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 215.65+0.3%Dec 29 3:59 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: eracer who wrote (151555)2/23/2005 4:25:08 PM
From: PetzRead Replies (1) of 275872
 
The Anandtech review was limited to single channel 355MHz DDR. Dual channel DDR-667 will provide 3.75 times as much bandwidth as the Dothan had in that review.

But with an FSB of 666, the most bandwidth the Yonah can use is only 1.88x larger than what Anand (and all the other reviewers) could accomplish. But for 2 cores, it wants 2x the bandwidth and it won't get it.

Perhaps you should look at x86-secrets benchmarks of the Sonama platform, which runs at 533 FSB, 80% of what Yonah will have (100% of the server version of Yonah, FSBs can't be as fast when paralleled). The Sonoma platform has a 2.13 GHz CPU and uses dual channel DDR2 memory.
translate.google.com
Benchmarks are on previous page,
babelfish.altavista.digital.com

They found hardly any improvement at all and they were comparing a
2.13 GHz Model 770, 533 FSB, dual channel DDR2 to
2.1 GHz Model 765, 400 FSB, single channel DDR333

Results: PovRay: 0%, Kribibench +12%, DivX +4%, RAR 0%.

From conclusions: Indeed, if the performances of Pentium M are more and more recognized and were already excellent on Centrino first generation, they benefit only little from the passage to a bus 533. Side report, Dual GDR-II 533 comes to take the changing of the growing old DDR333. Malheureusement, the flow offered by this memory being twice higher to the maximum which Pentium M can use, the surplus could be used only to the graphic controller integrated and, some share, to use memory very expensive and powerful in a portable only equipped with a chipset with graphic controller integrated, it is a waste that few manufacturers can allow. With final, vis-a-vis the old generation, Sonoma can pride itself on a rise of the flow report [memory bandwidth] from 15 to 20% on synthetic tests. Alas, this profit is perceived only little in practice since our Sonoma platform appeared only very slightly more powerful than the preceding one. A negligible profit which should not however not occult the benefits of this new platform.

Indeed, if, just like for the passage of the i865 to the i915, the rise of performances is negligible compared to the preceding generation...
but more features, blah, blah, blah

Convinced yet? I was shocked how little improvement the Sonoma platform and higher FSB gave. I think that is because the higher latency of DDR2 hurts a shorter pipeline architecture like Pentium M proportionately more than the Netburst architecture.

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