[Home] [This version is outdated, a new version is here]
*Title...
*Search:...
*Read Me/FAQ/General Info...
*_IBM...
*ACC Micro...
*ALD...
*ALi...
*AMD . . . . . . . [no datasheets, some info]...
*Chips & Technologies...
*Contaq . . . . . [no datasheets, some info]...
*Efar Microsystems [no datasheets, some info]...
*ETEQ...
*Faraday...
*Forex . . . . . . [List only, no datasheets found]...
*Intel...
**82395DX High Performance Smart Cache 06/18/90
***Notes:...
***Info:
The 82395DX High Performance 82395DX Smart Cache is a low cost, high
integration, 32-Bit peripheral for Intel's i386 DX Microprocessor. It
stores a copy of frequently accessed code or data from main memory to
on chip data RAM that can be accessed in zero wait states. The 82395DX
enables the 386 DX Microprocessor to run at near its full potential by
reducing the average number of wait states seen by the CPU to nearly
zero. The dual bus architecture allows another bus master to access
the System Bus while the 386 DX Microprocessor can operate out of the
82395DX's cache on the Local Bus. The 82395DX has a snooping
mechanism which maintains cache coherency during these cycles.
The 8239SDX is completely software transparent, protecting the
integrity of system software. High performance, low cost and board
space saving are achieved due to the high integration and new write
buffer architecture.
1.0 82395DX FUNCTIONAL OVERVIEW
1.1 Introduction
The primary function of a cache is to provide local storage for
frequently accessed memory locations. The cache intercepts memory
references and handles them directly without transferring the request
to the System Bus. This results in lower traffic on the System Bus and
decreases latency on the local bus. This leads to improved
performance for a processor on the Local Bus. By providing fast access
to frequently used code and data, the cache is able to reduce the
average memory access time of the 386 DX Microprocessor based system.
The 82395DX is a single chip cache subsystem specifically designed for
use with the 386 DX Microprocessor. The 82395DX integrates 16KB
cache, the Cache Directory and the Cache Control Logic onto one chip.
The 82395DX is expandable such that larger cache sizes are supported
by cascading 82395DXs. In a single 82395DX system, the 82395DX can map
4 Giga bytes of main memory into a 16KB cache. In the maximum con-
figuration of a four 82395DX system, the 4 Giga bytes of main memory
are mapped into a 64KB cache. The cache is unified for code and data
and is transparent to application software. The 82395DX provides a
cache consistency mechanism which guarantees that the cache has the
most recently updated version of the main memory. Consistency support
has no performance impact on the 386 DX Microprocessor. Section 1.2
covers all the 82395DX features.
The 8239SDX cache architecture is similar to the i486 Microprocessor’s
on-chip cache. The cache is four Way set associative with Pseudo LRU
replacement algorithm. The line size is 16B and a full line is
retrieved from the memory every cache miss. A TAG is associated with
every 16B line.
The 82395DX architecture allows for cache read hit cycles to run on
the Local Bus even when the System Bus is not available. 82395DX
incorporates a new write buffer cache architecture, which allows the
386 DX Microprocessor to continue operation without waiting for write
cycles to actually update the main memory.
A detailed description of the cache operation and parameters is
included in chapter 2.
The 82395DX has an interface to two electrically isolated busses. The
interface to the 386 DX Microprocessor bus is referred to as the Local
Bus (LB) interface. The interface to the main memory and other system
devices is referred to as the 82395DX System Bus (SB) interface. The
SB interface emulates the 386 DX Microprocessor. The SB interface, as
does the 386 DX Microprocessor, can be pipelined.
in addition, it is enhanced by an optional burst mode for Line
Fills. The burst mode provides faster line fills by allowing
consecutive read cycles to be executed at a rate of up to one DW per
clock cycle. Several bus masters (or several 82395DXs) can share the
same System Bus and the arbitration is done via the SHOLD/SHLDA/SBREQ
mechanism (similar to the i486 Microprocessor) along with
SFHOLD#. Using these arbitration mechanisms, the 82395DX is able to
support a multiprocessor system (multi 386 DX Microprocessor/82395DX
systems sharing the same memory).
