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: Estephen who wrote (126932)	10/28/2000 11:01:11 AM
From: Jim McMannis	Respond to of 1580833

RE:"My mistake, I realized on my drive home the other night that I meant 75% of workstations." You're old enough to drive? <G> Jim

To: Estephen who wrote (126932)	10/28/2000 1:00:52 PM
From: Estephen	Read Replies (1) \| Respond to of 1580833

Thoughts on the DDR Launch by: h0db (40/M/Tysons Corner, VA) 10/28/00 10:58 am Msg: 179535 of 179579 Okay, on the AMD760 launch--this is 1/10 real. They will announce the chipset, and you will see an EXPLOSION of AMD760 benchmarks around the web. Tom's hardware is going to more than make up for his recent silence on DDR and Rambus. The NDA for the AMD760 expires midnight EST on 30 October. These benchmarks are going to "prove" that DDR provides a 10% performance boost in typical applications and benchmarks. How can I be so sure that it will be 10%. Simple--and it has nothing to do with DDR. You are going to see systems with a double-clocked 133MHz (266MHz DDR) frontside bus. Current systems based on the Athlon and Duron AMD processors use a 100MHz (200MHz DDR) frontside bus. In other words, we are going to see that increasing the FSB by 33% provides a real world performance improvement of about 10%. This is EXACTLY what we saw when Intel release the i820 and later the i815, which use a 133MHz FSB. Why isn't the improvement a flat 33%? Because FSB bandwidth is only part of the problem with current platforms. You still have peripheral bus latencies. So get ready for the BS to hit the fan. Everybody and their brother are going to be screaming about how great DDR is, and they are going to flood this board with benchmark "proof." The only things that the benchmarks prove is something that Intel proved a year ago--if you raise the frontside bus by 33%, you will see a real improvement in performance. It has NOTHING to do with DDR. Pure memory bandwidth is not the problem. The only thing that will shut up the AMDroids is the Pentium-4, and some courts in Germany and the US. Pentium-4 will be the first system where the theoretical memory bandwidth provided by Rambus or DDR is actually utilized. This is because of the Pentium-4's headroom (that is a 2GHz chip, air-cooled, at .18-mu-fab process die) and because of the 400MHz memory interface tied to a 400/800MHz memory subsystem. How much difference with this make in benchmarks? If you look at a pure memory throughput benchmarks, the Pentium-4 @ 1.5GHz will have 3-4 times the memory throughput of a 1.2GHz Athlon/PC2100 DDR system. It won't even be close. The Athlon will pull close on integer applications, but remember this--the Athlon and even the Mustang cores will never be as close to Pentium-4 performance as they are on the day the Pentium-4 launches. From the time the first pentium-4 hits the streets, each new iteration and improvement will only see it pulling farther and farther away from the AMD competition. messages.yahoo.com