SI
SI
discoversearch

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: Petz who wrote (19106)11/14/2000 8:06:34 PM
From: fyodor_Read Replies (3) | Respond to of 275872
 
<Petz: Programming double precision SPECfp stuff using...>

I didn't know there was any double precision fp stuff in SPECfp95/2k. Are you sure there is?

And might I ask where you got the 30% increase in SPECfp using SSE2? If accurate, this might be due to some indirect improvements, such as prefetching.

-fyo



To: Petz who wrote (19106)11/14/2000 8:52:42 PM
From: Scott D.Read Replies (1) | Respond to of 275872
 
re: unavailable Intel compiler

I have the beta, which will be available soon:
developer.intel.com

It certainly supports all Intel x86 processor features through Pentium 4, and generates great code.

It generates great Athlon code, too!



To: Petz who wrote (19106)11/15/2000 11:36:26 AM
From: jcholewaRespond to of 275872
 
> rogramming double precision SPECfp stuff using SSE2's many new registers
> is a lot more efficient than messing around with the x86 stack.

Assuming a free FXCH (you know, the instruction that gives a stack the flexibility of a 2-operand flat register set, other than the increased decode bandwidth requirements, do you see any serious penalties that the x87 stack has compared to SSE2?

&nbsp;&nbsp;&nbsp;&nbsp;-JC