Cache consistency is maintained by the SAHOLD/SEADS# snooping
mechanism, similar to the i486 microprocessor. The 82395DX is able to
run a zero wait state 386 DX Microprocessor non-pipelined read cycle
it the data exists in the cache. Memory write cycles can run with zero
wait states if the write buffer is not full.
The 82395DX cache organization provides a higher hit rate than other
standard configurations. The 82395DX, featuring the new high
performance write buffer cache architecture, provides full concurrency
between the electrically isolated Local Bus and System Bus. This
allows the 82395DX to service read hit cycles on the Local Bus while
running line fills or buffered write cycles on the System Bus.
Moreover, the user has the option to expand his cache system up to
64KB.
1.2 Features
1.2.1 82385-LIKE FEATURES
o The 82395DX maps the entire physical address range of the 386 DX
Microprocessor (4GB) into 16KB, 32KB, or 64KB cache (with one, two,
or four 82395DXs respectively).
o Unified code and data cache.
o Cache attributes are handled by hardware. Thus the 82395DX is
transparent to application software. This preserves the integrity of
system software and protects the users software investment.
o Double Word, Word and Byte writes, Double Word reads.
o Zero wait states in read hits and in buffered write cycles. All 386
DX Microprocessor cycles are non-pipelined. (Note: The 386 DX
Microprocessor must never be pipelined when used with the 82395DX -
NA# must be tied to Vcc).
o A hardware cache FLUSH# option. The 82395DX will invalidate all the
Tag Valid bits in the Cache Directory and clear the System Bus line
butter when FLUSH# is activated for a minimum of four CLK’s. The
line buffer is also FLUSH #ed.
o The 8239SDX supports non-cacheable accesses. The 82395DX internally
decodes the 387 DX Math Coprocessor accesses as Local Bus cycles.
o The system bus interface emulates a 386 DX Microprocessor interface.
o The 82395DX supports pipelined and non-pipelined system interface.
o Provides cache consistency (snooping): The 82395DX monitors the
System Bus address via SEADS# and invalidates the cache address if
the System Bus address matches a cached location.
1.2.2 NEW FEATURES
o 16KB on chip cache arranged in four banks, one bank for each way. In
Read hit cycles, one DW is read. In a write hit cycle, any byte
within the DW can be written. In cache fill cycle, the whole line
(16B) is written. This large line size increases the hit rate over
smaller line size caches.
o Cache architecture similar to the i486 Microprocessor cache: Four
Way SET associative with Pseudo LRU replacement algorithm. Line size
is 16B and a full line is retrieved from memory for every cache
miss. Tag. Tag Valid Bit and Write Protect Bit are associated with
every Line.
o New write buffer architecture with four DW deep write buffer
provides zero wait state memory write cycles. I/O, Halt/ Shutdown
and LOCK#ed writes are not buffered.
o Concurrent Line Buffer Cacheing: The 82395DX has a line buffer that
is used as additional memory. Before data gets written to the cache
memory at the completion of a Line Fill it is stored in this buffer.
Cache hit cycles to the line buffer can occur before the line is
written to the cache.
o Expandable: two 82395DXs support 32KB cache memory, four 82395DXs
support 64KB cache memory. This gives the user the option of config-
uring a system to meet their own performance requirements.
o In 387 DX Math Coprocessor accesses, the 82895DX drives the READYO#
in one wait state if the READYI# was not driven in the previous
clock.
Note that the timing of the 82395’5 READYO# generation for 387 DX
Math Coprocessor cycles is incompatible with 80287 timing.
o The 82395DX optionally decodes CPU accesses to Weitek 3167
Floating-Point Coprocessor address space (COOOOOOOH-ClFFFFFFH) as
Local Bus cycles. This option is enabled or disabled according to
the LBA# pin value at the falling edge of RESET.
o An enhanced System Bus interface:
a) Burst option is supported in line-fills similar to the i486
Microprocessor. SBRDY# (System Burst READY) is provided in
addition to SRDY#. A burst is always a 16 byte cache update which
is equivalent to four DW cycles. The i486 Microprocessor burst
order is supported.
b) System cacheability attribute is provided (SKEN#). SKEN# is used
to determine whether the current cycle is cacheable. If is used
to qualify Line Fill requests.
