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.
Politics : Formerly About Advanced Micro Devices

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Steve Porter who wrote (66382)7/21/1999 10:09:00 PM
From: kapkan4u  Read Replies (1) of 1575622
 
<Prefetch doesn't return in time.. the computation in the algorithms core are done before the first prefetch has been completed. The message boards at 3dnow.org include this information.>

I looked for prefetch discussions in 3dnow.org forums but could not find any. Without seeing the code it is hard for me to be precise in suggesting how to use prefetch. In general you have to unroll the loop and prefetch several iterations back to allow for prefetch to complete. In addition try using streaming stores (to bypass the cache when writing data). If you give me a pointer to the SETI sources I can try to optimize the inner loop code for you using the SSE streaming instructions. The K7 optimizations will have to wait until I get my killer K7 box.

Kap.

Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext