarded to the expansion bus. The CPU then does a write of the same address, which is directed to main memory. After the BIOS is shadowed, the attributes for that memory area are set to read only so that all writes are forwarded to the expansion bus. Table 7. PAM Register and Associated Memory Segments PAM Reg. PAM0[3:0] PAM0[7:4] PAM1[3:0] PAM1[7:4] PAM2[3:0] PAM2[7:4] PAM3[3:0] PAM3[7:4] PAM4[3:0] PAM4[7:4] PAM5[3:0] Attribute Bits Reserved R R R R R R R R R R CE CE CE CE CE CE CE CE CE CE WE WE WE WE WE WE WE WE WE WE RE RE RE RE RE RE RE RE RE RE 0F0000h - 0FFFFFh 0C0000h - 0C3FFFh 0C4000h - 0C7FFFh 0C8000h - 0CBFFFh BIOS Area ISA Add-on BIOS ISA Add-on BIOS ISA Add-on BIOS Memory Segment Comments 59h 59h 5Ah 5Ah 5Bh 5Bh 5Ch 5Ch 5Dh 5Dh 5Eh Offset
0CC000h - 0CFFFFh ISA Add-on BIOS 0D0000h - 0D3FFFh 0D4000h - 0D7FFFh 0D8000h - 0DBFFFh ISA Add-on BIOS ISA Add-on BIOS ISA Add-on BIOS
0DC000h - 0DFFFFh ISA Add-on BIOS 0E0000h - 0E3FFFh BIOS Extension
PRELIMINARY
38
Extended Temperature 82439TX (MTXC) Datasheet
Table 7. PAM Register and Associated Memory Segments PAM Reg. PAM5[7:4] PAM6[3:0] PAM6[7:4] R R R Attribute Bits CE CE CE WE WE WE RE RE RE Memory Segment 0E4000h - 0E7FFFh 0E8000h - 0EBFFFh 0EC000h - 0EFFFFh Comments BIOS Extension BIOS Extension BIOS Extension 5Eh 5Fh 5Fh Offset
NOTES: The CE bit should not be changed while the L2 cache is enabled. DOS Application Area (00000h-9FFFh) Read, write, and cacheability attributes are always enabled and are not programmable for the 0-640-Kbytes DOS application region. Video Buffer Area (A0000h-BFFFFh) This 128-Kbytes area is not controlled by attribute bits. CPU-initiated cycles in this region are always forwarded to PCI for termination. This area is not cacheable. Expansion Area (C0000h-DFFFFh) This 128-Kbytes area is divided into eight 16-Kbytes segments. Each segment can be assigned one of four Read/Write states: read-only, write-only, read/write, or disabled Memory that is disabled is not remapped. Cacheability status can also be specified for each segment. Extended System BIOS Area (E0000h-EFFFFh) This 64-Kbytes area is divided into four 16-Kbytes segments. Each segment can be assigned independent cacheability, read, and write attributes. Memory segments that are disabled are not remapped elsewhere. System BIOS Area (F0000h-FFFFFh) This area is a single 64-Kbytes segment. This segment can be assigned cacheability, read, and write attributes. When disabled, this segment is not remapped. Extended Memory Area (100000h-FFFFFFFFh) The extended memory area can be split into several parts: * * * Flash BIOS area from 4 Gbytes to 4 Gbytes-512 Kbytes (aliased on ISA at 16 Mbytes-15.5 Mbytes) DRAM Memory from 1 Mbytes to a maximum of 512 Mbytes PCI Memory space from the top of DRAM to 4 Gbytes-512 Kbytes
On power-up or reset the CPU vectors to the Flash BIOS area, mapped in the range of 4 Gbytes to 4 Gbytes-512 Kbytes. This area is physically mapped on the expansion bus. Since these addresses are in the upper 4-Gbytes range, the request is directed to PCI. The DRAM memory space can occupy extended memory from a minimum of 1 Mbytes up to 256 Mbytes. This memory is cacheable. PCI memory space from the top of main memory to 4 Gbytes is always non-cacheable.
PRELIMINARY
39
Extended Temperature 82439TX (MTXC) Datasheet
3.1.22. DRBDRAM ROW BOUNDARY REGISTERS 60-65h 02h Read/Write
Address Offset: Default Value: Access:
The MTXC supports 6 rows of DRAM. Each row is 64-bits wide. The DRAM Row Boundary Registers define upper and lower addresses for each DRAM row. Contents of these 8-bit registers represent the boundary addresses in 4-Mbytes granularity. DRB0=Total amount of memory in row 0 (in 4 Mbytes) DRB1=Total amount of memory in row 0 + row 1 (in 4 Mbytes) DRB2=Total amount of memory in row 0 + row 1 + row 2 (in 4 Mbytes) DRB3=Total amount of memory in row 0 + row 1 + row 2 + row 3 (in 4 Mbytes) DRB4=Total amount of memory in row 0 + row 1 + row 2 + row 3 + row 4 (in 4 Mbytes) DRB5=Total amount of memory in row 0 + row 1 + row 2 + row 3 + row 4 + row 5 (in 4 Mbytes) The DRAM array can be configured with 512-KB, 1-MB, 4-MB, or 16-MB deep by 32- or 36-bit wide SIMMs. Each register defines an address range that will cause a particular RAS# line to be asserted (e.g., if the first DRAM row is 8 Mbytes, accesses within the 0 to 8-Mbytes range will cause RAS0# to be asserted). NOTE When programming the DRB registers, the following programming consideration must be followed: When DRB3 is written, DRB4 and DRB5 are also modified with the value written into DRB3. When DRB4 is written, DRB5 is also modified with the value written into DRB4. To avoid data corruption in the DRB4 and DRB5 registers, program DRB3 first, followed by DRB4 and then DRB5. If either DRB3 or DRB4 are written, this sequence should be followed. Bit 7 6:0 Reserved. Row Boundary Address. This 7-bit value is compared against the address lines A[28:22] to determine the upper address limit of a particular row (i.e., DRB minus previous DRB=row size). Description
Row Boundary Address These 8 bit values represent the upper address limits of the 6 rows (i.e., this row minus previous row=row size). Unpopulated rows have a value equal to the previous row (row size=0). DRB5 reflects the maximum amount of DRAM in the system. The top of memory is determined by the value written into DRB5. If DRB5 is greater than 256 Mbytes, then 256 Mbytes of DRAM are available. BIOS must make sure that the DRB registers do not reflect more than 256M of Main memory. As an example of a general purpose configuration where 3 physical rows are configured for either singlesided or double-sided SIMMs, the memory array would be configured like the one shown in Figure 3. In this configuration, the MTXC drives two RAS# signals directly to the SIMM rows. If single-sided SIMMs are populated, the even RAS# signal is used and the odd RAS# is not connected. If double-sided SIMMs are used, both RAS# signals are used.
PRELIMINARY
40
Extended Temperature 82439TX (MTXC) Datasheet
RAS5# RAS4# RAS3# RAS2# RAS1# RAS0#
SIMM-5 Back SIMM-5 Front SIMM-3 Back SIMM-3 Front SIMM-1 Back SIMM-1 Front
SIMM-4 Back SIMM-4 Front SIMM-2 Back SIMM-2 Front SIMM-0 Back SIMM-0 Front
DRB5 DRB4 DRB3 DRB2 DRB1 DRB0
CAS7# CAS5# CAS3# CAS1# CAS6# CAS4# CAS2# CAS0#
simm_drb
Figure 3. SIMMs and Corresponding DRB Registers
The following 2 examples describe how the DRB Registers are programmed for cases of single-sided and double-sided SIMMs on a motherboard having a total of four 8-byte or eight 4-byte SIMM sockets. Example #1 The memory array is populated with four single-sided 1 MB x 32 SIMMs, a total of 16 MBytes of DRAM. Two SIMMs are required for each populated row making each populated row 8 Mbytes in size. DRB0=02h DRB1=04h DRB2=04h DRB3=04h DRB4=04h DRB5=04h populated (2 SIMMs, 8 Mbytes this row) populated (2 SIMMs, 8 Mbytes this row) empty row empty row empty row empty row
Example #2 As an another example, the memory array is populated with two 2 Mbytes x 32 double-sided SIMMs (one row), and four 4 Mbytes x 32 single-sided SIMMs (two rows), yielding a total of 96 Mbytes of DRAM. The DRB Registers are programmed as follows: DRB0=04h DRB1=08h DRB2=10h DRB3=18h DRB4=18h DRB5=18h populated with 16 Mbytes, 1/2 of double-sided SIMMs the other 16 Mbytes of the double-sided SIMMs populated with 32 Mbytes, one of the sided SIMMs the other 32 Mbytes of single-sided SIMMs empty row empty row
PRELIMINARY
41
Extended Temperature 82439TX (MTXC) Datasheet
3.1.23. DRTHDRAM ROW TYPE REGISTER HIGH 67h S0000000 Read/Write
Address Offset: Default Value: Access:
This 8-bit register identifies the type of DRAM (EDO, SPM (standard page mode)), or SDRAM (synchronous DRAM) used in rows 4 and 5 and should be programmed by BIOS for optimum performance if EDO DRAMs or SDRAMs are used. The MTXC uses these bits to determine the correct cycle timing to use before a DRAM cycle is run. Bit 7 of this register is used for Host Frequency Detection (HFD). Bit 2 of this register is used to determine the muxing results of CKE/MAA0 and CKEB/MAA1. NOTE This register should not be written while DRAM refresh is enabled. Bit 7 Description Host Frequency Detection (HFD). 1=66 MHz. 0=60 MHz. This bit is initialized to the inverted level on the A27 signal at the rising edge of the RST#. Since A27 pin contains an internal weak pulldown, unless an external resistor exits, the field is initialized to 1, indicating 66 MHz. Subsequent writes to this field will override the reset strap value. BIOS can use the value to determine if the system is 60 MHz (external pull-up) or 66 MHz (no strapping). DRAM Row Type (DRT). The DRT bits select the DRAM type installed in each physical DRAM Row. Each one-of-four bit pairs in this register corresponds to the DRAM row identified by the corresponding DRB register. DRT Bits 5,1 4,0 DRT 0,0 0,1 1,0 1,1 6,3 2 Reserved. Memory Address Select Enable (MASELEN). When this bit is set to 1, CKE and CKEB are used to propagate the second copy of the MA0 and MA1 lines. CKE is muxed with MAA0 and CKEB is muxed with MAA1. When this bit is set to 0, the CKE and CKEB functionality is propagated across these lines. This bit defaults to 0 and BIOS must set it to 1 to take advantage of the second copy of the MA0 and MA1 lines. DRAM Row 5 4 DRAM Type value definitions SPM DRAM EDO DRAM SDRAM Reserved
5:4, 1:0
PRELIMINARY
42
Extended Temperature 82439TX (MTXC) Datasheet
3.1.24. DRTL--DRAM ROW TYPE REGISTER LOW 68h 00h Read/Write
Address Offset: Default Value: Access:
This 8-bit register identifies the type of DRAM (EDO, SPM (standard page mode)), or SDRAM (synchronous DRAM) used in rows 0 to 3 and should be programmed by BIOS for optimum performance if EDO DRAM's or SDRAMs are used. The hardware uses these bits to determine the correct cycle timing to use before a DRAM cycle is run. Bit 7:0 Description DRAM Row Type (DRT). The DRT bits select the DRAM type installed in each physical DRAM Row. Each one-of-four bit pairs in this register corresponds to the DRAM row identified by the corresponding DRB register. DRT Bits 7,3 6,2 5,1 4,0 DRT 0,0 0,1 1,0 1,1 DRAM Row 3 2 1 0 DRAM Type value definitions SPM DRAM EDO DRAM SDRAM reserved
3.1.25.
MTTMULTI-TRANSACTION TIMER REGISTER (RESERVED TEST MODE REGISTER) 70h 20h Read/Write
Address Offset: Default Value: Access:
MTT is an 8-bit register that controls the amount of time that the MTXC's arbiter allows a PCI initiator to perform multiple transactions on the PCI bus. The MTT guarantees the minimum time, measured in PCLKs, that the PCI agent retains the ownership of the PCI bus from the initial assertion of grant. Bit 7:2 Description MTT Count value. The number of clocks programmed in the MTT represents the guaranteed time slice (in PCLKs) allotted to the current agent, after which the MTXC will grant the bus as soon as another PCI agent requests the bus. The value of 00h disables this function. The count value should be set to multiples of 4 (i.e., 2 lsbs are ignored). Reserved. Hardwired to 0. (i.e., counter has a resolution of 4 PCLKs)
1:0
PRELIMINARY
43
Extended Temperature 82439TX (MTXC) Datasheet
3.1.26. ESMRAMCEXTENDED SYSTEM MANAGEMENT RAM CONTROL REGISTER 71h 00h Read/Write
Address Offset: Default Value: Access:
The Extended SMRAM register controls the configuration of Extended SMRAM space. MTXC supports two types of SMRAM memory: Compatible and Extended. The Compatible SMRAM (C_SMRAM) memory provides an uncacheable SMRAM memory space below 1 Mbytes in the A and B segments. The Extended SMRAM (E_SMRAM) memory provides a writeback cacheable SMRAM memory space that is above 1 Mbytes. This register provides the following types of control over SMRAM space: * * * * Where the memory space is located (above 1 Mbytes, below 1 Mbytes) Enabling of SMRAM memory (TSEG, 128 Kbytes, 256 Kbytes, 512 Kbytes or 1 Mbytes of additional SMRAM memory) for Extended SMRAM space only. Cacheability control (for the Extended SMRAM space only) Protection of SMRAM space for non-SMM accesses Bit 7 Description High SMRAM Enable (H_SMRAME). 1=Enable. 0=Disable. This bit enables the high SMRAM memory space to appear in the appropriate physical address locations between 100A0000h and 100F0000h. Extended SMRAM Error (E_SMERR). This bit is set when CPU accesses the defined memory ranges in Extended SMRAM (High Memory and T-segment) while not in SMM space and with the D-OPEN bit=0. It is software's responsibility to clear this bit. The software must write a 1 to this bit to clear it. SMRAM Cache Strategy (SM_CACHE). Hardwired to 0. This bit determines how Extended SMRAM space is cached (writethru or writeback). Since the MTXC supports only writeback for extended SMRAM space, this bit is hardwired to 0. SMRAM_L1_EN (SM_L1). This bit should be set to 1 if Extended SMRAM is being used and the system wishes to L1 writeback cache this memory space. Default value for this bit is 0. SMRAM_L2_EN (SM_L2). This bit should be set to 1 if Extended SMRAM is being used, and there is less than 32 Mbytes of DRAM in the system. Setting of this bit when SM_L1 bit=1 allows the Extended SMRAM to be writeback cached in the L2. Default value for this bit is 0. TSEG_SZ[1-0] (T_SZ). Selects the size of the TSEG memory block, if enabled. This memory is taken from the top of DRAM space, which is no longer claimed by the memory controller (all accesses to this space are sent to the PCI bus if TSEG_EN is set). This memory appears at the physical memory space of 256 Mbytes plus the top of memory (TOM) minus the size of TSEG. This field decodes as follows: Bits[1,0] 00 01 10 11 0 Description (TOM-128 KB) to TOM (TOM-256 KB) to TOM (TOM-512 KB) to TOM (TOM-1 MB) to TOM
6
5
4 3
2:1
TSEG_EN (T_EN). When G_SMRAME=1 and T_EN=1, the TSEG is enabled to appear in the appropriate physical address space.
