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.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: fyodor_ who wrote (74844)	3/18/2002 1:46:11 PM
From: AK2004	Read Replies (2) \| Respond to of 275872

fyodor re: Joe sorry, I need some coffee :-)) re: That's a 20% difference? That point did not escape me and it does not add to my comfort :-)) but my point was that given that dP/df "jumps" even within one processor type you might need a bigger sample to get to anything that might be conclusive. Again, I looked at it as samples without regard for the business so I obviously ignored anything that cpu specific - non my area of expertise. re: I chose 7 benchmarks fairly arbitrarily you were completely fair about it, what I referred is actual frequency points. Even if there is such a thing as "true" dP/df can you really say that the results are conclusive given that in a small set you encountered at least 2 ouliers I know that std is not really at issue here but it needed to establish degree of confidence re: it does show that AMD needs to either change its quantispeed formula since model numbers are compared to intel frequency then I agree with you. But I recall that at the QS intro someone at AMD said that they would use simplified formula over some rage of frequencies and adjust the formula when needed so, at this point, I do not see amd disagreeing with you on that either. Should they change it right now or let it go for another few speed grades would make very little difference Regards -Albert

To: fyodor_ who wrote (74844)	3/18/2002 9:07:55 PM
From: Petz	Read Replies (2) \| Respond to of 275872

fy., For AthlonXP, dP/dQf is around the 4300 mark. For Northwood, dP/df is around the 4900 mark. That's a 20% difference… in Northwoods favor. Ahem... 4900 is less than 15% above 4300, not "more than 20%!" It's totally bogus to compare dP/dQf for Athlon to dP/dF for P4 and then claim that AMD's IPC needs improvement. The AMD quantispeed ratings are not governed by any formula that says that 66IntelMhz=100AMDMhz. OTOH, it would be silly for AMD to have model ratings like "2049" instead of "2000+" or "1968" instead of "1900+." AMD gave themselves a lot of slack with the 1600+, which clearly could have been labeled an 1800+. In short the "+" means something; it means "faster than xMHz Intel," but 2000+ does NOT mean "100 MHz faster than 1900+," it just means, "faster than Intel 1900 MHz. But with Northwood vs. Palomino, "1.5 qF per F" is over. I'm pretty confident that there will be some archtectural improvements in TBred which give it 10% additional IPC, so much the more if they increase the bus speed. This should allow AMD to continue to use round numbers for their model ratings. Petz