c) SHOLD/SHLDA/SBREQ system bus arbitration mechanism is supp-
orted. the same as in the i486 Microprocessor. A Multi 386
DX/82395DX cluster can share the same System Bus via this
mechanism.
d) SNENE# output (Next Near) is provided to simplify the interface
to DRAM controllers. DRAM page size of 2K is supported.
e) Fast HOLD function (SFHOLD#) is provided. This function allows
for multiprocessor support.
f) Cache invalidation cycles supported via SEADS#. This is the
mechanism used to provide cache coherency.
o Full Local Bus/System Bus concurrency is attained by:
a) Servicing cache read hit cycles on the Local Bus while completing
a Line Fill on the System Bus. The data requested by the 386 DX
Microprocessor was provided over the local bus as the first part
of the Line Fill.
b) Servicing cache read hit cycles on the Local Bus while executing
buffered write cycles on the system bus.
c) Servicing cache read hit cycles on the Local Bus while another
bus master is running (DMA, other 386 DX Microprocessor, 82395DX,
i486 Microprocessor, etc...) on the System Bus.
d) Buffering write cycles on the Local Bus while the system bus is
executing other cycles.
o Write protected areas are supported by the SWP# input. This enables
caching of ROM space or shadowed ROM space.
o No Post Input (NPI#) provided for disabling of write buffers per
cycle. This option supports memory mapped I/O designs.
o A20M# input provided for emulation of 8086 address wrap-around.
o SRAM test mode. in which the TAGRAM and the cache RAM are treated as
standard SRAM, is provided. A Tristate Output test mode is also pro-
vided for system debugging. in this mode the 82395DX is isolated
from the other devices in the board by floating all its outputs.
o Single chip, 196 lead PQFP package, 1 micron CHMOS-lV technology.
***Versions:...
***Features:...
**82395SX Smart Cache 12/17/90...
**82396SX Smart Cache 12/17/90...
**82485 Turbo Cache (and 485Turbocache) c90...
**82489DX Advanced Programmable Interrupt Controller 10/12/92...
**82495DX/490DX DX CPU-Cache Chip Set <Sep91
***Notes:...
***Info:
The 50 MHz Intel486 DX CPU-Cache Chip Set provides a high performance
solution for servers and high-end desktop systems. This binary
compatible solution has been optimized to provide 50 MHz, zero wait
state performance. The CPU-Cache chip set combines the 50 MHz Intel486
Microprocessor with the 82495DX/82490DX cache subsystem. It delivers
integer performance of 41 V1.1 Dhrystone MlPs and a SPEC integer
rating of 27.9. The cache subsystem features the 82495DX Cache
Controller and the 82490DX Dual Ported Data RAM. Dual ported buffers
and registers of the 82490DX allow the 82495DX Cache Controller to
concurrently handle CPU bus, memory bus, and internal cache operations
for maximum performance.
The CPU-Cache Chip Set offers many features that are ideal for multi-
processor based systems. The Write-Back feature provides efficient
memory bus utilization by reducing bus traffic through eliminating
unnecessary writes to main memory. The CPU-Cache chip set also
supports MESI protocol and monitors the memory bus to guarantee cache
coherency.
The 50 MHz Intel486 DX CPU and 82495DX/82490DX Cache subsystem are
produced on Intel's latest CHMOS V process which features submicron
technology and triple layer metal.
3.0 ARCHITECTURAL OVERVIEW
3.1 Introduction
The Intel486 CPU-cache chip set provides a tightly coupled processing
engine based on the Intel486 microprocessor and a cache subsystem
comprised of the 82495DX cache controller and multiple 82490DX cache
components. Figure 3.1 [see datasheet] diagrams the basic config-
uration.
The cache subsystem provides a gateway between the CPU and the memory
bus. All CPU accesses that can be serviced locally are transparent to
the memory bus and serve to avoid bus traffic. As a result, the cache
chip set reduces memory bus bandwidth to both increase Intel486
processor performance and support efficient multiprocessor systems.
The cache subsystem also decouples the CPU from the memory bus to
provide zero-wait-state operation at high clock frequencies while
allowing relatively slow and inexpensive memories.