PRELIMINARY
44
Extended Temperature 82439TX (MTXC) Datasheet
3.1.27. SMRAMCSYSTEM MANAGEMENT RAM CONTROL REGISTER 72h 02h Read/Write
Address Offset: Default Value: Access:
The SMRAMC register controls how accesses to Compatible and Extended SMRAM spaces are treated. MTXC supports two types of SMRAM memory: Compatible and Extended. The Open, Close, and Lock bits function only when G_SMRAME bit is set to a 1. Also, the OPEN bit be reset before the LOCK bit is set. Bit 7 6 Reserved. SMM Space Open (D_OPEN). When D_OPEN=1 and D_LCK=0, the SMM space DRAM is made visible even when SMIACT# is negated. This is intended to help BIOS initialize SMM space. Software should ensure that D_OPEN=1 is mutually exclusive with D_CLS=1. When D_LCK is set to a 1, D_OPEN is reset to 0 and becomes read only. SMM Space Closed (D_CLS). When D_CLS=1, SMM space DRAM is not accessible to data references, even if SMIACT# is asserted. Code references may still access SMM space DRAM. This will allow SMM software to reference "through" SMM space to update the display, even when SMM space is mapped over the VGA range. Software should ensure that D_OPEN=1 is mutually exclusive with D_CLS=1. SMM Space Locked (D_LCK). When D_LCK is set to 1, D_OPEN is reset to 0 and both D_LCK and D_OPEN become read only. D_LCK can be set to 1 via a normal configuration space write but can only be cleared by a power-on reset. The combination of D_LCK and D_OPEN provide convenience with security. The BIOS can use the D_OPEN function to initialize SMM space and then use D_LCK to "lock down" SMM space in the future so that no application software (or BIOS itself) can violate the integrity of SMM space, even if the program has knowledge of the D_OPEN function. Global SMRAM Enable (G_SMRAME). If set to a 1, then Compatible SMRAM functions is enabled, providing 128 KB of DRAM accessible at the A0000h address while in SMM (ADS# with SMIACT#). To enable Extended SMRAM function this bit has be set to 1. Refer to the section on SMM for more details. Compatible SMM Space Base Segment (C_BASE_SEG). This field programs the location of SMM space. SMM DRAM is not remapped. It is simply "made visible" if the conditions are right to access SMM space; otherwise, the access is forwarded to PCI. C_BASE_SEG=010 selects the SMM space as A0000-BFFFFh. All other values are reserved. PCI initiators are not allowed to access to SMM space. These bits are hardwired to 010. Description
5
4
3
2:0
PRELIMINARY
45
Extended Temperature 82439TX (MTXC) Datasheet
Table 8 summarizes the operation of SMRAM space cycles targeting the SMI space addresses. Table 8. SMRAM Space Cycles G_ S M R A M E
0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
D_ L C K
D_ C L S
D_ O P E N
S M I A C T#
x 0 0 0 0 1 x x x x 0 0 0 0 x 0 0 0 0 1 0 0 0 0
H_ S M R A M E
x 0 0 1 1 x 0 0 1 1 0 0 1 1 x 0 0 1 1 x 0 0 1 1
T S E G_ E N
x 0 1 0 1 x 0 1 0 1 0 1 0 1 x 0 1 0 1 x 0 1 0 1 ADRAM ADRAM APCI APCI APCI ADRAM ADRAM APCI APCI APCI ADRAM ADRAM APCI APCI APCI ADRAM ADRAM APCI APCI ADRAM ADRAM ADRAM APCI
Code Fetch
Data Access
x 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
x 0 0 0 0 x 0 0 0 0 1 1 1 1 1 0 0 0 0 x 1 1 1 1
x 0 0 0 0 0 1 1 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0
SPCI SPCI SPCI
TPCI TPCI TDRAM
APCI ADRAM ADRAM APCI APCI APCI ADRAM ADRAM APCI APCI APCI APCI APCI APCI ADRAM ADRAM APCI APCI APCI APCI APCI APCI APCI
SPCI SPCI SPCI
TPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM SPCI SPCI SPCI TPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM SPCI SPCI SPCI TPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM SPCI SPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM SPCI SPCI SPCI SPCI Invalid SPCI SPCI TPCI TPCI TPCI TPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM Invalid SPCI SPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM SPCI SPCI SPCI TPCI TPCI TDRAM
SDRAM TPCI SDRAM TDRAM SPCI SPCI SPCI SPCI SPCI TPCI TPCI TPCI TPCI TPCI
SDRAM TPCI SDRAM TDRAM
NOTES: 1. A=A Segment, S=100A0000h to 100FFFFFh, and T=T Segment. The Code Fetch and Data Access columns indicate whether the access is to the PCI bus or to main memory DRAM.
PRELIMINARY
46
Extended Temperature 82439TX (MTXC) Datasheet
3.1.28. MCTLMISCELLANEOUS CONTROL REGISTER 79h 00h Read/Write Description Reserved. ACPI Control Register Enable (ACRE). 0=Any CPU access to I/O address 0022h is passed on to the PCI bus. 1=Any CPU access to I/O address 0022h is processed internally in the MTXC. This bit must be set to 1 before accessing the "Arbiter Disable" bit in the PM2_CNTRL Register (0022h). Suspend Refresh Type (SRT). 0=CBR refresh. 1=Self refresh. This bit determines what type of DRAM refresh is used during Power On Suspend (POS) or Suspend to DRAM modes. This bit applies to EDO/FPM DRAM only. SDRAM always uses self refresh, regardless of the state of this bit. Normal Refresh Enable (NREF_EN). Setting this bit to 1 switches MTXC from suspend refresh to normal refresh. After the reset, this bit must be set by software executing out of EPROM. MTXC waits for this bit to be set before exiting out of suspend refresh mode. Reserved. Internal Clock Control (Gated Clock) Disable (ICC). 1=Disable. 0=Enable. This bit, when set to 0, allows the MTXC to reduce its power consumption (via turning off its internal clocks, to specific interfaces) when in chip standby mode. This bit defaults to 0. Reserved.
Address Offset: Default Value: Access: Bit 7 6
5
4
3 2
1:0
PRELIMINARY
47
Extended Temperature 82439TX (MTXC) Datasheet
4.0. 4.1.
FUNCTIONAL DESCRIPTION Host Interface
The Host Interface of the MTXC is designed to support the Pentium microprocessor. The host interface of the MTXC supports 60-, and 66-MHz bus speeds. The Intel 430TX PCIset supports the Pentium microprocessor with a full 64-bit data bus, 32-bit address bus, and associated internal writeback cache logic. Host bus addresses are decoded by the MTXC for accesses to main memory, PCI memory, and PCI I/O. The MTXC also supports the pipelined addressing capability of the Pentium microprocessor.
4.2.
Secondary Cache Interface
The MTXC integrates a high performance writeback second level cache controller using internal/external tags and provides a full first level and second level cache coherency mechanism. The second level cache is direct mapped, nonsectored, and supports a writeback, no write allocate (lines are not allocated on write misses) write policy. The second level cache can be configured to support either a 256-KB or 512-KB cache using synchronous pipelined burst SRAM or DRAM Cache. One additional PCIset signal (KRQAK) is required to support DRAM Cache. 64-Mbytes cacheability coverage is obtained with 8Kx8 standard SRAM to store the tags for 256-KB configuration. For the 512-KB configurations, a 16Kx8 standard SRAM is used to store the tags and the valid bits for 64-MB cacheability. A second level cache line is 32-bytes wide. In the 256-KB configurations, the second level cache contains 8K lines, while the 512-KB configurations contain 16K lines. Valid and modified status bits are kept on a per line basis. Cacheability of the entire memory space in first level cache is supported, while only the lower 64 MB of main memory is cacheable in the second level cache. Table 9 shows the tag sizes needed to support different sizes of cacheability. Only main memory controlled by the MTXC DRAM interface is cached. PCI memory is not cached. Table 9. Cacheability Cache Size 256 Kbytes 512 Kbytes 8K by 8 bits 16K by 8 bits (including valid bit) Tag Size 64 Mbytes 64 Mbytes Cacheability
The following table shows the different standard SRAM access time requirements for different host clock frequencies. Table 10. SRAM Access Time Requirements Host Clock Frequency (MHz) 60 66 Pipelined Burst Clock-to-Output Access Time (ns) 10 8.5 Tag RAM Cycle Time (ns) 15 15
PRELIMINARY
48
Extended Temperature 82439TX (MTXC) Datasheet
Figure 4 and Figure 5 show the connections between the MTXC and the external tag RAM and data SRAM.
HA[17:5] MTXC TIO[7:0] TW E# 8Kx8 T RAM ag D[7:0] W E# OE# HCLK A[12:0] HA[17:3] COE# CCS# CADS# CAD V# ADS# GW E# BW E# HBE[7:4]# 32Kx32 SRAM CLK A[14:0] OE# CS# ADSC# AD V# ADSP# GW E# BW E# D[31:24] D[23:16] D[15:8] BE[3:0]# GW E# BW E# W E1# HBE[3:0]# BE[3:0]# D[7:0] D[31:24] D[23:16] D[15:8] D[7:0] HD[63:56] HD[55:48] HD[47:40] HD[39:32] HD[31:24] HD[23:16] HD[15:8] HD[7:0] zz zz
mtxc_01
Figure 4. MTXC Connections for 256K Second Level Cache with PBSRAM
Figure 5 shows a 512-KB implementation using four 32Kx32 SRAM. Two 64Kx32 devices could also be used. In this case, HA18 would not be connected to CS2# (i.e., CS2 and CS2# should be connected to an active state). HA18 should be connected to one of the address lines on the 64Kx32 SRAM and is still required for the TAG RAM.
PRELIMINARY
49
Extended Temperature 82439TX (MTXC) Datasheet
HA[18:5] MTXC TIO[7:0] TW E# 16Kx8 Tag RAM D[7:0] W E# OE# HCLK A[13:0] HA[17:3] COE# CCS# CADS# CAD V# ADS# HA18 GND GW E# BW E# HBE[7:4]# 32Kx32 SRAM CLK A[14:0] OE# CS1# ADSC# AD V# ADSP# CS2 CS2# GW E# BW E# BE[3:0]# GW E# BW E# HBE[3:0]# BE[3:0]# zz zz
D[31:24] D[23:16] D[15:8] D[7:0] D[31:24] D[23:16] D[15:8] D[7:0]
HD[63:56] HD[55:48] HD[47:40] HD[39:32] HD[31:24] HD[23:16] HD[15:8] HD[7:0]
32Kx32 SRAM HCLK HA[17:3] COE# CCS# CADS# CAD V# ADS# Vcc HA18 GW E# BW E# HBE[7:4]# GW E# BW E# HBE[3:0]# CLK A[14:0] OE# CS1# ADSC# AD V# ADSP# CS2 CS2# GW E# BW E# BE[3:0]# GW E# BW E# BE[3:0]# zz zz
D[31:24] D[23:16] D[15:8] D[7:0] D[31:24] D[23:16] D[15:8] D[7:0]
HD[63:56] HD[55:48] HD[47:40] HD[39:32] HD[31:24] HD[23:16] HD[15:8] HD[7:0]
mtxc_02
Figure 5. MTXC Connections for 512K Second Level Cache with PBSRAM
PRELIMINARY
50
Extended Temperature 82439TX (MTXC) Datasheet
4.2.1. CLOCK LATENCIES
Table 11 lists the latencies for various processor transfers to and from the second level cache. Table 11. Second Level Cache Latencies with Pipelined Burst SRAM Cycle Type Burst Read Burst Write (write back) Single Read Single Write Pipelined Back-to-Back Burst Reads 3-1-1-1 3-1-1-1 3 3 3-1-1-1,1-1-1-1 (note 1) HCLK Count
NOTES: 1. The back to back cycles do not account for CPU idle clocks between cycles. 4.2.2. SNOOP CYCLES
The snoop (or inquire) cycle is used to probe the first level and second level caches when a PCI master attempts to access main memory. This is done in order to maintain coherency between the first and second level caches and main memory. To maintain optimum PCI bandwidth to DRAM, the MTXC utilizes a snoop ahead algorithm. Once the snoop for the first cache line of a transfer has completed, the MTXC automatically snoops the next sequential cache line. This algorithm enables the MTXC to continue burst transfers across cache line boundaries. Reads Snoop cycles are performed by driving the PCI master address onto the host address bus and asserting EADS#. The processor then performs a tag lookup to determine whether the addressed memory is in the first level cache. If the snoop hit is to a Modified Line in the first level cache (HITM# asserted), then the line in the first level cache is posted to the DRAM Posted Write buffers. The line in the second level cache (if it exists) is invalidated. The line in the first level cache is not invalidated if the INV pin on the CPU is tied to the KEN# signal from the MTXC. KEN#/INV will be driven low by the MTXC with EADS# assertion during PCI master read cycles. At the same time as the first level snoop cycle, the MTXC performs a tag lookup to determine whether the addressed memory is in the second level cache. A hit to a modified line in the second level cache also results in a writeback to DRAM posted write buffers if HITM# is not asserted. The PCI data is serviced from the DRAM after the line has been retired to DRAM. Writes PCI Master write cycles never result in a write directly into the second level cache. A snoop hit to a modified line in either the first or second caches results in a writeback of that line to main memory. If both the first and second level caches have modified lines, then the line is written back from the first level cache. In all cases lines in the first and second level caches are invalidated and the PCI write to main memory occurs after the writeback completes. A PCI master write snoop hit to an unmodified line in either the first or second level caches results in the line being invalidated. KEN#/INV will be driven high by the MTXC with EADS# assertion during PCI master write cycles.
PRELIMINARY
51
Extended Temperature 82439TX (MTXC) Datasheet
4.2.3. DRAM CACHE SECOND LEVEL CACHE MODE
DRAM Cache L2 cache implementation is similar to Pipelined Burst SRAM, except for the addition of the KRQAK bi-direct refresh handshake signal between the MTXC and L2 SRAM. A DRAM Cache type L2 is assumed present when the KRQAK pin is sampled high during the negation of the reset signal. An internal weak pull-down is used on the MTXC KRQAK pin to default to a non DRAM Cache L2 mode, if this pin is left unconnected. An external pull-up (10 k) must be used on KRQAK when DRAM Cache SRAM is used. Note that there is no configuration bit associated with the L2 Pseudo SRAM mode. The SRAM can operate in either master or slave mode via the M/S# strapping bit. In master mode, the SRAM drives the KRQAK pin to request a refresh. A slave device never drives KRQAK, but only monitors it to determine when a refresh period begins. Only one SRAM device within the L2 cache is master enabled. The other SRAM devices must be slaves. During reset, the master SRAM and MTXC tri-state their KRQAK outputs. After the SRAM RESET pin is negated, KRQAK remains tri-stated for one whole refresh interval and is then driven high by the master SRAM. The SRAM signals a refresh request by driving KRQAK low for 1 clock, high the next clock, and then tri-states on the following clock and waits, sampling the KRQAK pin. The MTXC after sampling the SRAM's request on KRQAK and after the SRAM has tri-stated its KRQAK output, waits for a host bus dead clock and grants an L2 refresh by driving its KRQAK pin in an identical fashion to the SRAM's request signaling. When all SRAM's see the refresh grant from the MTXC, they begin their internal refresh cycle for a period of 20 clocks.
