Why AMD's L2 Cache only has 2MB maximum?

[Mic_Ma]

I don't know the answer to your question (I guess something to do with economics) but here is an interesting link comparing cache sizes. They really do seem to make a difference:

http://www.tomshardware.com/2007/10/24/does_cache_size_matter/index.html

[lots]

AMD's K8 and related chips can get away with smaller cache sizes because the design is less dependant on them. Since the K8 (Athlon64 and Opteron) have a direct connection to system RAM that is both high bandwidth and low latency, the cache size does not play a big role in overall performance (at least as big a role as it does in Intel chips anyway :P). Instead you see larger gains on the AMD platform if you have RAM with really low timings. AMDs chips just don't need the cache as badly as Intel's.

With Intel's platform, you have a northbridge chip connected to the RAM which is then connected to the CPU. Because of this extra hop, there is a larger latency, and to make up for this, Intel must load its CPUs up with alot of cache space. This is OK though, since Intel has spend alot of money developing inexpencive and fast L2 cache.

In either case, it is mainly because the two companies approach CPU design from different angles. If you look at benchmarks of AMD chips with small L2, and compare to ones with large L2, you really don't notice much of a gain. AMD's memory sub system is much more advanced than Intels. Even today. It probably wont be true for much longer though, as Intel is developing a similar tech to AMD.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.