July 2004
Intermediate to advanced
1744 pages
35h 3m
English
The 486 processor implemented a unified code/data cache that was subject to contention when the instruction Prefetcher and the Execution Unit issued simultaneous requests to the cache.
Starting with the Pentium® processor, all IA32 processors implement separate code and data caches, thus eliminating the contention issue related to a unified cache.
Refer to Figure 20-5 on page 477. The code or instruction cache in the P5 and the P54C versions of the Pentium® was an 8KB, 2-way set-associative, read-only cache. The cache was read-only and only implemented the SI subset of the MESI protocol. Each 4KB cache way contained 128 cache lines of 32 bytes each. There was a separate 128 entry directory ...
Read now
Unlock full access