4.3.
DRAM Interface
The MTXC integrates a DRAM controller that supports a 64-bit memory array from 4 Mbytes to 256 Mbytes of main memory. The MTXC supports Standard Page Mode (FPM), Extended Data Out (EDO) and Synchronous DRAM (SDRAM) memories using 32-bit wide SIMM modules, 64-bit wide unbuffered DIMM modules and 64-bit wide unbuffered SO-DIMM modules. DRAM parity is not supported, and for loading reasons, parity modules should not be used. All three memory types can be mixed and matched. The MTXC generates all DRAM control signals and multiplexed addresses for the DRAM array. The address and data flows through the MTXC for all DRAM accesses. The DRAM controller interface is fully configurable through a set of control registers. Complete descriptions of these registers are given in the MTXC configuration register description. A brief overview of these registers is provided in this section. The MTXC supports page mode DRAMs and EDO (Extended Data Out) DRAMs; otherwise known as Hyper Page mode. The twelve multiplexed address lines, MA[11:0], allow the MTXC to support 4-Mbit, 16-Mbit, and 64-Mbit memory, both symmetrical and asymmetrical addressing. The MTXC has six RAS# lines enabling the support of up to six rows of DRAM. Eight CAS# lines allow byte control over the array during write operations. The MTXC targets 60 ns (also supports 50 ns and 70 ns) DRAMs, and supports both single- and doublesided Dram modules. The MTXC provides CBR refresh and extended CBR refresh in the normal mode and self refresh or CBR (for EDOs only) during suspend mode. The MTXC also supports SDRAMs. The fourteen multiplexed address lines, MA[13:0], allow the MTXC to support 16-Mbit and 64-Mbit SDRAM devices. The MTXC has six CS# lines (i.e. muxed onto RAS#[5:0]). Although six CS# signals are provided, due to loading concerns, 5 rows of SDRAM maximum is recommended. Eight DQM lines (i.e., muxed with CAS#[7:0]) allow byte control over the array during the write operation. Two copies of SRAS# and SCAS# signals are provided for encoded SDRAM commands. The MTXC targets 60- and 66-MHz SDRAMs and supports both single- and double-sided SDRAM modules. The DRAM interface of the MTXC is configured by the DRAM Control Mode Register (DRAMC), DRAM Extended Control Register (DRAMEC), DRAM Timing Register (DRAMT), SDRAM Control Register (SDRAMC), six DRAM Row Boundary (DRB) Registers, and the DRAM Row Type (DRT) Registers. The DRB registers define the size of each row in the memory array.
PRELIMINARY
52
Extended Temperature 82439TX (MTXC) Datasheet
Seven Programmable Attribute Map (PAM) Registers are used to specify the cacheability, PCI enable, and read/write status of the memory space between 640 Kbytes and 1 Mbytes. Each PAM Register defines a specific address area enabling the system to selectively mark specific memory ranges as cacheable, read only, write only, read/write, or disabled. When a memory range is disabled, all CPU accesses to that range are forwarded to PCI. The MTXC also supports one of two memory holes, either from 512 KB-640 KB or from 14/15 MB-16 MB in main memory. Accesses to the memory holes are forwarded to PCI. The memory hole can be enabled/disabled through the DRAM Control register. All other memory from 1M to 256 MB is read/write L1 cacheable, and is L2 cacheable up to 64 MB. An optional Extended SMRAM DRAM memory space is also supported in the 256-MB to 512-MB address range. It consists of the 640-KB-1-MB DRAM area aliased at the 256-MB memory segment, and also an optional 128K/256K/512K/1M DRAM area chopped from the Top-of-DRAM memory and aliased above 256 MB in a similar manner. 4.3.1. DRAM ORGANIZATION
The MTXC integrates a DRAM controller that supports EDO, FPM, and SDRAM. SDRAM, EDO and FPM DRAM's can be mixed between rows, however, a given row must contain only one type of DRAM. When DRAM types are mixed (EDO, FPM and SDRAM) each row will run optimized for that particular type of DRAM. The MTXC supports six rows of memory (six RAS#/CS# lines). For maximum memory flexibility and performance, it is recommended that a DRAM configuration of four rows be used. This allows 64-Mbit DRAM devices to be used as well as the mixing of SDRAM and EDO/FPM. Figure 6 shows an EDO/FPM configuration using x32 SIMM modules and Figure 7 shows a four row EDO/FPM/SDRAM configuration using x64 DIMM modules (or x64 SO-DIMM). NOTE It is not recommended to mix SDRAM (which are 3V devices) with 5V EDO/FPM SIMMs, unless the SDRAM and EDO/FPM are properly isolated (e.g., isolate the memory data lines with Qswitches). Mixing 5V and 3V memory is not recommend for reliability reasons. Not all SDRAMs are 5V tolerant.
PRELIMINARY
53
Extended Temperature 82439TX (MTXC) Datasheet
MTXC
HD[63:0] M D[63:0] RAS[3:0]# CAS[3:0]# CAS[7:4]# M A[11:0]
RAS[3:2]# 32-Bit SIM M 32-Bit SIM M
W EB#
RAS[1:0]# 32-Bit SIM M 32-Bit SIM M
W EA#
Host Data Bus
mtxc_03
Figure 6. FPM/EDO Four Row SIMM Configuration
MTXC
HD[63:0] M D[63:0] RAS/CS[3:0]# CAS/DQM [3:0]# CAS/DQM [7:4]# M A[13:0] CKEB [3:2]# 64-Bit DIM or S0-DIM M M W EB# SRASB# SCASB# CKEA [1:0]# 64-Bit DIM or S0-DIM M M W EA# SRASA# SCASA#
Host Data Bus
mtxc_04
NOTES: 1. In a configuration that supports suspend to RAM, only CKE is used. This is because CKEB is not part of the suspend well the MTXC maintains during a suspend to RAM state. 2. In a desktop system that supports EDO/FPM, CKE and CKEB should be used as the second pair of MA0 and MA1 lines. CKE is used as MAA0 and CKEB is used as MAA1. In this case, they should be evenly distributed throughout the system along with the first pair of MA0 and MA1 lines.
Figure 7. FPM/EDO/SDRAM Four Row DIMM or SO-DIMM Configuration
PRELIMINARY
54
Extended Temperature 82439TX (MTXC) Datasheet
Rules for Populating SIMM Modules (or x32 SO-DIMM modules) * SIMM sockets can be populated in any order (i.e., memory for RAS0# does not have to be populated before memory for RAS[2:1]# or RAS[4:3]# are used). * SIMM socket pairs (i.e., two, 32-bit wide SIMMs) need to be populated with the same densities. For example, SIMM sockets for RAS0# should be populated with identical densities. However, SIMM sockets for RAS[2:1]# can be populated with different densities than the SIMM socket pair for RAS0#. * EDOs and standard page mode can both be used; however, only one type should be used per SIMM socket pair. For example, in the table shown below SIMM sockets for RAS[2:1]# can be populated with EDOs while SIMM sockets for RAS[4:3]# can be populated with standard page mode. If different memory is used for different rows, each row will be optimized for that type of memory. * The DRAM Timing Register which provides the DRAM speed grade control for the entire memory array must be programmed to use the timings of the slowest DRAMs installed. Rules for Populating DIMM or SO-DIMM modules * DIMM or SO-DIMM sockets can be populated in any order (i.e., memory for RAS0# does not have to be populated before memory for RAS[2:1]# or RAS[4:3]# are used). 4.3.2. CONFIGURATION REQUIREMENTS
General Configuration Requirements * In a system that uses 64-Mbit SDRAM, the RAS4#/CS4#/BA1 and RAS5#/CS5#/MA13 signals are used to provide two additional address lines (BA1 and MA13), and KRQAK/CS4_64# is used to provide the 5th CS# line, if required. To enable 64-Mbit support for four rows of SDRAM, set SDRAMC[bit 1] to 1 (offset 54h). To enable 64-Mbit support for five rows of SDRAM, SDRAMC[bit 1] must be set to 1, and DRAM cache must not be present in the system (indicated by CEC[bit 5]=0, offset 53h). In a five row SDRAM system that supports 64-Mbit SDRAM devices, the KRQAK/CS4_64# signal provides the fifth CS# (or CS4_64#) function. This means that a system that supports DRAM Cache, can not support five rows of 64-Mbit SDRAM. However, four rows of 64-Mbit SDRAM with DRAM Cache is supported. In a FPM/EDO only configuration, there are no restrictions on using 64-Mbit devices (i.e., all six rows can support 64Mbit DRAM devices. However, SDRAMC[bit 1] must be set to 1 if more than four rows of EDO/FPM are used. This allows the RAS4# and RAS5# functions to be used. Driven on RAS5#/CS5#/ MA13 Bit 1, reg 54h=0 Bit 1, reg 54h=1 and DRAM Cache is present* Bit 1, reg 54h=1 and DRAM Cache is not present1 RAS5#/CS5# MA13 Driven on RAS4#/CS4#/ MA13 RAS4#/CS4# BA1 (Bank Select) Driven on KRQAK/ CS4_64# KRQAK KRQAK 64-Mbit (SDRAM) no Yes (4 rows) Yes (5 rows) 64-Mbit (EDO/FPM) yes (6 rows) Yes (4 rows)
MA13
BA1 (Bank Select)
RAS4#/ CS4_64#
Yes (5 rows)
NOTES: 1. The presence of DRAM cache is indicated by the value in bit 5, register 53h. * * * Due to loading, using SDRAM x4 devices is not recommended. Buffering of SDRAM Rows is not supported In a five row system, the 5th row is intended to be implemented with DRAM devices that are soldered down on the motherboard. If a DIMM or a SIMM is used in the 5th row, it should not be used as an
PRELIMINARY
55
Extended Temperature 82439TX (MTXC) Datasheet
upgrade path by the end user; the size and type of DRAM that can be implemented in the 5th row is limited (see the bullets below). The total memory supported is 256 MB, even though it is possible to populate the six rows with more than 256 MB. This limit must be ensured by the system BIOS.
*
EDO/FPM only configuration Requirements * If more than four rows of x4 DRAM devices + one row of x8 DRAM devices of memory is supported, it is recommended that all six rows be buffered. MA and MWE# enable signals should be buffered. In a system that only supports x8 or x16 devices (i.e., x4 devices not supported), six rows of memory can be supported without buffering. Maximum load supported without buffers: Four rows of x4 DRAM devices + one row of x8 DRAM devices. A second pair or MA0 and MA1 signals are provided by muxing CKE with MAA0 and CKEB with MAA1. In a desktop system, it is required that the second pair of MA lines be used to support 5-2-2-2 EDO performance in more than two rows of memory. The second pair of MA lines are not required in a mobile system, assuming x4 devices are not used. The MA functionality is selected via DRAMC[bit 2] (67h).
* *
SDRAM only configuration Requirements * Maximum rows supported; Five rows of x8 devices.
SDRAM/EDO/FPM mixing configuration Requirements * If SDRAM and EDO/FPM are mixed in a system, the configuration is limited to a maximum of four rows (two rows of x4 EDO/FPM and two rows of x8 or x16 SDRAM). If only x8 or x16 EDO/FPM and SDRAM devices are used (i.e., not x4's), five rows can be supported. * SDRAMs can be mixed with EDO/FPM on a row by row basis (e.g., row 0 can be populated with SDRAMs while row 3 is populated with EDO/FPM). * A second pair or MA0 and MA1 signals are provided by muxing CKE with MAA0 and CKEB with MAA1. In a desktop system, it is required that the second pair of MA lines be used to support 5-2-2-2 EDO performance in more than two rows of memory. The second pair of MA lines are not required in a mobile system, assuming x4 devices are not used. The MA functionality is selected via DRAMC[bit 2] (67h). Table 12 provides a summary of the characteristics of memory configurations supported by the MTXC. Minimum values listed are obtained with single-sided SIMMs or DIMMs. Maximum values are obtained with double-sided SIMMs or DIMMs. Note that, for a 64-bit wide memory array, a minimum of two 32-bit wide DRAM SIMMs are required in any specific row. The minimum values used are also the smallest upgradeable memory size. Please note that EDO/FPM can also come on x64 DIMM modules.
PRELIMINARY
56
Extended Temperature 82439TX (MTXC) Datasheet
Table 12. Minimum (Upgradeable) and Maximum Memory Size for each configuration (DRAM) DRAM Tech. 4M DRAM Density 512K 1M 16M 1M 1M 2M 4M 4M 64M 2M 4M 4M 8M 16M DRAM Width 8 4 16 16 8 4 4 32 16 16 8 4 DRAM SIMM SS x32 512K 1M 1M 1M 2M 4M 4M 2M 4M 4M 8M 16M DS x32 1M 2M 2M 2M 4M 8M 8M 4M 8M 8M 16M 32M DRAM Addressing Asymmetric Symmetric Symmetric Asymmetric Asymmetric Symmetric Asymmetric Asymmetric Symmetric Asymmetric Asymmetric Symmetric Address Size Row 10 10 10 12 11 11 12 12 11 12 12 12 9 10 10 8 10 11 10 9 11 10 11 12 Col DRAM Size Min. (UP) (1 row) 4 MB 8 MB 8 MB 8 MB 16 MB 32 MB 32 MB 16 MB 32 MB 32 MB 64 MB 128 MB Max. (6 rows) 24 MB 48 MB 48 MB 48 MB 96 MB 192 MB 192 MB 96 MB 192 MB 192 MB 256 MB 256 MB
Table 13. Minimum (Upgradeable) and Maximum Memory Size for each configuration (SDRAM) SDRA M Tech. 16M SDRAM Density 1M 2M 4M 64M 2M 2M 4M 8M 16M SDRAM SDRAM DIMM Width 16 8 4
1
SDRAM Addressing Asymmetric Asymmetric Asymmetric Asymmetric Asymmetric Asymmetric Asymmetric Asymmetric
Address Size Row 12 12 12 12 13 14 14 14 Column 8 9 10 9 8 8 9 10
SDRAM Size Min. (UP) (1 row) 8 MB 16 MB 32 MB 16 MB 16 MB 32 MB 64 MB 128 MB Max. (6 rows) 48 MB 96 MB 192 MB 96 MB 96 MB 192 MB 256 MB 256 MB
SS x64 1M 2M 4M 2M 2M 4M 8M 16M
DS x64 2M 4M 8M 4M 4M 8M 16M 32M
32 32 16 8 4
1
NOTES: 1. Functionally the 430TX supports x4 SDRAM devices. However, due to loading reasons, it is not recommended that x4 devices be used in 60-MHz and 66-MHz designs.
