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 194.80-2.8%12:01 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: DDB_WO who wrote (227988)3/19/2007 6:10:23 AM
From: RinkRead Replies (3) of 275872
 
AMD's GDC 2007 slide set (with many SSE128 details)
By Dresdenboy (Recent posts)
Monday, March 19, 2007 3:17 AM EST

--------------------------------------------------------------------------------
Slide 24 now clearly tells us, that SSE128 won't make scalar SSE code (and x87 code) faster than it is in K8. So for this kind of code only the other core improvements would apply.

The INQ was unfortunately wrong on this. But there wouldn't have been the bandwidth (decode, L1, L/S) to support a sustained double throughput of scalar SSE code.

Also the L1 still only supports 64 bit stores. Load latency increased by one cycle.

Matrix math will be 85% faster per core. Other math intensive apps gain 10% to 50%.

Etc.. Just read yourself.

I just found the pdf using google looking for something different :)

Link: developer.amd.com
--------------------------------------------------------------------------------

Above post copied from Aces: aceshardware.com

DDB, your find made it to the Inq: uk.theinquirer.net

The Inq article mentions 8-core desktops this year. I'm sure this will be in the form of dual socket Torrenza A64 FX boards though. Still it sounds kinda cool.

Regards,

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