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: DRBES who wrote (79386)5/5/2002 11:43:23 AM
From: Ali ChenRead Replies (1) | Respond to of 275872
 
DRBES: "How do you intelligently share a single FPU that two or more applications are waiting for."

Your premise is incorrect. Modern applications do not
wait for FPU/SSE, in 50%-80% time they are waiting for memory
to deliver data. The whole idea is based on this behavior:
while one thread is waiting or data, another thread is
using the same execution units that otherwise would idle.
Given the ever increasing gap between CPU speed and memory,
this is indeed a very smart idea.

Of course, efficient use of the multithreading capabilities
would require thorough re-thinking and re-engineering
of all individual applications. The simplest way is to load
the system with several independent and loosely-related
tasks, which what Intel is already doing by steering
and re-specifying benchmarks.

Regards,

- Ali



To: DRBES who wrote (79386)5/5/2002 1:09:24 PM
From: Monica DetwilerRead Replies (2) | Respond to of 275872
 
SOME THOUGHTS ON AMD"S INABILITY TO DELIVER ANYTHING MORE THAN 2-WAY SMP SYSTEMS
If AMD wanted to exploit this technology it already has, in house, far more capable expertise (renegades and escapes from inteL's run at DEC debacle) than inteL is likely "home grow" any time in the reasonable foreseeable future.


Interesting.

Then just how did Intel manage to implement Hyperthreading years before AMD?