PRELIMINARY
57
Extended Temperature 82439TX (MTXC) Datasheet
The memory organization shown below represents the maximum 256 MB of address space. Accesses to memory space above Top-of-DRAM (< 256 MB), video buffer, or the memory gaps (if enabled) are forwarded to PCI, and these regions are not cacheable. Below 1 MB, there are several memory segments which have selectable cacheability. None of the DRAM space occupied by the video buffer (except for SMM usage) or the memory space gaps is remapped (and is therefore "lost").
4 GB Forwarded to PCI (Non-Cacheable) 256 M + T B op-of-DRAM Optional TSEG (Cacheable) 128 KB/256 KB/512 KB/1 M B Forwarded to PCI (Non-Cacheable) aliased Optional HiSM RAM Shadow Area 384 KB (Cacheable) Forwarded to PCI (Non-Cacheable) aliased Non-Cacheable in L2, Cacheable in L1 Cacheable T op-of-DRAM (DRB5) T op-ofM M ory ain em 16 M B TSEG Optional M ory Space Gap em 14/15 M B Cacheable 1M B System DRAM Expansion and BIOS Region (Cacheable Segm ents) 768 KB Video Buffer (SM MSpace Non-Cacheable) Optional M ory Space Gap em DOS Applications (No Read/W Protect) rite (Always Cacheable) 0
mtxc_08
256 M + 640 KB B 256 M B 64 M B
640 KB 512 KB
Figure 8. Memory Space Organization
PRELIMINARY
58
Extended Temperature 82439TX (MTXC) Datasheet
4.3.3. DRAM ADDRESS TRANSLATION
The multiplexed row/column address to the DRAM memory array is provided by the MA[11:0] signals (MA[13:0] for SDRAM 64-Mbit support). The MA bits are derived from the host or PCI address bus as defined by the Table 14. The MTXC supports a 2K byte page size only. The MA lines are translated from the address lines A[26:3] for all memory accesses. Table 14. MTXC DRAM Address Map Summary
ADDR MA13 MA12/ MA11/ MA10 BA1 BA0 MA9 MA8 MA7 MA6 MA5 MA4 MA3 MA2 MA1 MA0
Row Col
A24
A23 A23
A11 A26/ A11
A22 A11/ A25 "V"
A21 A11/ A24/ A26
A20 A11/ A22/ A23/ A25
A19 A10
A18 A9
A17 A8
A16 A7
A15 A6
A14 A5
A13 A4
A12 A3
NOTES: 1. V=Valid level (either 0 or 1) used for SDRAMs. It is 1 during the initialization sequence. It is 0 during normal mode of operation. 2. BA0 and BA1 are the muxed bank selects for SDRAM. Bank select BA1 is required for 64-Mbit SDRAM support. 4.3.4. DRAM PAGING
If DRAMC[bit 4]=1, the MTXC keeps the page open until a page or row miss occurs. If DRAMC[bit 4]=0 (default), the DRAM page is kept open when: * * CPU host bus is non-idle, or PCI interface owns the bus. DRAM TYPES FPM Mode
4.3.5. 4.3.5.1.
The MTXC, as a default, supports the standard fast page mode (FPM) DRAM. 4.3.5.2. EDO Mode
Extended Data Out (or Hyper Page Mode) DRAM is designed to improve the DRAM read performance. EDO DRAM holds the memory data valid until the next CAS# falling edge. Compared to standard page mode DRAM which tri-states the memory data when CAS# negates to precharge. With EDO, the CAS# precharge overlaps the memory data valid time. This allows CAS# to negate earlier while still satisfying the memory data valid window time. 4.3.5.3. SDRAM Mode
Synchronous DRAM (SDRAM) implements a fully synchronous interface as compared to a conventional DRAM whose timing delays are related to the rising and falling edges of the RAS#, CAS#, and WE# input signals. The 430TX supports all of the features and timings as shown in the "SDRAM PC" specification. The objective of the SDRAM PC Specification is to enable low cost and easily manufacturable SDRAMs for the main stream volume desktop and Mobile PC's. There are three grade parts defined for the 430TX. All of the
PRELIMINARY
59
Extended Temperature 82439TX (MTXC) Datasheet
speed grade conform to the SDRAM PC Specification. For information on the performance of each of the Speed grade parts, refer to the DRAM performance section. The Three speed grade parts are shown in Table 15. Table 15. SDRAM Speed Grade Parts Speed Grade 66.67 MHz 66.67 MHz 66.67 MHz CAS latency (CL) 3 3 2 RAS to CAS (Trcd) 3 2 2 System Frequency 60/66 MHz 60/66 MHz 60/66 MHz
SDRAM Command Reference The 430TX supports the following commands: Command Mode Register Set (MRS) Activate Bank (ACT) Read Bank (RD) Write Bank (WR) Precharge All Banks (PALL) Deselect Device Command No Operation (NOP) Auto Refresh CBR (REFR) Data Write/Output Enable Data Mask/Output Disable Self Refresh Entry Self Refresh Exit
Table 16 MRS command (Mode Register Set) Supported by the MTXC.
Table 16. Command Fields A11 0 A10 0 A9 0 A8 0 A7 0 A[6:4] CL A3 WT A[2:0] BL
CAS Latency Field (CL) Bits[6:4] 010 011 All Other CAS Latency 2 3 X
Wrap Type Field (WT) Bit 3 0 1 Type X Interleave
Burst Length Field (BL) Bits[2:0] 010 All Other Burst Length 4 X
The linear order addressing is not supported.
NOTES: 1. X=Don't Care. These Modes are Don't Care for MTXC specific implementation.
PRELIMINARY
60
Extended Temperature 82439TX (MTXC) Datasheet
4.3.6. AUTO DETECTION
The SDRAM, FPM, and EDO detection is performed by BIOS. Note that when accessing any of the DRAM related registers (i.e., 54h-68h), refresh should be turned off via the DRAM Control register (DRAMC). 4.3.7. DRAM PERFORMANCE
The DRAM performance is controlled by the DRAM timing register, processor pipelining, and by the type of DRAM used (EDO or FPM or SDRAM). Table 17 depicts both EDO and standard page mode optimum timings. For read cycles, clocks counts are measured from ADS# to BRDY#. For write cycles, the measurement is broken up into two parts. The first part consists of the rate of posting data in to the CPU to DRAM posted write buffers. This is measured from ADS# to BRDY#. The second part consists of the retire rate from posted write buffers to the DRAM. The leadoff for retiring is measured from the clock after BRDY# assertion to the CAS# assertion. Table 17 lists the performance summary for 60 ns EDO/FPM DRAMs. The four row column is assuming each row is populated with a maximum of 16, x4 devices=64 DRAM devices. The five row column is assuming each of the first four rows is populated with a maximum of 16, x4 devices and the fifth row is populated with a maximum of eight, x8 devices=72 DRAM devices. The six row column assumes that each of the six rows can be populated with a maximum of 16, x4 devices. The FELO and SLD bits are used to control the leadoff for read cycles (page hit, row miss, and page miss). Each bit removes one clock from the leadoff, when enabled. Note that FELO impacts EDO only and must be disabled for FPM. The DLT bits are used to control the base starting point for the leadoff for read/write cycles (page miss and row miss, only).
PRELIMINARY
61
Extended Temperature 82439TX (MTXC) Datasheet
Table 17. EDO/ Standard Page Mode Performance Summary (60 ns DRAMs) Processor Cycle Type (pipelined) 60/66 MHz w/ four rows 5-2-2-2 8-2-2-2 11-2-2-2 60/66 MHz w/ five rows 6-3-3-3 9-3-3-3 12-3-3-3 6-3-3-3-4-3-3-3 7-4-4-4 9-4-4-4 12-4-4-4 7-4-4-4-4-4-4-4 3 6 9 3-1-1-1 -3-3-3 2 0 1 0 1 1 1 0 60/66 MHz w/ six rows Buffered 6-3-3-3 10-3-3-3 13-3-3-3 6-3-3-3-4-3-3-3 7-4-4-4 9-4-4-4 12-4-4-4 7-4-4-4-4-4-4-4 3 7 10 3-1-1-1 -3-3-3 3 0 1 0 1 1 0 0 DRAM Type
Burst Read Page Hit Read Row Miss
1
EDO EDO EDO EDO FPM FPM FPM FPM EDO/FPM EDO/FPM EDO/FPM EDO/FPM EDO/FPM EDO/FPM EDO/FPM EDO FPM EDO/FPM EDO/FPM EDO/FPM EDO/FPM
Read Page Miss
Back-to-Back Burst Reads Page Hit 5-2-2-2-3-2-2-2 Burst Read Page Hit Burst Read Row Miss
1
6-3-3-3 9-3-3-3 12-3-3-3 6-3-3-3-3-3-3-3 3 6 9 3-1-1-1 -2-2-2 2
5 6 6
Burst Read Page Miss Back-to-Back Burst Read Page Hit Write Page Hit
2,3,4 2,3,4 2,3,4
Write Row Miss
3,4
Write Page Miss Posted Write
Write retire rate from Posted Write Buffer Single writes Reg 56h, Bit 4 (SLD)
0 1 0 2 2 1 0
Reg 56h, Bit 5 (FELO) Reg 56h, Bit 5 (FELO)
Reg 58h, Bits[6:5] (DRBT) Reg 58h, Bits[4:3] (DWBT) Reg 58h, Bits[1:0] (DLT) Reg 56h, Bit 6 (RRA)
NOTES: 1. The row miss cycles assume that the new page is closed from the prior cycle. Due to the MA[13:0] to RAS# setup requirements, if the page is open, 2 clocks are added to the leadoff. 2. This cycle timing assumes the write buffer(DWB) is empty. 3. Write timing is measured from the clock after BRDY# is returned to the CPU up to CAS# assertion for that cycle. 4. Write data is always posted as 3-1-1-1 (ADS# to BRDY#), if write buffers is available. 5. This bit (SLD) should be set to a 1 (speculative leadoff disable) in systems with cache and to 0 in systems without cache. 6. When set to 1, enables fast timing for EDO timing only. Enables one HCLK pull in for page hit, page miss, and row miss cycles.
PRELIMINARY
62
Extended Temperature 82439TX (MTXC) Datasheet
Table 18 lists the performance summary for SDRAM. The CL= 3 column represents a CAS latency of 3 part with a RAS to CAS (Trcd) of two clocks. The CL=2 column represents a CAS latency of two part. The performance numbers in Table 18 assume each row is populated with a maximum of eight, x8 devices=40 SDRAM devices. The SLD bit (page hit, row miss, and page miss) is used to control the leadoff for read cycles. This bit removes one clock from the leadoff, when enabled. Table 18. SDRAM Performance Summary Processor Cycle Type 60/66 MHz CL=3 Five Rows (Max) 7-1-1-1 9-1-1-1 12-1-1-1 7-1-1-1 2-1-1-1 3 6 9 3-1-1-1 -1-1-1 1 0 0
4
60/66 MHz CL=2 Five Rows (Max) 6-1-1-1 8-1-1-1 11-1-1-1 6-1-1-1 2-1-1-1 3 5 8 3-1-1-1 -1-1-1 0 1 1 0
Burst Read Page Hit Read row Miss
1
Read Page Miss Back-to-Back Burst Reads Page Hit Write Page Hit
2,3 2,3 2,3
Write Row Miss
Write Page Miss Posted Write
2,3
Write retire rate from Posted Write Buffer Reg 54h, Bit 5 (RCO) Reg 54h, Bit 4 (CL) Reg 54h, Bit 3 (RT) Reg 56h, Bit 4 (SLD)
5
0
NOTES: 1. The row miss cycle assumes that the new page is closed from the prior cycle. 2. This cycle timing assumes the write buffer(DWB) is empty. 3. Write data is always posted as 3-1-1-1 (ADS# to BRDY#), if write buffers is available. 4. This bit (SLD) must be set to a 1 (speculative leadoff disable) in systems with cache and to 0 in systems without cache. 5. For a CL=3 part that can not meet a RAS to CAS timing (Trcd) of two HCLKs, RCO can be set to 0. This will add an HCLK to the leadoff cycle for Row miss and Page miss cycles.
PRELIMINARY
63
Extended Temperature 82439TX (MTXC) Datasheet
4.3.8. DRAM REFRESH
MTXC supports CAS-before-RAS# (CBR) refresh and Self refresh. The refresh rate is controlled via the DRAM Refresh Rate field in the DRAM Control Register (DRAMC). When a refresh request is generated, it is placed in a four entry queue. The DRAM controller services a refresh request when the refresh queue in not empty and the controller has no other requests pending. When the refresh queue is full, refresh becomes the highest priority request and will be serviced next by the controller. Refresh is only performed on rows that are populated (i.e., "smart refresh"). The controller determines which rows are populated by looking at the DRB registers. Note that Refresh has to be disabled before the refresh rate is changed. Refer to bit 5 in the MCTL register (offset 79h) for suspend refresh information.
4.4.
4.4.1.
PCI CLK Control (CLKRUN#)
CLOCKING STATES
There are three main states in the clocking protocol: * * * Clock Running: The clock is running and the bus is operational. About to Stop: The central resource has indicated on the CLKRUN# line that the clock is about to stop. Clock Stopped: The clock is stopped with CLKRUN# being monitored for a restart OPERATION
4.4.2.
The MTXC is a CLKRUN# Master device and behaves according to the rules for a master device. The PIIX4 companion chip controls the clocks in the system and is the CLKRUN# Central Resource. Please refer to the latest "PCI Mobile Design Guide" for more information.
4.5.
