Advanced Micro Devices - Moderated (AMD) Message Board

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD)

AMD 204.52

-2.6%

3:40 PM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: jcholewa who wrote (60018)	10/24/2001 10:26:43 PM
From: Elmer	Read Replies (1) of 275872

Why must we conclude this? It's been established many times that specfp2000 scales extremely nicely with average memory bandwidth. Could it not simply be that Intel's 3.20GB/s peak memory bandwidth and hardware prefetching allow it to attain these high scores? I think the data supports it. Consider this: www6.tomshardware.com Here we have an application, Flask, which is benchmarked in several compiled versions, old x87, new x87 and SSE2. A striking point to note is that we see a 3.66-X speedup with a simple Intel compiler x87 recompile alone! Apparently there are some really bad compilers out there producing x87 code and perhaps benchmarks based on "old code" are not worth much, the Intel compiler has been attacked frequently but it even produces code they runs very on Athlon, but I digress.... If we compare the x87 optimized code with the SSE2 code we still see an additional 35% improvement, not the 5% that you seemed to suggest. Additionally Intel engineers said they could have done a lot more optimizing but time did not permit and this seemed to be an old version of the Intel compiler, ver 5.1. I can't find any other benchmarks right now that use SSE2 and also compare them to x87 code compiler by a good compiler (Intel). Perhaps you have other examples? Now I ask, did you mean to say that SSE2 only provided an approx 5% speedup over x87 code overall, or were you specifically referring to the code used to measure SPECfp_2000? EP

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next