Applied Materials No-Politics Thread (AMAT) Message Board

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 : Applied Materials No-Politics Thread (AMAT)

AMAT 307.20

+2.0%

Jan 12 3:59 PM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: Big Bucks who wrote (7405)	10/4/2003 1:29:25 AM
From: Ira Player	Read Replies (2) of 25522

Three items to think about: 1. Do the current CAE tools support Wafer Scale Integration? My feeling: Not yet. The tools available are barely supporting the move to 90 nm because of issues with timing estimates at that level. Better timing estimates...more processing time...slower iterations. Kind of a chicken and egg issue. You don't have the power until you have massive parallel... 2. How do you remove the heat from this wafer? This is a non-trivial issue. Some modern MPU's would fail in seconds without a heat sink, minutes without a fan. A 12 inch wafer, "fully loaded" with advanced circuitry would dissipate several kilowatts 3. Are there alternatives to this technique? Again, I believe using known good die in a multichip module (MCM) is more cost effective. Processors and Memory use very different techniques. Placing both on a single wafer increases the number of process steps. MCM's (possibly using silicon as the substrate for the chips to deal with thermal mismatch related failures) allows the chips to be made on an optimized wafer process. Stacking can be used to reduce trace lengths within the modules. "Generic" Modules can be made and used in various configurations. Modules are identical rather than the unique configurations that would result from the bypassing of malfunctioning chips on each wafer. While a dense MCM also has heat dissipation issues, each module is identical, so a generic solution works. Each wafer would have a different thermal signature and the cooling system would have to adjust or (more likely) be over engineered to handle the worst case load. (A fully operation, 100% yielded wafer.) Intel flirted with MCM methodology with the "Pentium Pro" but called it a "Dual cavity" package...using a second chip for the cache. Just another opinion regarding Wafer Scale Integration. Will it eventually happen? I believe it will, for some special applications. But not yet. Maybe it's like X-ray Lithography. It's the "technology of tomorrow" and always will be... However, I completely agree with the premise that packaging is going to play a more significant role in electronics than it has in the past. Ira

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next