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: milo_morai who wrote (48250)	7/18/2001 5:17:12 PM
From: pgerassi	Read Replies (1) \| Respond to of 275872

Dear Milo: Given that P4, Tbird, and MP are all thermally limited (cooling allows them to go faster) not speed path limited, anything that evens the temperature out or gets rid of heat faster can only help output. In the case of P4, it might be more helpful as there are two known hot spots on it. Low K dielectric, smaller voltage and smaller sizes probably will be more help but, every little bit counts. Ison-28 will probably add no more than 5% to speed of AMD's CPUs. Low K will add better than linearly to speed as lower K means lower capacities and lower currents means reductions are squared for thermal purposes (the limits due to speed increases in slew rates will rise slower than the limits by thermal output but, were higher to begin with. Imagine that CPUs at room temperature run at same speed as them being cooled to -40C. Thus, a 30% to 40% improvement is not out of the question. Votage reductions help in thermal output by reducing both components current and voltage. It does nothing for speed path limits except by reductions in feature sizes. It has the same characteristics as above but with lower ultimate limits. It also increases power lost in the voltage regulators relative to CPU thermal power. This is forgotten by many when sizing power supplies. A 70 watt CPU also causes the regulators to lose between 20 and 30 watts. Tyan was correct to require increases of greater than 100W for the second CPU. Regulator power losses increase by a factor with linear and quadratic terms. To supply 2x the current loses more than 2x in losses. Going from 300W to 460W adds to the margin and helps stability. The last, smaller sizes, reduces current but, without the decrease in voltage, is generally a wash in terms of thermal speed. It also increases thermal power density. It does, however, increase limits by speed paths linearly. These last two probably account for between 10 and 15% from 0.18 to 0.13u. Copper has another 30% at 0.18u or 20% at 0.13u. So, you can see why AMD is looking at SOI, low K, and even ISON-28 for 0.13u. The potential gains are probably a 100% increase to almost 3 GHz if all are implemented together. Pete