SMRAM Memory Space
The MTXC supports the use of main memory as System Management RAM (SMRAM), enabling the use of System Management Mode. The MTXC supports two SMRAM options; Compatible SMRAM (C_SMRAM) and Extended SMRAM (E_SMRAM). 4.5.1. COMPATIBLE SMRAM (C_SMRAM)
This is the traditional SMRAM feature supported in Intel PCIsets. When this function is enabled via C_BASE_SEG[2:0]=010 and G_SMRAME=1 of the SMRAMC register, the MTXC reserves 000A0000h through 000BFFFFh (A and B segments) of the main memory for use as Noncacheable SMRAM. CPU accesses to segments A and B while not in SMM (i.e., SMIACT# is negated) are always forwarded to the PCI bus. CPU accesses to segments A and B while in SMM (i.e., SMIACT# is asserted) are forwarded to either DRAM or PCI bus, depending on the value of bits[6:0] of the SMRAMC register. PCI masters cannot access the SMRAM area of the main memory. When a PCI master tries to access the SMRAM space, the MTXC does not respond to the PCI cycle (i.e., DEVSEL# is not asserted). 4.5.2. EXTENDED SMRAM (E_SMRAM)
This feature in the MTXC extends the SMRAM space up to 1 Mbytes and provide writeback cacheability. This feature requires that SMI handlers execute above 1 Mbytes which will require rewriting the existing code to
PRELIMINARY
64
Extended Temperature 82439TX (MTXC) Datasheet
operate properly above 1 Mbytes. However once this is done, then SMI handlers execute at full processor performance. An error status bit is set in the Extended SMRAM Control register if the CPU tries to access the extended SMRAM space while SMIACT# is negated and D_OPEN bit is 0. This access is forwarded to PCI bus and may result in a Master Abort condition. Extended SMRAM feature allows up to 1 Mbyte of SMRAM space to be writeback cacheable. This memory space consists of any DRAM not used by the system (as shadow space etc.) between 640 Kbytes and 1 Mbyte (this memory space is referred to as High Memory in this document), and an optional block of memory referred to as the "TSEG". The TSEG is either a 128 Kbyte, 256 Kbyte, 512 Kbytes, or 1 Mbytes block of memory, as defined by TSEG_SZ[1:0] of the SMRAMC register. When TSEG is enabled, the TSEG block of memory is disabled from the top of memory and the system BIOS should report a main memory size of (memorize - TSEG) to the OS. The two areas of memory available for SMRAM when Extended SMRAM is enabled are: Physical Address 100A0000h to 100FFFFFh 10000000h plus TOM minus TSEG_SZ to 10000000h plus TOM DRAM Address 000A0000h to 000FFFFFh TOM minus TSEG_SZ to TOM (High Mem) (TSEG)
Extended SMRAM option has the following DRAM memory available to it:
Table 19. Extended SMRAM DRAM memory regions DRAM Area A Segment B Segment C Segment D Segment E Segment F Segment TSEG Size/Availability 64 Kbytes always available if enabled (i.e., H_SMRAM=1 and G_SMRAME=1) 64 Kbytes always available if enabled (i.e., H_SMRAM=1 and G_SMRAME=1) 64 Kbytes available if not used for shadowing (as defined by PAM register) and enabled (i.e., H_SMRAM=1 and G_SMRAME=1) 64 Kbytes available if not used for shadowing (as defined by PAM register) and enabled (i.e., H_SMRAM=1 and G_SMRAME=1) 64 Kbytes available if not used for shadowing (as defined by PAM register) and enabled (i.e., H_SMRAM=1 and G_SMRAME=1) 64 Kbytes only available for suspend/resume (as defined by PAM register) if enabled (i.e., H_SMRAM=1 and G_SMRAME=1) 128K, 256K, 512K or 1M bytes available if enabled (i.e., TSEG_EN=1 and G_SMRAME=1)
PRELIMINARY
65
Extended Temperature 82439TX (MTXC) Datasheet
As with the Compatible SMRAM solution, MTXC does not claim any bus master access to the Extended SMRAM memory ranges defined above. The CPU can access these memory ranges by one of the following mechanisms: * * * The processor generating an access to one of the defined memory ranges while in the SMM (SMIACT# is active). A processor access to any of the defined ranges while not in SMM (SMIACT# is inactive) and with the D_OPN bit reset will be forwarded to PCI bus and a status bit is set in the SMRAMC register. The processor generating an access to one of the defined memory ranges while the D_OPN bit is set. Any modified write access of the processor is allowed to write into the SMRAM space, regardless of the state of the D_OPN, D_CLS, or SMIACT# signals.
The cacheability of SMRAM space is dependent on how much physical DRAM is available in the system. If the system has less than 32 Mbytes of DRAM, the SMRAM is cached in both the L1 and L2. If the system has more than 32 Mbytes of DRAM, the SMRAM is cached in only the L1. 4.5.3. SMRAM PROGRAMMING CONSIDERATIONS
When using the Extended SMRAM configuration, the SMI handler software must be extremely careful when accessing DRAM memory in the 100A0000h to 100FFFFFh memory range. First, if this area of memory is accessed while the CPU is not in SMM mode and the D_OPN bit is not set, the MTXC will forward the cycle to PCI bus which may cause a fatal system error and system shutdown. Second, only areas within the 100A0000h to 100FFFFFh region that have been selected as SMRAM space should be accessed; otherwise, the L1 and L2 caches will become incoherent, which will cause a future system error. Any memory in normal DRAM space that is not used in OS or application space can be used as SMRAM memory.
4.6.
Low Power States
MTXC supports five types of low power states: Chip Standby, Power On Suspend (POS), Suspend to RAM (STR), Suspend to Disk (STD), and dynamic stop clock. The Table 20 summarizes the various MTXC's Low power states. Table 20. 430TX Low Power State Summary PM Mode Chip Standby Dynamic Stop Clock Description When MTXC's CPU and PCI busses are both idle, MTXC enters this state. Exit Latency Target No delay
<10 ms MTXC provides provisions that enable transitioning the CPU in and out of the stop clock state in an active system. This includes the ability to disable the system arbiter and transition the memory controller in and out of the suspend refresh state. System PLLs are powered down, only running clock is the RTC clock and the SUSCLK. MTXC maintains DRAM refresh using SUSCLK. CPU complex (CPU and L2) and PCI interface are powered off. Only the RTC clock and SUSCLK are running. MTXC maintains DRAM refresh using SUSCLK. CPU complex (CPU and L2), DRAM and PCI interface are powered off. <10 ms
Powered On Suspend (POS) Suspend to RAM (STR) Suspend to Disk(STD)
~1 sec
~30 sec
PRELIMINARY
66
Extended Temperature 82439TX (MTXC) Datasheet
The 430TX system maintains a very low power CPU complex by utilizing the different power down features available from the CPU, cache data RAMs and utilizing leading edge low power design techniques in the 430TX system components. The 430TX components work in unison to dynamically control the CPU complexes power state without adversely affecting performance. The following gives a brief description of how the 430TX system components achieve these low power states. The MTXC and PIIX4 work in unison to maintain a very low power L2 subsystem without adversely affecting peak performance. NOTE There are some system restrictions when DRAM Cache is implemented in a system that supports STP_CLK, POS, and STR power management modes. Since KRQAK is not implemented in the "Suspend Well," the correct operation of KRQAK is not guaranteed when the system enters the above mentioned power management modes. To avoid data corruption in the L2 cache, a system that implements the STP_CLK, POS, and STR modes must abide by the following rules: 1. Before entering these power management modes, the DRAM cache must be flushed so that all modified lines end up in system memory. 2. After exiting these power management modes, the DRAM Cache must be reinitialized. 4.6.1. CHIP STANDBY
The MTXC also supports a chip standby mode. When the MTXC determines that both its CPU interface and PCI interface are idle, it will dynamically place itself into a very low power state. While in chip standby state the MTXC is able to respond to new CPU or PCI bus master accesses with no performance penalty. This provides very optimized power/performance characteristics because the CPU interface are idle for large periods of time. The MTXC enters Chip Standby mode when the following conditions are true: * * * * * Host Bus idle PCI bus Idle Normal Mode (i.e., not Test Mode) Not in RESET state Internal operations idle
Entering the Chip Standby state is not dependent on any timer expiration. When the above conditions are met, the MTXC can enter the chip standby state as soon as it can. 4.6.2. SUSPEND/RESUME
The MTXC supports POS, STR, STD and SOFF (Soft Off) suspend states. The MTXC supports the POS mode by maintaining all of its power planes when in the suspend state. The MTXC supports the STR modes by isolating its CPU and PCI interfaces, and only maintaining the DRAM refresh off the SUSCLK signal. When exiting the STR modes, the MTXC's core well is reset and its context is lost (the power management context is not lost however). The MTXC supports the STD and SOFF modes by being totally powered off.
PRELIMINARY
67
Extended Temperature 82439TX (MTXC) Datasheet
CPU/L2 Interface
Core Well
RSM Well DRAM Refresh Suspend
RASx#, CASx#, WEx# SUSCLK, SUSSTAT1# CKE
MTXC
PCI Interface
mtxc_05
Figure 9. MTXC Power Planes
4.6.2.1.
Power Transition Changes
The MTXC supports several suspend modes that support the PIIX4 system suspend states. Table 21 illustrates what suspend mode the MTXC enters upon the appropriate PIIX4 suspend mode. Table 21. Power Transition States PIIX4 Suspend State POS STR STD Off/Soft Off MTXC Suspend State PonS PoffS Off Off MTXC Description All interfaces enabled, clocks stopped. CPU, L2, PCI interfaces disabled Chip is off. Chip is off.
The core logic should be reset when the PCI bus is reset. This means that the refresh logic and power sequencing logic is not reset during resumes (part of the resume well). 4.6.2.2 MTXC Transition
The MTXC will transition from suspend refresh to normal refresh when it samples SUSCLK active (high) with the first sample of SUS_STAT# inactive (high). Samples are taken on the rising edge of the PCI clock. The MTX will reset its resume well (DRAM configuration registers go to default settings) when it samples SUSCLK inactive (low) with the first sample of SUS_STAT# inactive (high). The mechanism which allows this functionality is based on the PIIX4 deasserting the SUS_STAT# signal in relation to the SUSCLK,
PRELIMINARY
68
Extended Temperature 82439TX (MTXC) Datasheet
synchronized to the PCI clock. To not reset the suspend well within the MTXC, the PIIX4 will drive the SUS_STAT# signal just after the SUSCLK signal has gone high. To reset the resume well within the MTXC, the PIIX4 will drive the SUS_STAT# signal while the SUSCLK is disabled (low). Because of this functionality, SUSCLK should not be inverted for any reason in applications.
4.7.
PCI Interface
The MTXC integrates a high performance interface to the PCI local bus taking full advantage of the high bandwidth and low latency of PCI. The MTXC is fully PCI 2.1 compliant. Table 22 lists the PCI bus commands supported. Five PCI masters are supported by the integrated arbiter including the PIIX4 and four general PCI masters. The MTXC acts as a PCI master for CPU accesses to PCI. The PCI bus is clocked at one half the frequency of the CPU clock. This divided synchronous interface minimizes latency for CPU-toPCI cycles and PCI-to-main memory cycles. The MTXC integrates posted write buffers for CPU memory writes to PCI. Back-to-back sequential memory writes to PCI are converted to burst writes on PCI. This feature allows the CPU to continue posting DWord writes at the maximum bandwidth for the Pentium processor for the highest possible transfer rates to the graphics frame buffer. Read prefetch and write posting buffers in the MTXC enable PCI masters to access main memory at up to 120 MB/sec. The MTXC incorporates a snoop ahead feature that allows PCI masters to continue bursting on both reads and writes even as the bursts cross cache line boundaries. The MTXC forwards each of the CPU shutdown, Halt, and Stop Grant cycles to the PCI bus as special cycles. These cycles are terminated on PCI as master abort and a BRDY# is returned to the CPU. The Stop Grant cycle is propagated with 0002h in the message field and 0012h in the message dependent data field. Table 22. PCI Commands C/BE# 0000 0001 0010 0011 0100 0101 0110 0111 1000 1001 1010 1011 1100 1101 Command Interrupt Acknowledge Special cycle I/O read I/O write reserved reserved Memory read Memory write reserved reserved Configuration Read Configuration Write Memory Read Multiple Dual Address Cycle Target Support NO NO YES YES NO NO YES YES NO NO NO NO As Memory Read NO Initiator Support YES YES YES YES NO NO YES YES NO NO YES YES NO NO
PRELIMINARY
69
Extended Temperature 82439TX (MTXC) Datasheet
Table 22. PCI Commands C/BE# 1110 1111 Command Memory Read Line Memory Write and Invalidate Target Support As Memory Read As Memory Write Initiator Support NO NO
4.8.
System Arbitration
The MTXC's PCI Bus Arbiter allows PCI peer-to-peer traffic concurrent with CPU main memory/second level cache cycles. The arbiter supports five PCI masters. REQ[3:0]#/GNT[3:0]# are used by PCI masters other than the PCI-to-ISA expansion bridge (PIIX4). PHLD#/PHLDA# are the arbitration request/grant signals for the PIIX4 and provide guaranteed access time capability for ISA masters. PHLD#/PHLDA# also optimize system performance based on the PIIX4 known policies.
PHLD# REQ0# REQ1# REQ2# REQ3#
Arbiter
PHLD A# GNT0# GNT1# GNT2# GNT3#
mtxc_06
Figure 10. PCI Arbiter
4.8.1.
PRIORITY SCHEME AND BUS GRANT
The highest priority requester is determined by a fixed order queue together with a highest priority pointer. Although the priority ring is fixed, the highest priority pointer moves to determine which PCI agent is at the top (and bottom) of the queue. The arbiter counts three grant assertions to requesters different than the one it is currently granting (and all grants within MTT are collapsed to one) to decide when it's time to let the host in. The grant signals (GNTx#) are normally negated after recognition of FRAME# assertion, or 16 PCLKs from grant assertion, if no cycle has started.
PRELIMINARY
70
Extended Temperature 82439TX (MTXC) Datasheet
C P U /P C I Priority Q ueue
P C I P riority Q ueue
CPU
PHLD#
PCI
REQ0#
PCI
REQ1#
PCI
PHLD#
REQ2#
REQ3#
mtxc_07
NOTES: 1. In the PCI Priority Queue, the last agent granted is always dropped to the bottom of the queue for the next arbitration cycle, but the order of the chain is always preserved. 2. In the PCI Priority Queue, if PHLD is at the bottom of the queue, the upper PHLD slot is masked. This prevents back-toback PHLD grants, if other PCI request are pending. 3. In the CPU/PCI priority Queue, the CPU is granted high priority status after 3 consecutive PCI grants. If three consecutive PCI grants have not been counted down, then the CPU can be granted the bus as the low priority agent.
Figure 11. Arbitration Priority Rotation
PRELIMINARY
71
Extended Temperature 82439TX (MTXC) Datasheet
Multi-Transaction Timer (MTT) The priority chain algorithm has been enhanced by the Multi-Transaction Timer (MTT) mechanism. Once a PCI agent is granted, the MTT is started. This timer then counts down in PCI clocks from its preset value to zero. Until the timer expires, that agent will be promoted to being the highest priority PCI agent for the next grant event. 4.8.2. CPU POLICIES
The CPU never explicitly requests the bus. Instead, the arbiter grants the bus to the CPU when: * * The CPU is the highest priority PCI agents do not require main memory (peer-to-peer transfers or bus idle) and the PCI bus is not currently locked by a PCI master
When the CPU is granted as highest priority, the MLT timer is used to guarantee a minimum amount of system resources to the CPU before another requesting PCI agent is granted.
5.0. 5.1.
CLOCKS AND RESET Clock Generation and distribution
The MTXC and CPU should be clocked from one clock driver output to minimize skew between the CPU & MTXC.
5.2.
RESET Sequencing
The MTXC is asynchronously reset when the RST# signal is asserted. The MTXC arbiter includes support for PCI central resource functions. These functions include driving the AD[31:0],C/BE[3:0]#, and the PAR signals when no one is granted the PCI bus and the PCI bus is idle. The MTXC drives 0's on these signals during these times, plus during RESET.
6.0. ELECTRICAL TIMING SPECIFICATIONS 6.1. Absolute Maximum Ratings
Case Temperature under Bias............................................... -40oC to +115oC Storage Temperature ............................................................ -55oC to +150oC Voltage on 5V tolerant pins with Respect to Ground.............. -0.3 to REFVDD5 + 0.3 Voltage on 3.3V pins with Respect to Ground ....................... -0.3 to VDD3 + 0.3 Supply Voltage with Respect to Vss ...................................... -0.3 to +3.6 V (2.5V CPU) Supply Voltage with Respect to Vss ................... -0.2 to +2.7 V Maximum Power Dissipation ................................................. 1.0 W WARNING: Stressing the device beyond the "Absolute Maximum Ratings" may cause permanent damage. These are stress ratings only. Operating beyond the "Operating Conditions" is not recommended and extended exposure beyond "Operating Conditions" may affect reliability.