The CPU-cache chip set prevents latency and bandwidth bottlenecks
across a variety of uniprocessor and multiprocessor designs. The
processor’s on-chip cache supports a very wide CPU data bus and
high-speed data movement. The second-level cache greatly extends the
capabilities of the on-chip cache resources, enabling a larger portion
of memory cycles to be satisfied independently of the memory bus.
3.2 CPU-Cache Chip Set Description
The chip set is comprised of three functional blocks:
3.2.1 CPU
The chip set includes a special version of the Intel486DX micropro-
cessor at 50 MHz. The Intel486DX Microprocessor Data Sheet provides
complete component specifications.
3.2.2 CACHE CONTROLLER
The 82495DX cache controller is the main control element for the chip
set. providing tags and line states. and determining cache hits and
misses. The 82495DX executes all CPU bus requests and coordinates all
main memory accesses with the memory bus controller (MBC).
The 82495DX controls the data paths of the 82490DX cache components
for cache hits and misses and furnishes the CPU with needed data. The
controller dynamically adds wait states as needed using the most
recently used (MRU) prediction algorithm.
The 82495DX also performs memory bus snoop operations in shared memory
systems and drives the cycle address and other attributes during
memory bus accesses. Figure 3.2 [see datasheet] diagrams the 82495DX.
3.2.3 CACHE SRAM
Multiple 82490DX cache components provide the cache SRAM and data
path. Each component includes the latches, muxes and logic needed to
work in lock step with the 82495DX to efficiently serve both hit and
miss accesses. The 82490DX components take full advantage of VLSI
silicon flexibility to exceed the capabilities of discrete
implementations. The 82490DX components support zero-wait-state hit
accesses and concurrent CPU and memory bus accesses, and they
replicate MRU bits for autonomous way prediction. During memory bus
cycles. the 82490DX components act as a gateway between CPU and memory
buses. Figure 3.3 [see datasheet] diagrams an 82490DX cache component.
3.3 Secondary Cache Features
The 82495DX cache controller and 82490DX cache components provide a
unified, software transparent secondary data and instruction cache.
The cache enables a highspeed processor core that provides efficient
performance even when paired with a significantly slower memory bus.
The secondary cache interprets CPU bus cycles and can service most
memory read and write cycles without accessing main memory. I/O and
other special cycles are passed directly to the memory bus. The cache
has a dual-port structure that permits concurrent CPU and memory bus
operation.
The 82495DX cache controller contains the 8K tag entries and logic
needed to support a cache as large as 256K. Combinations of between 4
and 9 82490DX cache SRAMs are used to create caches ranging from 128K
to 256K, with or without data parity.
The MBC provides logic needed to interface the CPU, 82495DX and
82490DX to the memory bus. Because the MBC also affects system
performance. its design can be the basis of product differentiation.
***Configurations:
128K cache without parity:
82495DX 50Mhz
+ 4x 82490DX
128K cache with parity:
82495DX 50Mhz
+ 5x 82490DX
256K cache without parity:
82495DX 50Mhz
+ 8x 82490DX
256K cache with parity:
82495DX 50Mhz
+ 9x 82490DX
***Features:...
**82495XP/490XP Cache Controller / Cache RAM (for i860) 06/05/91...
**82496/491 Cache Controller / Cache RAM (for P5 Pentium) 03/22/93...
**82497/492 Cache Controller / Cache RAM (for P54 Pentium) <Nov94...
**82498/493 Cache Controller / Cache RAM (for P54 Pentium) <Nov94...
**
**Later chipsets (basic spec):
**440 series:...
**450NX (?) 06/29/98:...
**????? (Profusion) c:99...
**800 series...
*Headland/G2...
*HMC (Hulon Microelectronics)...
*Logicstar...
*Motorola...
*OPTi...
*PC CHIPS/Amptron/Atrend/ECS/Elpina/etc...
*SIS...
*Symphony...
*TI (Texas Instruments)...
*UMC...
*Unresearched:...
*VIA...
*VLSI...
*Western Digital...
*Winbond...
*ZyMOS...
*General Sources:...
(c) Copyright mR_Slugs Warehouse - All rights Reserved