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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Kevin K. Spurway who wrote (39309)	10/14/1998 4:56:00 PM
From: Tenchusatsu	Read Replies (2) \| Respond to of 1573718

Kevin, re: <Don't leave us hanging.... What does it [instruction trace cache] do?> Whoops, sorry. I was talking mainly to Ali, who I assume knows more about the thing than I do. As I understood it, the ITC will store decoded instructions so that the next time the instructions are needed, in loops for example, the x86 instructions wouldn't have to be decoded again. I'm not sure what the exact benefits are. I'll try to find out more about it. Tenchusatsu

To: Kevin K. Spurway who wrote (39309)	10/14/1998 6:59:00 PM
From: Ali Chen	Read Replies (1) \| Respond to of 1573718

Kevin, <"instruction trace cache." What does it do?> Instruction trace cache is a variant of branch prediction algorithm, where a processor caches not only a few static sequences of branch instructions (to train the branch predicator), but stores a whole trace of recent instructions, identifies code pattern ("basic block"), and predict taken/not-taken branches for the whole "basic block". For more readings, visit drona.csa.iisc.ernet.in and look through list of publications. IMHO, the implementation of this idea will cost a lot of silicon, with vague return in performance. Much better results may be achieved by improving compiler techniques. -Ali