PRELIMINARY
72
Extended Temperature 82439TX (MTXC) Datasheet
6.2. Thermal Characteristics
The Extended Temperature 82439TX (MTXC) is designed for operation at case temperatures between -40oC and 115oC. The thermal resistance of the MTXC is provided in Table 1.
Table 23. MTXC Package Thermal Resistance
Parameter 0 (0) Thetaja (oC/Watt) Thetajc (oC/Watt) 34 8 Air Flow Meters/Second (Linear Feet per Minute) 1.0 (196.9) 26
PRELIMINARY
73
Extended Temperature 82439TX (MTXC) Datasheet
6.3. MTXC DC Characteristics
Table 2. MTXC DC Characteristics
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE = -40oC to +115oC) Symbol VIL1 Parameter Input Low Voltage Min -0.3 0.8 Max V Unit Notes Notes 1, 2, 3 VCC =3.135V VIL2 Input Low Voltage -0.3 0.7 V Notes 1 VCC(CPU) = 2.375V VIH1 Input High Voltage (3.3V signals) Input High Voltage (2.5V signals) Input High Voltage (5V signals) Output Low Voltage IOL = 1mA (all signals except as noted below) IOL = 3mA (Note 4) IOL = 6mA (Note 5) VOL2 Output Low Voltage IOL = 100uA IOL = 1mA IOL = 2mA VOH1 Output High Voltage IOH = -1mA (all signals except as noted below)IOH = -2mA (Notes 4,5) 2.4 0.2 0.3 0.4 V 3.3V signals (Note 7) V 2.5V Signals Note 1 2.2 VCC + 0.3 V Notes 1, 2, 3 VCC = 3.6V 1.7 VCC + 0.3 V Notes 1 VCC = 2.7V 2.2 VREF + 0.3 V Notes 2 VREF = 5.25V 0.4 V 3.3V signals (Note 7)
VIH2
VIH3
VOL1
PRELIMINARY
74
Extended Temperature 82439TX (MTXC) Datasheet
VOH2
Output High Voltage IOH = -100uA IOH = -1mA IOH = -2mA 2.1 1.9 1.7 1.5 1.25 1.5 1.25 10 300
V
2.5V Signals Note 1
VT1 VT2 IIL1 IIL2
Threshold Voltage (3.3V Signals) Threshold Voltage (2.5V Signals) Input Leakage Current Input Leakage Current
V V uA uA
Note 8 Note 8 0VCIN COUT CI/O
Input Capacitance Output Capacitance I/O Capacitance
12 12 12
pF pF pF
FC = 1 MHz FC = 1 MHz FC = 1 MHz
PRELIMINARY
75
Extended Temperature 82439TX (MTXC) Datasheet
NOTES:
1.
These signals are CPU Vcc (3.3V or 2.5V).: A[31:3], BE[7:0]#, BRDY#, NA#, AHOLD, EADS#, HD[63:0], KEN#/INV, HLOCK#, M/IO#,D/C#, W/R#, ADS#, HITM#, CACHE#, SMIACT#, HCLKIN, VCC(CPU) These signals are 3.3V with 5.0V tolerance: TIO[7:0], AD[31:0], C/BE[3:0]#, PLOCK#, FRAME#, IRDY#, TRDY#, DEVSEL#, STOP#, PAR, REQ[3:0]#, PCLKIN, PHLD#, MD[63:0], TESTIN#, PCLKIN, VREF, LOCK#, PHLDA#, CLKRUN#, RST#. These signals are 3.3V: KRQAK, TWE#, BWE#, GWE#, COE#, CCS#, CADS#, CADV#, CKEB, CKE, SCAS[A,B]#, MWE#, MWEB#, MA[11:0], CAS[7:0]# or DQM[7:0], RAS[5:0]# or CS[5:0]#, VCC(SUS), VCC, SUSTAT1#, SUSCLK, GNT[3:0]#. I ol and I oH apply to the following signals: AD[31:0], C/BE[3:0]#, PAR I ol and I oH apply to the following signals: FRAME#, IRDY#, TRDY#, DEVSEL#, STOP#, LOCK#. I IL applies to the following signals: HD[63:0], MD[63:0], KRQAK, TIO[7:0], A27. These signals have internal pulldown resistors. All signals from note 1 when the CPU Vcc is 3.3V ; All signals from note 2, 3. Threshold voltage for all delay and pulse width measurements.
2.
3.
4. 5. 6.
7. 8.
PRELIMINARY
76
Extended Temperature 82439TX (MTXC) Datasheet Table 3: MTXC Pin States in Various System Modes
Power Plane VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(CPU) VCC(SUS) VCC(SUS) VCC VCC(SUS) VCC VCC VCC(SUS) VCC VCC During Reset Low** Input Input High High High High High Input Input Input Input Low Input Hi-Z Undef Undef Undef High High High Undef Undef Hi-Z After Reset Hi-Z Input Input High High Low High High Input Input Input Input Low Input Hi-Z High Undef Undef High High High High High Hi-Z Chip Standby Hi-Z Input Input High High Low High High Input Input Input Input Low Input Hi-Z During POS Hi-Z Input Input High High Low High High Input Input Input Input Low Input Hi-Z During STR Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn
MTXC Signals Host Interface A[31:3] BE[7:0] ADS# BRDY# NA# AHOLD EADS# BOFF# HITM# W/R# HLOCK# CACHE# KEN#/INV SMIACT# HD[63:0] RAS[5:0]/ CS[5:0]# CAS[7:0]#/ DQM[7:0] MA[11:0] MWE# MWEB# SRAS[A,B]# SCAS[A,B]# CKE CKEB MD[63:0]
Type I/O I I O O O O O I I I I O I I/O O O O O O O O O I/O
Buffer* 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3/2.5V 3.3 V 3.3 V 3.3 V 3.3 V 3.3V 3.3V 3.3V 3.3V 3.3/5V
Max Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling
DRAM Interface
Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Toggling Low Low Pwrdn Toggling High High High Toggling Toggling Low Pwrdn Toggling Toggling Low Pwrdn High High Low Low High High Low Pwrdn Toggling Hi-Z Hi-Z Pwrdn
CADV# VCC O 3 .3V High High Toggling High High Pwrdn CADS# VCC O 3 .3V High High Toggling High High Pwrdn CCS# VCC O 3 .3V Low Low Low Low Low Pwrdn COE# VCC O 3.3 V High High Toggling High High Pwrdn GWE# VCC O 3.3 V High High Toggling High High Pwrdn BWE# VCC O 3.3 V High High Toggling High High Pwrdn TIO[7:0] VCC I/O 3.3 V Low Hi-Z Toggling Hi-Z Hi-Z Pwrdn TWE# VCC O 3.3 V Low High Toggling High High Pwrdn KRQAK / FRCL# VCC I/O 3.3V Input Input Toggling Input Input Pwrdn * 3.3/2.5V indicates the buffer is 3.3V or 2.5V only, depending upon the VCC(CPU) voltage. 3.3/5V indicates that the output is 3.3V, and input is 3.3V with 5V tolerance. 5V indicates 3.3V input with 5V tolerance. ** A[31:26] are inputs during Reset.
L2 Cache Interface
PRELIMINARY
77
Extended Temperature 82439TX (MTXC) Datasheet
MTXC Signals PCI AD[31:0] Interface C/BE[3:0]# FRAME# DEVSEL# IRDY# TRDY# STOP# LOCK# REQ[3:0]# GNT[3:0]# PHLD# PHLDA# PAR CLKRUN# RST# Test / Clock Signals Power Mgmt TEST# HCLKIN PCLKIN SUSCLK SUS_STAT#
Power Plane VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC VCC(CPU) VCC VCC(SUS) VCC(SUS)
Type I/O I/O I/O I/O I/O I/O I/O I/O I O I O I/O I/O I I I I I I
Buffer 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3 V 3.3/5 V 3.3 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/5 V 3.3/2.5V 3.3/5 V 3.3 V 3.3 V
During Reset Low Low Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Input Hi-Z Input High Low Hi-Z Input Input Input Input Input Input
After Reset Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Input High Input High Undef Hi-Z Input Input Input Input Input Input
Chip Max Standby Toggling Last Toggling Last Toggling Hi-Z Toggling Hi-Z Toggling Hi-Z Toggling Hi-Z Toggling Hi-Z Toggling Hi-Z Input Input Toggling High Input Input Toggling High Toggling Undef Low Low High Input Input Input Input Input Input Input Input Input Input Input
During POS Last Last Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Hi-Z Input High Input High Undef High Input Input Input Input Input Input
During STR Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Pwrdn Input Input
For the Standby State, the following assumptions are made: * * * * * * Host Bus is idle PCI bus is idle DRAM bus is idle except for DRAM refresh cycles All external clocks are running: HCLK (66Mhz), PCLOCK (33MHz), SUSCLK (32KHz) Not in any suspend state No PCI activity
For the MAX state, the following assumptions are made: * * * Host Bus cycle in progress PCI bus cycle in progress DRAM bus cycle in progress
PRELIMINARY
78
Extended Temperature 82439TX (MTXC) Datasheet
6.4. MTXC AC Characteristics
All timings are in nanoseconds (ns), unless otherwise specified.
Table 4. HOST CLOCK TIMING; 66 MHz (MTXC)
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE =-40oC to +115oC) Symbol Parameter 66 MHz Figures Min t1 t1s t1H t1L t1r t1f HCLKIN Period HCLKIN Period Stability HCLKIN High Time HCLKIN Low Time HCLKIN Rise Time HCLKIN Fall Time 5.5 5.5 1.5 1.5 15.0 Max 20.0 250 12 12 12 12 12 pS Notes
PRELIMINARY
79
Extended Temperature 82439TX (MTXC) Datasheet
Table 5. CPU INTERFACE TIMING; 66 MHz (MTXC)
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE = -40oC to +115oC) Symbol Parameter 66 MHz Min t2 t3 t4 t5 t6 t7 t8 t9 t11 ADS# Setup Time to HCLKIN Rising W/R# Setup Time to HCLKIN Rising BE[7:0]# Setup Time to HCLKIN Rising HITM# Setup Time to HCLKIN Rising CACHE# Setup Time to HCLKIN Rising M/IO# Setup Time to HCLKIN Rising D/C# Setup Time to HCLKIN Rising HLOCK#, SMIACT# Setup Time to HCLKIN Rising ADS#, HITM#, W/R#, M/IO#, D/C#, BE[7:0], HLOCK#, CACHE#, SMIACT# Hold Time from HCLKIN Rising A[31:0] Setup Time to HCLKIN Rising A[31:0] Hold Time from HCLKIN Rising A[31:0] Valid Delay from HCLKIN Rising A[31:0] Output Enable From HCLKIN Rising A[31:0] Float Delay from HCLKIN Rising HD(63:0) Setup Time to HCLK Rising HD(63:0) Hold Time from HCLK Rising HD(63:0) Valid Delay from HCLK Rising HD(63:0) Flow Through Delay from MD[63:0], 66MHz, 5-2-2-2 BRDY# Valid Delay from HCLKIN Rising NA# Valid Delay from HCLKIN Rising AHOLD Valid Delay from HCLKIN Rising BOFF# Valid Delay from HCLKIN Rising EADS# Valid Delay from HCLKIN Rising 5.0 5.7 3.4 5.3 5.0 5.3 5.0 4.0 1.0 Max 15 15 15 15 15 15 15 15 15 Figures Notes
t12 t13 t14 t15 t16 t17 t18 t19 t20 t21 t22 t23 t24 t25
3.5 1.0 2.0 0.0 0.0 3.75 1.0 1.5 1.5 1.5 1.5 1.5 1.5 1.5 7.0 6.0 8.0 8.0 7.0 7.0 7.0 13.0 13.0 13.0
15 15 14 19 0 pF
0 pF 0 pF 14 14 14 14 14 0 pF 0 pF 0 pF 0 pF 0 pF
PRELIMINARY
80
Extended Temperature 82439TX (MTXC) Datasheet
t26
KEN#/INV Valid Delay from HCLKIN Rising
1.5
7.0
14
0 pF
Table 6. SECOND LEVEL CACHE TIMING; 66 MHz (MTXC)
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE = -40oC to +115oC) Symbol Parameter 66 MHz Min t27 t28 t29 COE# Valid Delay from HCLKIN Rising GWE# Valid Delay from HCLKIN Rising BWE# Valid Delay from HCLKIN Rising 2.0 2.0 2.0 Max 8.0 9.5 9.0 14 14 3 0 pF 0 pF 0 pF Figures Notes
t30 t31 t32 t33 t34 t35
KRQAK Valid Delay from HCLKIN KRQAK setup time to HCLKIN KRQAK Hold Time from HCLKIN TIO[7:0] Valid Delay from HCLKIN Rising TIO[7:0] Setup time to HCLKIN Rising TIO[7:0] Hold time to HCLKIN Rising
1.5 2.8 1.0 2.0 2.2 2.0
7.0
0 pF
8.0
3 15 15
0 pF
t36 t37 t38 t39
CCS# Valid Delay from HCLKIN Rising CADS# Valid Delay from HCLKIN Rising CADV# Valid Delay from HCLKIN Rising TWE# Valid Delay
1.5 1.5 1.5 2.0
7.0 7.0 7.0 9.0
14 14 14 14
0 pF 0 pF 0 pF 0 pF
PRELIMINARY
81
Extended Temperature 82439TX (MTXC) Datasheet Table 7. EDO/FPM DRAM INTERFACE TIMING; 66 MHz (MTXC)
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE = -40oC to +115oC) Symbol Parameter 66 MHz Min t40 t41 t42 t43 t44 t45 RAS[5:0]# Valid Delay from HCLK Rising CAS[7:0]# Valid Delay from HCLKIN Rising MWE#, MWEB# Valid Delay From HCLKIN Rising MA[13:0] Flow Through Delay from HA (read col addr) MA[13:0] Valid Delay from HCLK Rising (read row addr) MA[11:0] Valid Delay from HCLK Rising (read col addrs burst cycles) MA[1:0] Valid Delay from HCLK Rising (Write Row and Col Addr) MD[63:0] set up to HCLK Rising MD[63:0] hold time from HCLK Rising MD[63:0] Valid delay from HCLK Rising SDRAM Interface Timing t51 t52 t56 t57 t58 t59 t60 t61 t62 t62a SRAS[B:A]# Valid Delay from HCLKIN Rising (two clock path) SCAS[B:A]# Valid Delay from HCLKIN Rising (two clock path) MWE#,MWEB# Valid delay from HCLKIN Rising (Two Clock Path) CKE, CKEB Valid Delay from HCLKIN Rising CS[5:0]# Valid Delay from HCLKIN Rising DQM[7:0] Valid Delay from HCLKIN Rising MA[11:0] Valid Delay from HCLKIN Rising MD[63:0] set up to HCLK Rising MD[63:0] hold time from HCLK Rising MD[63:0] Valid Delay from HCLK Rising 5.0 5.0 5.0 1.5 1.5 1.5 2.0 3.5 1.0 2.0 8.0 19.0 19.0 19.0 6.0 7.0 6.0 16.0 3 3 3 14 0 pF 0 pF 0 pF 0 pF 0 pF 0 pF 0 pF 0 pF 0 pF 0 pF 1.5 1.5 1.5 2.0 2.0 2.0 Max 7.0 6.0 19.0 8.0 9.0 6.5 14 14 14 0 pF 0 pF 0 pF 0 pF 0 pF 0 pF (Also applies when CKE pins are used as copies of MA[1:0]) 0 pF Fig. Notes
t46 t48 t49 t50
2.0 1.10 4.0 2.0
10.0
8.0
0 pF
PRELIMINARY
82
Extended Temperature 82439TX (MTXC) Datasheet
Table 8. PCI CLOCK TIMING; 66 MHz (MTXC)
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE = -40oC to +115oC) Symbol Parameter 66 MHz Min PM and TEST Timing t63 t64 t64s t64t SUS_STAT# Setup Time SUS_STAT# Hold Time SUSCLK TESTIN# 7 2 Async Async Max Figure notes
PCI CLOCKS t65 t66 t66r t66f PCLKIN High Time PCLKIN Low Time PCLKIN Rise Time PCLKIN Fall Time 12.0 12.0 3.0 3.0 12 12 12 12
PRELIMINARY
83
Extended Temperature 82439TX (MTXC) Datasheet
Table 9. PCI INTERFACE TIMING; 66 MHz (MTXC)
Functional Operating Range (VCC = 3.13 V to 3.6V; VCC(CPU) = 2.37 V to 2.62 V / 3.13V to 3.6V; VREF = 5V 5%; TCASE = -40oC to +115oC) Symbol Parameter 66 MHz Min t67 t68 AD[31:0] Valid Delay AD[29:0] Setup Time AD[31:30] t63 t70 AD[31:0] Hold Time from PCLKIN C/BE[3:0]#, FRAME#, TRDY#, IRDY#, STOP#, LOCK#, PAR, DEVSEL# Valid Delay from PCLKIN Rising C/BE[3:0]#, FRAME#, TRDY#, IRDY#, STOP#, LOCK#, PAR, DEVSEL# Output Enable Delay from PCLKIN Rising C/BE[3:0]#, FRAME#, TRDY#, IRDY#, STOP#, LOCK#, PAR, DEVSEL# Float Delay from PCLKIN Rising C/BE[3:0]#, FRAME#, TRDY#, IRDY#, STOP#, LOCK#, PAR, DEVSEL# Setup Time to PCLKIN Rising C/BE[3:0]#, FRAME#, TRDY#, IRDY#, STOP#, LOCK#, PAR, DEVSEL# Hold Time from PCLKIN Rising PHLDA# Valid Delay from PCLKIN Rising GNT[3:0] # Valid Delay from PCLKIN Rising REQx#, PHLD# Setup Time from PCLKIN Rising REQx#, PHLD# Hold Time from PCLKIN Rising RST# Low Pulse Width CLKRUN# 2 7 7.7 0 2.0 11.0 Max 11 14 15 4 15 14 Figures Notes
t71
2.0
11.0
19
t72
2.0
11.0
16
t73
7.0
15
t74
0.0
15
t75 t76 t77 t78 t79 t80
2 2 12.0 0.0 1 ms Async
9.0 9.0
14
18
PRELIMINARY
84
Extended Temperature 82439TX (MTXC) Datasheet
7.0. MTXC Timing Diagrams
Period High Time
VIHmin HCLKIN PCLKIN Fall Time VILmax
VIHmin Rise Time VILmax
Low Time
clocktm.drw
Figure 12. Clock Timing
Input
VT Propagation Delay
Output
VT
prop_del.drw
Figure 13. Propagation Delay
PRELIMINARY
85
Extended Temperature 82439TX (MTXC) Datasheet
Clock
VT
Valid Delay
Output
VT Valid Delay
Output
VT
val_del.drw
Figure 14. Valid Delay From Rising Clock Edge
Clock Setup Time
VT Hold Time
Input
VT
VT
sethold.drw
Figure 15. Setup and Hold Times
PRELIMINARY
86
Extended Temperature 82439TX (MTXC) Datasheet
Input
VT Float Delay
Output
floatdel.DRW
Figure 16. Float Delay
HCLKIN
ADS# A[31:3], BE[7:0] t43 MA[11:0],
flowthru.drw
NOTES: The flow through delay is for the leadoff cycle during a DRAM access.
