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)
AMD 256.12+0.5%Oct 31 9:30 AM EST
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Scumbria who wrote (33454)3/25/2001 11:36:16 PM
From: dale_laroy of 275872
 
"Generally I prefer the concept of sophisticated compiler driven prefetching vs. dumb hardware prefetching."

Hardware prefetching need not be dumb. AMD has a patent on predictive cache prefetching that uses historical patterns to predict when prefetch will be useful, in much the same way that the Branch History Table is used to predict branches.

Even in the case of dumb hardware prefetch, cache flushing need not be much of a problem. As long as the prefetch only stays one cache line ahead of execution, it would have less impact than doubling the cache line size.

Personally, I would rather see AMD implement prefetch directly to L1, rather than to L2.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  

HomeMailHotSubjectMarksPeopleMarksKeepersSettings
Terms Of UseContact UsCopyright/IP PolicyPrivacy PolicyAbout UsFAQAdvertise on SI
© 2025 Knight Sac Media.  Data provided by Twelve Data, Alpha Vantage, and CityFALCON News