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: kapkan4u who wrote (27083)2/1/2001 1:57:48 AM
From: milo_moraiRead Replies (1) | Respond to of 275872
 
Well it's only on one benchmark, but I'm sure it should be similar gains as in EV6 Alpha systems going from one to two processors.

Milo.



To: kapkan4u who wrote (27083)2/1/2001 2:06:01 AM
From: fyodor_Respond to of 275872
 
Kap: I am somewhat dubious about the > 100% performance increase.

uhmm.. There's something I'm not quite clear about... Were both runs done on the same system? If so, I'm told that performance even on a single processor system can be increased nicely by running parts of the compile in parallel. Who knows the specific conditions under which this test was conducted? They certainly weren't given in the accompanying text.

Now that the Tyan board has been spotted several times, I'm sure we'll see more benchmarks - and from more reputable sources.

-fyo



To: kapkan4u who wrote (27083)2/1/2001 11:52:26 AM
From: Ali ChenRespond to of 275872
 
Kap, "somewhat dubious about the > 100% performance increase."

It is theoretically possible. Two threads with
non-overlapping data sets would obviously do
better on two processors (each one with it's own
cache) than on a single processor: no cache
trashing during process switching. The effect
can be huge if the data sets are about the size
of L2 cache.

Regards,
- Ali