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: pgerassi who wrote (70354)	2/4/2002 5:18:08 PM
From: wanna_bmw	Respond to of 275872

Pete, Re: "A 2.33GHz Throughbred (=Palomino) (assuming dumb shrink), would beat a 3.3GHZ NW (if it has 512K cache and a 166MHZ FSB, it would beat a 4GHz NW). A 2.67GHz Barton (=Palomino) would beat a 3.8GHz NW (if it too has a 512K L2 and 166 FSB, a 4.8GHz NW). However, since the NW 2.2GHz really only NCT (non clock throttled) to 2.6GHz at best, it is debatable if NW will reach 3GHz on 0.13u copper. A 3GHz Clawhammer will have about twice the IPC with lower latency to boot would out run a 4.5GHz NW and a 3GHz Sledgehammer would outrun a 10GHz NW (the ultra high multipliers will start taking a big chunk out of the NW's performance)." What are you basing these claims on? Some kind of linear scaling analysis, maybe, or perhaps better projected scaling from TBred or Barton? What are you estimating the added cache and front side bus scaling would add? Can you please break this down into simpler logic? Much appreciated. wbmw

To: pgerassi who wrote (70354)	2/4/2002 5:38:32 PM
From: Joe NYC	Read Replies (2) \| Respond to of 275872

Pete, A 2.33GHz Throughbred (=Palomino) (assuming dumb shrink), would beat a 3.3GHZ NW I think you are being too optimistic. 2.33 GHz is 10 x 66 MHz from 1.66 GHz, which is XP-2000+. So if the formula remains the same, 2.33 GHz would be 3000+. On performance basis, XP-2000+ is somewhere in neighborhood of NW 2 GHz to NW 2.2 GHz. The problem is that Tbred and Barton will have only 256K L2, and it will limit the performance scaling vs. NW with 512K L2. Also, NW will get a 4.2 GB/s of bandwidth soon, and XP is limited to 2.1 GB/s, which will again limit performance scaling. My prediction is that 2.33 GHz Tbred will at best equal NW 3.0 GHz. Being at performance parity with older core (Athlon) vs. Intel's newest (NW) is not such a bad thing. I think AMD should at this point emphasize maturity of design, breadth of supporting chipsets and low power. I think high performance with low power consumption is the ticket for AMD in 2002. AMD has 3 aces in their pocket: PowerNow, .13u, .13u SOI. I hope AMD will use them to the full advantage. Joe

To: pgerassi who wrote (70354)	2/4/2002 5:51:53 PM
From: Yousef	Read Replies (1) \| Respond to of 275872

Dear Pete, Re: " ... Are you falling into that other trap, that MHz equals performance?" Obviously in your "world", Model-Hurtz equals performance. <ggg> What do the technical people know ... Let marketing set us all straight. <ggg> Pretty sad, IMHO. Make It So, Yousef

To: pgerassi who wrote (70354)	2/4/2002 6:31:05 PM
From: dale_laroy	Read Replies (2) \| Respond to of 275872

>I did not see Jerry Sanders ever say that Barton would be slower than NW (link please) although he may have said that Intel so believes, to their horror when it does not pan out (do not confuse clock speed not matching with performance not matching.< Jerry didn't actually state that Barton would be slower than NW. What he stated was that Athlon could not keep pace with P4. Whether or not the fastest Barton is or is not higher performance than the fastest Northwood, there can be little doubt that with the bus speed rising by 33% over Willamette versus a possible 20% increase in bus speed for Barton over Palomino, combined with a doubling of the L2 cache size of Northwood versus Willamette and Barton having the same L2 cache size as Palomino, Northwood will be at the very least a more worthy challenger to Barton's performance crown than Willamette is to Palomino's performance crown. >The former I could believe, but, a 3.3GHz Palomino would beat a 3.8GHz NW more than a 1.67GHz Palomino beat a 2.2GHz NW.< Probably true, but there is a higher probability of NW hitting 3.8 GHz than Palomino(or even Barton) hitting 3.3 GHz. >But if Clawhammer is 50% faster than Barton at same speed and roughly same die size, why not go completely to Clawhammer?< I believe this is the plan. Clawhammer will be aggressively priced. The peak clock rate for Clawhammer should run about 20% higher than the peak clock rate for Barton, and just as AMD charges roughly the same for the highest clock rate Duron as the same clock rate Athlon, AMD will probably charge the same for the highest clock rate Barton as the same clock rate Clawhammer. Then, in mid-2003, AMD will introduce the value segment Hammer, effectively killing off the desktop Barton. Duron will still occupy the value segment, but Athlon will effectively be history by the end of 2003.