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: peter_luc who wrote (73341)	3/5/2002 7:32:51 AM
From: Gopher Broke	Read Replies (3) \| Respond to of 275872

Firstly, Paul DeMone has been so wrong in the past with his predictions that his arguments don't carry a lot of weight with me. Secondly, his estimates are based on the assumption that all the changes in Hammer are evolutionary rather than revolutionary. Therefore it is inevitable that he gets an answer indicating only an evolutionary increase in performance. If I net down what he is saying, it is that Hammer is pretty much the same except that it uses a better process and has an onboard memory controller that helps with memory latency. You have to wonder what the AMD engineers have been doing with their time. I have no evidence that his assumptions are wrong, obviously, but he could have plucked that 30% number out of the air without bothering to look for the envelope. Incidentally, I think that the area he will be most wrong in is the actual clock speed of Hammer. He is predicting mature top bin .13 SOI Hammers clock at 2400 GHz. Just doesn't sound right to me. I would expect Hammers to clock at 2GHz at intro and be well over 3 GHz when fully matured. Look at scaling of the Athlon on the .18 process for a comparison. He also doesn't make allowances for any other architectural enhancements AMD may have made, such as improved branch prediction and prefetch. Oh, and with his blinkered view he seems to have omitted to factor in that SledgeHammer is a dual processor core.

To: peter_luc who wrote (73341)	3/5/2002 9:17:01 AM
From: Charles Gryba	Respond to of 275872

peter, he is a little off. some amd mouthpiece suggested a spec int score of 1350 for the Hammer. Whether that's for the early or mature version I don't know but it's definitely above whatever Intel may have in the next 12 months ( Unless Intel surprises us with another P4 core sooner than Prescott ). C

To: peter_luc who wrote (73341)	3/5/2002 9:59:50 AM
From: Elmer	Respond to of 275872

PT, thread, Hammer performance estimate If this estimate is accurate (big IF I realize) then Hammer will have just about matched P4@3GHz-533 before P4 sees any significant architectural enhancements. P4 is here in high volume now, it is scaling remarkably well, the process is running extremely well and in short it's already proven itself while the Hammer hasn't. P4 will see significant enhancements before Hammer sees meaningful volume. If the best hope for Hammer is to match a 3GHz NorthWood P4 then I think the hype is way overblown. EP

To: peter_luc who wrote (73341)	3/5/2002 10:28:11 AM
From: heatsinker2	Respond to of 275872

Peter- Hammer performance estimate Well I don't particularly accept Paul DeMone's estimates. I think Hammer's performance will be better. Also note that Paul is looking at Hammer strictly as a 32bit CPU. For 64 bit applications, the Hammer will smoke the P4. So even if Paul's estimates are true, it's not such a bad situation for the Hammer: as fast as Intel's CPUs in 32 bit, but much faster for the 64 bit applications and OS. Also cheaper! So even with the bad performance estimate, Hammer looks pretty solid.