Figure 17. Flow Through Delay
PRELIMINARY
87
Extended Temperature 82439TX (MTXC) Datasheet
VT
VT
Pulse Width
pulsewid.drw
Figure 18. Pulse Width
Clock
VT Output Enable Delay
Output
outendel.drw
Figure 19. Output Enable Delay
PRELIMINARY
88
Extended Temperature 82439TX (MTXC) Datasheet
8.0.
PINOUT INFORMATION
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20 A
HD63
AD31
AD29
AD27
C/BE3#
AD21
AD18
C/BE2#
AD14
AD11
AD8
AD6
AD3
AD1
AD0
PAR
MD29
MD60
MD27
MD43
B
HD62 AD30 AD28 AD26 AD23 AD20 AD17 C/BE1# AD13 AD10 C/BE0# AD5 AD2 MD31 MD14 MD13 MD61 MD12 MD59 MD58
C
HD59 HD60 HD61 AD25 AD22 AD19 AD16 AD15 AD12 AD9 AD7 AD4 CLKRUN# REQ3# MD47 MD63 MD45 MD28 MD62 MD10 MD42
D
HD55 HD58 HD57 PHLD# PHLDA# LOCK# AD24 REQ0#
GNT0#
REQ1#
GNT1#
REQ2#
GNT2#
GNT3#
MD46
MD30
MD44
MD26
MD57
MD09
E
HD52 HD54 HD56 HD53 FRAME# VCC(CPU) IRDY# TRDY# DEVSEL# PCLKIN STOP# VCC MD15 VCC5REF VSS MD49 MD25 MD41 MD24 MD56
F
HD48 HD47 HD51 HD50 VCC
VCC
VCC
MD48
MD33
MD8
MD40
MD19
G
HD45 HD41 HD49 HD43 HLO- VCC(CPU) CK# M/IO# MD35 MD32
MD18
MD36
MD50
H
HD39 HD40 HD46 HD44 MD16 MD0
MD4
MD5
MD17
J
HD37 HD36 HD42 HD38 CACHE#
VSS
VSS
VSS
VSS
MD1
MD2
MD52
MD55
MD23
K
HD34 BE0# BE1# BE2# KEN#
VSS
VSS
VSS
VSS
HCLKIN
VCC
MD34
MD39
MD7
MD54
L
BE3# BE4# BE5# BE6# AHOLD
VSS
VSS
VSS
VSS
MD3
MD37
MD38
MD6
M
BE7# HD33 HD32 HD35 BRDY#
VSS
VSS
VSS
VSS
RAS5#/ MD53 CS5#/MA13
VCCSUS RAS 3#/ CS3#
SCASB# SRASA# RAS1#/ CS1# MWE#
MD20
MD22
N
HD27 HD30
HD29
HD31
NA#
MD21
MD51
P
HD23 HD26
HD25
HD28
BOFF#
VCC
VCCSUS VCCSUS VSS
CAS6#/ DQM6# RAS2#/ CS2# CAS3#/ DQM3# MA9
SCASA# SUS STAT1#
SRASB#
R
HD7 HD21 HD19 HD24
EADS# VCC (CPU)
VCC (CPU) D/C# HITM#
W/R#
VCC
MD11
T
HD12 HD17 HD22 HD20 ADS# VSS
ACT#
A22
SMI-
A6
TIO3
VCC
MA3
RST#
CAS7#/ CAS4#/ DQM7# DQM4# CAS0#/ CAS5#/ DQM0# DQM5#
SUSCLK
U
HD8 HD18 HD14 HD16 A20 A16 A12 A5 A23 A29 CADS# TIO6 TIO0 MA4 MA10 CKE/ MAA0
V
HD6 HD15 HD10 HD13 A19 A14 A9 A8 A21 A26 A3 COE# GWE# TIO2 MA0 MA1 KRQAK/ CS4_64# TEST # MWE- RAS4#/ CAS2#/ B# CS4#/BA1 DQM2#
W
HD4
HD5
HD9
HD11
A18
A15
A11
A31
A25
A24
A30
CADV#
CCS#
TIO7
TIO4
CKEB/ MAA1
MA6
MA8
CAS1#/ DQM1#
Y
HD0 HD2 HD1 HD3
A17 A13
A10
A7
A27
A28
A4
BWE#
TWE#
TIO1
TIO5
MA5
MA2
MA7
MA11/ BA0
RAS0#/ CS0#
mtxc_pin
Figure 20. MTXC Pinout (Top View)
PRELIMINARY
89
Extended Temperature 82439TX (MTXC) Datasheet
Table 24. MTXC Alphabetical Pin List Pin A10 A11 A12 A13 A14 A15 A16 A17 A18 A19 A20 A21 A22 A23 A24 A25 A26 A27 A28 A29 A3 A30 A31 A4 A5 A6 A7 A8 A9 AD0 Y07 W07 U07 Y06 V06 W06 U06 Y05 W05 V05 U05 V09 U10 U09 W10 W09 V10 Y09 Y10 U11 V11 W11 W08 Y11 U08 T11 Y08 V08 V07 A15 Ball
Table 24. MTXC Alphabetical Pin List Pin AD1 AD10 AD11 AD12 AD13 AD14 AD15 AD16 AD17 AD18 AD19 AD2 AD20 AD21 AD22 AD23 AD24 AD25 AD26 AD27 AD28 AD29 AD3 AD30 AD31 AD4 AD5 AD6 AD7 AD8 A14 B10 A10 C09 B09 A09 C08 C07 B07 A07 C06 B13 B06 A06 C05 B05 D06 C04 B04 A04 B03 A03 A13 B02 A02 C12 B12 A12 C11 A11 Ball
Table 24. MTXC Alphabetical Pin List Pin AD9 ADS# AHOLD BE0# BE1# BE2# BE3# BE4# BE5# BE6# BE7# BOFF# BRDY# BWE# C/BE0# C/BE1# C/BE2# C/BE3# CACHE# CADS# CADV# CAS0#/DQM0# CAS1#/DQM1# CAS2#/DQM2# CAS3#/DQM3# CAS4#/DQM4# CAS5#/DQM5# CAS6#/DQM6# CAS7#/DQM7# CCS# C10 T05 L05 K02 K03 K04 L01 L02 L03 L04 M01 P05 M05 Y12 B11 B08 A08 A05 J05 U12 W12 U18 W20 V20 T17 T19 U19 P17 T18 W13 Ball
PRELIMINARY
90
Extended Temperature 82439TX (MTXC) Datasheet
Table 24. MTXC Alphabetical Pin List Pin CKE/MAA0 CKEB/MAA1 CLKRUN# COE# D/C# DEVSEL# EADS# FRAME# GNT0# GNT1# GNT2# GNT3# GWE# HCLKIN HD0 HD1 HD10 HD11 HD12 HD13 HD14 HD15 HD16 HD17 HD18 HD19 HD2 HD20 HD21 HD22 U20 W16 C13 V12 T07 E09 R05 E06 D08 D10 D12 D14 V13 K16 Y01 Y03 V03 W04 T01 V04 U03 V02 U04 T02 U02 R03 Y02 T04 R02 T03 Ball
Table 24. MTXC Alphabetical Pin List Pin HD23 HD24 HD25 HD26 HD27 HD28 HD29 HD3 HD30 HD31 HD32 HD33 HD34 HD35 HD36 HD37 HD38 HD39 HD4 HD40 HD41 HD42 HD43 HD44 HD45 HD46 HD47 HD48 HD49 HD5 P01 R04 P03 P02 N01 P04 N03 Y04 N02 N04 M03 M02 K01 M04 J02 J01 J04 H01 W01 H02 G02 J03 G04 H04 G01 H03 F02 F01 G03 W02 Ball
Table 24. MTXC Alphabetical Pin List Pin HD50 HD51 HD52 HD53 HD54 HD55 HD56 HD57 HD58 HD59 HD6 HD60 HD61 HD62 HD63 HD7 HD8 HD9 HITM# HLOCK# IRDY# KEN# KRQAK/ CS4_64# LOCK# M/IO# MA0 MA1 MA10 MA11/BA0 F04 F03 E01 E04 E02 D01 E03 D03 D02 C01 V01 C02 C03 B01 A01 R01 U01 W03 T08 G05 E07 K05 V17 E05 H05 V15 V16 U16 Y19 Ball
PRELIMINARY
91
Extended Temperature 82439TX (MTXC) Datasheet
Table 24. MTXC Alphabetical Pin List Pin MA2 MA3 MA4 MA5 MA6 MA7 MA8 MA9 MD0 MD1 MD10 MD11 MD12 MD13 MD14 MD15 MD16 MD17 MD18 MD19 MD2 MD20 MD21 MD22 MD23 MD24 MD25 MD26 MD27 MD28 Y17 T14 U15 Y16 W18 Y18 W19 U17 H17 J16 C19 R20 B18 B16 B15 E13 H16 H20 G18 F20 J17 M19 N19 M20 J20 E19 E17 D18 A19 C17 Ball
Table 24. MTXC Alphabetical Pin List Pin MD29 MD3 MD30 MD31 MD32 MD33 MD34 MD35 MD36 MD37 MD38 MD39 MD4 MD40 MD41 MD42 MD43 MD44 MD45 MD46 MD47 MD48 MD49 MD5 MD50 MD51 MD52 MD53 MD54 MD55 A17 L17 D16 B14 G17 F17 K17 G16 G19 L18 L19 K18 H18 F19 E18 C20 A20 D17 C16 D15 C14 F16 E16 H19 G20 N20 J18 M17 K20 J19 Ball
Table 24. MTXC Alphabetical Pin List Pin MD56 MD57 MD58 MD59 MD6 MD60 MD61 MD62 MD63 MD7 MD8 MD9 MWE# MWEB# NA# PAR PCLKIN PHLD# PHLDA# RAS0#/CS0# RAS1#/CS1# RAS2#/CS2/# RAS3#/CS3# RAS4#/CS4#/ BA1 RAS5#/CS5#/ MA13 REQ0# REQ1# REQ2# REQ3# E20 D19 B20 B19 L20 A18 B17 C18 C15 K19 F18 D20 R18 V18 N05 A16 E10 D04 D05 Y20 P18 R17 N17 V19 M16 D07 D09 D11 D13 Ball
PRELIMINARY
92
Extended Temperature 82439TX (MTXC) Datasheet
Table 24. MTXC Alphabetical Pin List Pin RST# SCASA# SCASB# SMIACT# SRASA# SRASB# STOP# SUSCLK SUSSTAT1# TEST# TIO0 TIO1 T15 P19 M18 T10 N18 P20 E11 T20 R19 W17 U14 Y14 Ball
Table 24. MTXC Alphabetical Pin List Pin TIO2 TIO3 TIO4 TIO5 TIO6 TIO7 TRDY# TWE# W/R# VCC V14 T12 W15 Y15 U13 W14 E08 Y13 T09 F05, L16, R15, F15, E12, P15, F14, T13 Ball
Table 24. MTXC Alphabetical Pin List Pin VCC (CPU) VCC (SUS) VCC5REF VSS Ball F06, G06, R07, R06 R16, N16, P16 E14 E15, J9, J10 J11, J12, K09, K10, K11, K12, L09, L10, L11, L12, M09, M10, M11, M12, T06, T16
PRELIMINARY
93
Extended Temperature 82439TX (MTXC) Datasheet
9.0.
