Quote:
Originally posted by westothemax
I knnow you said a general explanation and I don't mean to threadjack but...
I was under the impression that too large of a cache will increase miss rates and penalties, and there is a point where the benefit from a larger cache is outweighed by the penalites from increased misses - so by adding more capacity to your cache you will eventually peak out your performance until the complexity of finding stuff in the cache starts sending your perfomance down the tubes.
There's something else that makes AMD's architecture use clock cycles more efficiently, but I'm not sure what that is.
Intel uses simultaneous multithreading, which their marketing dept calls "hyper-threading".
I am no expert on computer architecture by any means and if I am wrong please correct me.
|
Yeah, you have a point! I've always heard that L3 cache is the level that can decrease performance if too much is added. Take a look at the P4 Prescotts! They're not doing too well, and some believe it's because of the added L3 cache. I could be wrong, but I have heard of what you're talking about.
-Lasereth