AM486DX2 datasheet(28/67 Pages) AMD

28 / 67 page

Am5X86 Microprocessor

AMD

PRELIMINARY

Step 1 During a snoop to the on-chip cache that hits a

modified cache line, the HOLD signal cannot

be deasserted to the microprocessor until the

next clock cycle after HITM transitions active.

Step 2 After the write-back has commenced, the HOLD

signal should be asserted no earlier than the

next clock cycle after ADS goes active, and no

later than in the final BRDY of the last write.

Asserting HOLD later than the final BRDY may

allow the microprocessor to permit a pending

access to begin.

Step 3 If RDY is returned instead of BRDY during a

write-back, the HOLD signal can be reasserted

at any time starting one clock after ADS goes

active in the first transfer up to the final transfer

when RDY is asserted. Asserting RDY instead

of BRDY will not break the write-back cycle if

HOLD is asserted. The processor ignores

HOLD until the final write cycle of the write-back.

4.8.5.2

AHOLD Bus Arbitration Implementation

The use of AHOLD as the control mechanism is often

found in systems where an external second-level cache

is closely coupled to the microprocessor. This tight cou-

pling allows the microprocessor to operate with the least

amount of stalling from external snooping of the on-chip

cache. Additionally, snooping of the cache can be per-

formed concurrently with an access by the microproces-

sor. This feature further improves the performance of

the total system (see Figure 11).

Note: To maintain proper system timing, the AHOLD

signal must remain active for one clock cycle after HITM

transitions active. Deassertion of AHOLD in the same

clock cycle as HITM assertion may lead to unpredictable

processor behavior.

DRAM

Address Bus

Data Bus

L2 Cache

Address Bus

Data Bus

I/O Bus

Interface

Slow

Peripheral

CPU

Address Bus

Data Bus

Figure 11. Closely Coupled Cache Block Diagram

The following sections describe the snooping scenarios

for the AHOLD implementation.

4.8.5.3

Normal Write-Back

Scenario: This scenario assumes that a processor-ini-

tiated access has already started and that the external

logic can finish that access even without the address

being applied after the first clock cycle. Therefore, a

snooping access with AHOLD can be done in parallel.

In this case, the processor-initiated access is finished

first, then the write-back is executed (see Figure 12).

The sequence is as follows:

Step 1 The processor initiates an external, simple,

non-cacheable read access, strobing ADS = 0

and W/R = 0. The address is driven from the

CPU.

Step 2 In the same cycle, AHOLD is asserted to indi-

cate the start of snooping. The address bus

floats and becomes an input in the next clock

cycle.

Step 3 During the next clock cycles, the BRDY or RDY

signal is not strobed Low. Therefore, the pro-

cessor-initiated access is not finished.

Step 4 Two clock cycles after AHOLD is asserted, the

EADS signal is activated to start an actual

snooping cycle, and INV is valid. If INV is 0, a

read access caused the snooping cycle. If INV

is 1, a write access caused the snooping cycle.

Additional EADS are ignored due to the hit of a

modified line. It is detected after HITM goes in-

active.

Step 5 Two clock cycles after EADS is asserted, the

snooping signal HITM becomes valid. The line

is modified; therefore, HITM is 0.

Step 6 In this cycle, the processor-initiated access is

finished.

Step 7 Two clock cycles after the end of the processor-

initiated access, the cache immediately starts

writing back the modified line. This is indicated

by ADS = 0 and W/R = 1. Note that AHOLD is

still active and the address bus is still an input.

However, the write-back access can be execut-

ed without any address. This is because the

corresponding address must have been on the

bus when EADS was strobed. Therefore, in the

case of the core system logic, the address for

the write-back must be latched with EADS to

be available later. This is required only if

AHOLD is not removed if HITM becomes 0.

Otherwise, the address of the write-back is put

onto the address bus by the microprocessor.

AM486DX2 数据表(PDF) 28 Page - Advanced Micro Devices

AM486DX2 数据表(HTML) 28 Page - Advanced Micro Devices

Html Pages

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67

数据表下载

链接网址

AM486DX2 数据表(PDF) 28 Page - Advanced Micro Devices

AM486DX2 数据表(HTML) 28 Page - Advanced Micro Devices

Html Pages

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67

数据表 下载

链接网址

数据表下载