Silicon Investor (SI) -- The First Internet Community

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) -- Ignore unavailable to you. Want to Upgrade?

To: Elmer who wrote (56660)	10/2/2001 1:31:44 AM
From: Joe NYC	Respond to of 275872

Elmer, As far as the P4, how many times do people have to tell you that SSE2 code is new and that requires a recompile. Get it through your head. I think SSE2 has a great potential in computer marchitecture, since it can help in application that get benchmarked, or outright benchmarks. In the real world performance, it is practically worthless. Let me preface what I am going to say that I am aware of the fact that what matters is perception, not reality, so what I am going to say is irrelevant. In real world business apps (I have written way too many of these), there is basically no FP code. The occasional FP instruction here and there that does end up in a commercial app makes no difference in performance. You could go to the extreme and disable FP unit, ran FP emulation, you would have hard time measuring the difference in performance. A scenario that SSE2 needs is a series of densely packed FP instructions, and it just doesn't exist in business software. (I know you can find exception, but I am talking about the rule) A good case in point was a benchmark I have seen long time ago, in which a Cyrix chips at lower clock speed (equal PR rating - I believe 233), with it's lousy FP unit beat PII in an MSFT Excel tests of recalculating complex spreadsheet, which is as close as you will ever get to FP instruction use in a commercial application. The density of the FP instructions (in this rare app that uses FP instructions) was so low that it didn't do anything to help PII. Anyway, this is just a side note. I am perfectly aware that in the perception is reality world, this is irrelevant. What counts in this world is how computers perform on benchmarks, and as long as the 95% of people's use of computers get bunched into 1 benchmark (where SSE{2}) is irrelevant, and remaining 5% of the apps - games, graphics, multimedia {where SSE{2} may help) end up in other 19 benchmarks, Intel has some upside from the recompilation. Joe