MTXC PACKAGE INFORMATION
This specification outlines the mechanical dimensions for the MTXC. The package is a 324 pin ball grid array (BGA).
TOP VIEW Pin #1 Corner D D1
SIDE VIEW
Pin #1 I.D .
E1
E
A2 A1 A f c
mtxc_10
Figure 21. MTXC 324-pin Ball Grid Array (BGA)
PRELIMINARY
94
Extended Temperature 82439TX (MTXC) Datasheet
Pin #1 Corner
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 A
b
B C D E F G H J K L M N P R T U V W Y
e
324 Balls 20 X 20 Matrix
J I TOP VIEW
e
mtxc_09
Figure 22. MTXC 324-pin Ball Grid Array (BGA) Ball Pattern
PRELIMINARY
95
Extended Temperature 82439TX (MTXC) Datasheet
Table 25. MTXC 324-pin Ball Grid Array (BGA) Symbol Min1 A A1 A2 D D1 E E1 I J M2 N3 b c f Remark 1.44 REF. 1.44 REF. 20 (Depopulated) 324 0.60 0.32 8.05 REF. 2 Layer 0.76 0.36 0.90 0.40 26.80 1.95 0.50 1.12 26.80 2.13 0.60 1.17 27.00 24.00 27.00 24.00 e=1.27 (solder ball pitch) Nominal1 2.28 0.70 1.22 27.20 24.70 27.20 24.70 Max1
NOTES: 1. All dimensions are in millimeters. 2. `M' represents the maximum solder ball matrix size. 3. `N' represents the maximum allowable number of solder balls.
PRELIMINARY
96
Extended Temperature 82439TX (MTXC) Datasheet
10.0. TESTABILITY 10.1. NAND Tree Mode
A NAND tree mode is provided for Automated Test Equipment (ATE) board level testing. The NAND tree allows the tester to set the connectivity of each of the MTXC's signal pins. In Mobile/desktop mode, the NAND tree mode is activated by driving the test pin TEST# low when REQx# pins are at 0010. If TEST# is negated at any time, the test mode is deactivated and the MTXC goes back to normal operation. There is no guarantee that upon re-entering normal operation the chip will function properly, if the test mode was entered while the MTXC was not in a completely idle state. The MTXC has several test modes to improve board manufacturing. If the TEST# signal is asserted (driven low), the value on the REQ[3:0]# indicates the test mode to enable. The test mode enabled at the falling edge of TEST# will remain enabled until TEST# is negated. Table 26 shows each test mode and the value of REQ#[3:0] required to enable it. All other values of REQ[3:0]# while TEST# is active, are reserved and should not be asserted by the customer. Table 26. Test Modes REQ[3:0]# 0010 1110 1110 1111 TEST# 0 0 0 0 PHLD# x 1 0 x TEST Mode NAND Chain ID/REV code MID Code Disable test mode1 Description Float outputs, enable NAND chains on GNT[3:0]# Drives Device ID on AD[31:16] and revision ID on AD[7:0] Drives Manufacture ID on AD[31:0] Disables any active test mode, puts MTXC back into normal mode
NOTES: 1. It is recommended to assert RST# if this mode is used, to guarantee pins and PCIset will function normally.
10.2.
NAND Chain Mode
In NAND Tree mode, all outputs are tri-stated, except for GNT#[3:0]. These pins contain the NAND Chain. Note, also, that the internal pull-ups and pull-downs are still active. Because of the 282 pins in the NAND Chain, it must be separated into 4 chains. Two chains contain 72 pins each, 1 chain contains 70 pins, and 1 chain contains 68 pins. The MTXC remains in this mode until a new test mode is selected or RST# is asserted. The HCLK and PCLK are part of the NAND Chain and must be deactivated during this test.
PRELIMINARY
97
Extended Temperature 82439TX (MTXC) Datasheet
HCLK and PCLK need to run for a few clocks in the beginning to put the MTXC in the NAND chain mode. During the testing of chains 2 and 3, SUSSTAT# will be held high throughout the test. RST#, TEST#, and SUSSTAT1# are not part of the NAND chain. The following tables show the pin order for each chain: Table 27. Chain #0 (GNT#0) Pin Name A25 A7 A10 A21 A23 SMIACT# A31 A18 HD3 A12 A14 HD11 A19 A16 HITM# HD13 D/C# HD0 HD10 HD16 HD6 HD14 BOFF# HD18 Chain Element CH0_00 CH0_01 CH0_02 CH0_03 CH0_04 CH0_05 CH0_06 CH0_07 CH0_08 CH0_09 CH0_10 CH0_11 CH0_12 CH0_13 CH0_14 CH0_15 CH0_16 CH0_17 CH0_18 CH0_19 CH0_20 CH0_21 CH0_22 CH0_23 Table 27. Chain #0 (GNT#0) Pin Name HD24 HD8 HD19 HD7 HD31 HD26 HD27 HD33 AHOLD BE5# BE6# HD34 BE2# BE0# HD37 BE1# HD36 HD41 HD52 CACHE# HD47 HD44 HD49 HD54 Chain Element CH0_24 CH0_25 CH0_26 CH0_27 CH0_28 CH0_29 CH0_30 CH0_31 CH0_32 CH0_33 CH0_34 CH0_35 CH0_36 CH0_37 CH0_38 CH0_39 CH0_40 CH0_41 CH0_42 CH0_43 CH0_44 CH0_45 CH0_46 CH0_47 Table 27. Chain #0 (GNT#0) Pin Name HD55 HD57 HLOCK# HD53 HD60 HD62 HD61 AD30 AD25 AD22 AD24 AD27 AD19 C/BE3# REQ0# AD20 AD21 TRDY# AD16 C/BE2# C/BE1# AD14 AD12 AD13 Chain Element CH0_48 CH0_49 CH0_50 CH0_51 CH0_52 CH0_53 CH0_54 CH0_55 CH0_56 CH0_57 CH0_58 CH0_59 CH0_60 CH0_61 CH0_62 CH0_63 CH0_64 CH0_65 CH0_66 CH0_67 CH0_68 CH0_69 CH0_70 CH0_71
PRELIMINARY
98
Extended Temperature 82439TX (MTXC) Datasheet
Table 28. Chain #1 (GNT#1) Pin Name A24 A27 A26 A13 A8 W/R# A11 A17 A15 A5 A9 HD1 HD9 A20 HD2 HD5 HD4 ADS# HD15 EADS# HD20 HD17 HD22 HD12 Chain Element CH1_00 CH1_01 CH1_02 CH1_03 CH1_04 CH1_05 CH1_06 CH1_07 CH1_08 CH1_09 CH1_10 CH1_11 CH1_12 CH1_13 CH1_14 CH1_15 CH1_16 CH1_17 CH1_18 CH1_19 CH1_20 CH1_21 CH1_22 CH1_23
Table 28. Chain #1 (GNT#1) Pin Name HD28 HD21 NA# HD25 BRDY# HD23 HD35 HD29 HD30 BE7# HD32 BE3# BE4# HD39 HD45 HD42 HD38 HD40 HD48 KEN# HD46 HD43 HD51 HD58 Chain Element CH1_24 CH1_25 CH1_26 CH1_27 CH1_28 CH1_29 CH1_30 CH1_31 CH1_32 CH1_33 CH1_34 CH1_35 CH1_36 CH1_37 CH1_38 CH1_39 CH1_40 CH1_41 CH1_42 CH1_43 CH1_44 CH1_45 CH1_46 CH1_47
Table 28. Chain #1 (GNT#1) Pin Name HD56 HD50 M/IO# HD59 HD63 PHLD# AD31 LOCK# AD28 FRAME# AD29 PHLDA# IRDY# AD26 AD23 AD17 DEVSEL# AD18 REQ1# AD15 PCLKIN AD11 AD10 AD9 Chain Element CH1_48 CH1_49 CH1_50 CH1_51 CH1_52 CH1_53 CH1_54 CH1_55 CH1_56 CH1_57 CH1_58 CH1_59 CH1_60 CH1_61 CH1_62 CH1_63 CH1_64 CH1_65 CH1_66 CH1_67 CH1_68 CH1_69 CH1_70 CH1_71
PRELIMINARY
99
Extended Temperature 82439TX (MTXC) Datasheet
Table 29. Chain #2 (GNT#2) Pin Name A28 A29 A3 A30 A4 COE# BWE# CCS# MA5 MA10 KRQAK/ CS4_64# MA7 MA11 MA6 MA9 MWEB# RAS0# RAS4# RAS2# CAS7# CAS5# CAS6# CKE/MAA0 Chain Element CH2_00 CH2_01 CH2_02 CH2_03 CH2_04 CH2_05 CH2_06 CH2_07 CH2_08 CH2_09 CH2_10 CH2_11 CH2_12 CH2_13 CH2_14 CH2_15 CH2_16 CH2_17 CH2_18 CH2_19 CH2_20 CH2_21 CH2_22
Table 29. Chain #2 (GNT#2) Pin Name CAS4# SUSCLK SRASA MD11 MD53 SCASB# MD22 MD6 MD34 MD39 MD7 MD54 HCLKIN MD55 MD5 MD17 MD2 MD50 MD1 MD0 MD18 MD16 MD19 MD40 Chain Element CH2_23 CH2_24 CH2_25 CH2_26 CH2_27 CH2_28 CH2_29 CH2_30 CH2_31 CH2_32 CH2_33 CH2_34 CH2_35 CH2_36 CH2_37 CH2_38 CH2_39 CH2_40 CH2_41 CH2_42 CH2_43 CH2_44 CH2_45 CH2_46
Table 29. Chain #2 (GNT#2) Pin Name MD32 MD25 MD26 MD42 MD48 MD49 MD59 MD43 MD30 MD27 MD28 MD12 MD60 CLKRUN# AD0 AD2 STOP# AD4 AD3 AD5 AD7 AD6 REQ2# Chain Element CH2_47 CH2_48 CH2_49 CH2_50 CH2_51 CH2_52 CH2_53 CH2_54 CH2_55 CH2_56 CH2_57 CH2_58 CH2_59 CH2_60 CH2_61 CH2_62 CH2_63 CH2_64 CH2_65 CH2_66 CH2_67 CH2_68 CH2_69
PRELIMINARY
100
Extended Temperature 82439TX (MTXC) Datasheet
Table 30. Chain #3 (GNT#3) Pin Name A22 CADV# A6 TWE# GWE# CADS# TIO1 TIO3 TIO7 TIO6 TIO2 TIO5 TIO4 TIO0 MA0 MA2 MA4 CKEB/ MAA1 MA1 MA3 MA8 CAS1# CAS3# Chain Element CH3_00 CH3_01 CH3_02 CH3_03 CH3_04 CH3_05 CH3_06 CH3_07 CH3_08 CH3_09 CH3_10 CH3_11 CH3_12 CH3_13 CH3_14 CH3_15 CH3_16 CH3_17 CH3_18 CH3_19 CH3_20 CH3_21 CH3_22
Table 30. Chain #3 (GNT#3) Pin Name CAS0# CAS2# MWE# RAS1# RAS3# RAS5# SCASA MD21 SRASB# MD51 MD20 MD37 MD38 MD3 MD52 MD23 MD4 MD36 MD56 MD8 MD9 MD33 MD24 Chain Element CH3_23 CH3_24 CH3_25 CH3_26 CH3_27 CH3_28 CH3_29 CH3_30 CH3_31 CH3_32 CH3_33 CH3_34 CH3_35 CH3_36 CH3_37 CH3_38 CH3_39 CH3_40 CH3_41 CH3_42 CH3_43 CH3_44 CH3_45
Table 30. Chain #3 (GNT#3) Pin Name MD41 MD57 MD35 MD58 MD10 MD44 MD62 MD15 MD46 MD45 MD61 MD63 MD29 MD13 MD47 REQ3# MD14 PAR MD31 AD1 C/BE0# AD8 Chain Element CH3_46 CH3_47 CH3_48 CH3_49 CH3_50 CH3_51 CH3_52 CH3_53 CH3_54 CH3_55 CH3_56 CH3_57 CH3_58 CH3_59 CH3_60 CH3_61 CH3_62 CH3_63 CH3_64 CH3_65 CH3_66 CH3_67
PRELIMINARY
101
Extended Temperature 82439TX (MTXC) Datasheet
11.0. ERRATA 11.1. SDRAM Speculative Read Enable (SSRE)
PROBLEM: Due to a timing marginality during SDRAM read-page-hit cycles, the SSRE mode does not
function properly and must be disabled. This mode provided a five clock read lead off during CPU read-pagehit cycles to main memory. This mode impacts SDRAM only.
IMPLICATION: If SSRE mode is enabled, the system may not function properly. Intel has tested this mode and found the performance impact of disabling the SSRE mode is less than 0.20% for Winstone '96 and Winstone '97 benchmarks on the 430TX reference platform with 200MHz Pentium(R) processor with MMXTM technology, 16MB SDRAM, 512K PBSRAM L2, and hard drive in PIO Mode 4, WORKAROUND: This errata is avoided by disabling the SSRE mode by clearing bit 7 in register offset 56h to
"0". This bit has been changed to a reserved bit and must always remain in its default state (0).
STATUS: There are no plans to fix this erratum.
11.2.
Fast Back-to-Back, PCI Peer-to-Peer Cycles
PROBLEM: If a fast back-to-back cycle occurs on the PCI bus between the same master peer device and the
same slave peer device (i.e. MTXC and PIIX4 are not the intended targets), and at the same time a CPU cycle to PCI occurs, the MTXC may miss the second peer-to-peer cycle in the back-to-back sequence. The specific conditions that need to be met for this to occur, are as follows: 1. The PCI peer-to-peer, back-to-back transfer must happen at the same time the CPU is generating a host cycle to PCI. Specifically, between the rising edge of the first FRAME# (associated with the first cycle of the back-to-back transfer) to the rising edge of the second FRAME# (associated with the second cycle in the back-to-back transfer). 2. The PCI back-to-back transfer must be peer-to-peer and the transfer must not be targeting the PIIX4 or the MTXC. 3. The PCI master must be capable of running "fast" back-to-back cycles. 4. The second cycle of the back-to-back transfer must be targeting the same PCI slave device. 5. The PCI slave device must be doing a "fast" decode. 6. The first cycle in the back-to-back transfer must be a single transfer. 7. The first cycle in the back-to-back transfer must be a PCI write cycle (memory or I/O).
IMPLICATION: Intel observed this issue in a system simulation environment. WORKAROUND: If deemed necessary by the OEM, this issue can be avoided by clearing bit 3 in register offset 50h to "0". When set to "0", this bit prevents CPU bus access during PCI peer-to-peer transfers. STATUS: There are no plans to fix this erratum.
PRELIMINARY
102

▲Up To Search▲

Price & Availability of FW82439TX

	To Download FW82439TX Datasheet File
If you can't view the Datasheet, Please click here to try to view without PDF Reader .