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: Petz who wrote (6451)	8/24/2000 2:07:33 PM
From: Pravin Kamdar	Respond to of 275872

Petz, It looks like the Dresden process gives roughly a 100 Mhz speed bump for the same power as compared to the Austin process. Still, at 50+ watts, AMD's designs and processes use way too much power as compared to Intel's. AMD needs SOI. Pravin.

To: Petz who wrote (6451)	8/24/2000 3:42:49 PM
From: THE WATSONYOUTH	Read Replies (2) \| Respond to of 275872

At Intel things are actually the reverse. The 1.13 GHz PIII requires a huge jump in current and a 0.1 volt increase in voltage. It appears that Intel can not further increase freq. by channel length reduction alone (constant Vcc/constant Tj/ constant Tox/constant groundrules). Given a choice, it is preferable to reduce channel length rather than raising Vcc,reducing junction temperature, or shrinking groundrules. We know Intel has done the last three. By optimizing the device design to be able to run at shorter channel lengths, you can increase freq. with only modest increases in power since the reduced gate capacitance will partially offset the power increase due to higher freq. I don't believe Intel has made any further optimization of its device design since the notch process was introduced. I predicted the original .18um process as described in Dec 98 IEDM would max out at a 800MHz shippable bin (900MHz with guardband) Since then, Intel has improved their device design with the notch process (10%) / increased voltage from 1.5V to 1.8V (20%) / did an approx 15% process (same device) shrink (6%) / and reduced Tj 25 degrees (6%). So, 900MHz1.11.21.061.06 = 1335MHz which with 12% guardband is a 1168MHZ shippable part. I don't think they can increase Vcc any more. (they could not have gone to 1.8V without the lower Tj) They also can not lower Tj further. They can further shrink the process but I suspect they will not until the completely new .13um device and process. Further device optimization to shorter channel lengths is possible but they are probably at the minimum channel length in terms of (Vtlin/Vtsat/Ioff/noise margin/etc.) all ready. Even a 75A reduction in minimum channel length would add about 8%-9% freq improvement or about 100MHz. The last jump to 1.13 GHz seems to be from the process shrink and increased Vcc alone. So, 1.2GHz is possible at a very low volume or at larger volumes with an additional process shrink. But not very likely. all speculation THE WATSONYOUTH