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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: kapkan4u who wrote (62381)	6/19/1999 12:08:00 AM
From: Process Boy	Read Replies (2) \| Respond to of 1572777

kapkan - <Ali...Could this mean a process problem?> I seriously doubt Ali has ever worked in IC Process Development, or any concern related to it. Why would you ask him? PB

To: kapkan4u who wrote (62381)	6/19/1999 12:14:00 AM
From: Ali Chen	Read Replies (1) \| Respond to of 1572777

<CuMine is testing fine at 600 MHz at room temperature, but degrades with higher temperatures. Could this mean a process problem?> Could be anything, too little information. Actually, if a 0.25u design can routinely run at 600MHz at room temperature, those 600MHz for 0.18 is not an achievement at all. It is called "sux". Speaking theoretically, the shrink to next feature size is supposed to double the speed of design, so we should expect CuMine at about 666Mhz at start (since 0.25 P-II were introduced at 333). BTW, in Intel roadmaps the CuM was listed at 533+, so officially they are on tracks. I guess the interconnect takes it's toll. It was predicted at a time when Intel's officials proclaimed that copper in not necessary for 0.18. Time to pay for missing "inflection point". - Ali

To: kapkan4u who wrote (62381)	6/19/1999 12:42:00 AM
From: kash johal	Read Replies (3) \| Respond to of 1572777

kap, Re:" comment on Paul's possible slip of the tong that CuMine is testing fine at 600 MHz at room temperature, but degrades with higher temperatures. Could this mean a process problem?" I can see several scenarios: 1. The designers didn't model the routing paths properly. This could require chip relayout and rerouting. 2. Xtalk Noise due to coupling - which degrades as temp increases and also voltage increases (might explain why "jacking up the voltage" was not option" 3. He's just saying parts work fine at RT. You still need to guardband for temp and voltage so you need 20% margins. I am sure there are many more. I doubt the problem is "process" related as I am sure they understand their transistors very well and ran SRAM test chips etc. The solution may be to brute force the process another 10-15% as their isn't time to completely redo the design coupled with limited metal fixes. Still november looks like a tough deal, but I imagine the best and brightest will be working on this one. Regards, KAsh