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 : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: combjelly who wrote (162482)	3/18/2002 9:54:59 PM
From: wanna_bmw	Read Replies (2) \| Respond to of 186894

Combjelly, Re: "it is very likely that a SledgeHammer in 90nm would be closer to 100mm^2 than 150mm^2..." Maybe not. My WAG for Sledghammer die size is ~130mm^2. I think the 512KB of L2 will add about 15mm^2 to the die, and the other two HT links will add 5mm^2 each. Given that the Clawhammer is already 104mm^2, that would put Sledgehammer at about 130mm^2. I base this on the projection for the Thoroughbred die size, and the difference between it and Clawhammer. An 80mm^2 TBred subtracted from a 104mm^2 Clawhammer leaves 24mm^2. I believe it adds up like this: 4mm^2 for x86-64, 8mm^2 for 256KB extra cache, 8mm^2 for on-die memory controller, and 4mm^2 for micro-architectural changes such as deeper buffers, larger BHT, TLB, OOO queues, etc. wbmw

To: combjelly who wrote (162482)	3/19/2002 1:10:35 AM
From: dale_laroy	Read Replies (1) \| Respond to of 186894

>Seems to be on the large side. AMD is estimating the 90nm ClawHammer will be 60mm^2. Now we know that 192k of L2 cache is around 25mm^2 on 180nm. A straight optical shrink to 90nm would yield 768k in the same area. Ok, that wouldn't really work, but it is very likely that a SledgeHammer in 90nm would be closer to 100mm^2 than 150mm^2< Morgan is 106mm2 versus 128mm2 for Palomino. This is 22mm2 for 192KB L2 cache. AMD estimates 90nm the die size of Clawhammer at 64mm2. Add 768KB of L2 cache at the same size as 768KB L2 cache at 180nm and this increases to 86mm2. The circuitry for the extra memory controller and extra HTT ports(s) would increase the die size even more. An optical shrink from 180nm to 130nm would result in a die size 169/324 times as large, which would be a shrink of about 48%, the shrink from 130nm to 90nm would be 81/169, which would be about 52%. Thus, with Palomino being 60% larger than Thoroughbred, we could expect the 130nm Sledgehammer to be about 73% larger than the 90nm Sledgehammer. Just assuming 86mm2 for the 90nm Sledgehammer would therefore imply about a 148mm2 130nm Sledgehammer. BTW, just as I suspected that Palomino was actually larger than Thunderbird because of AMD avoiding all reference to die size, I am now beginning to suspect that Clawhammer will have only 256KB of L2 cache because AMD freely talks about the amount of L2 cache on Sledgehammer, but has yet to disclose the amount of L2 cache on Clawhammer. That is why I base my calculations on Sledgehammer having 768KB more L2 cache than Clawhammer. If Clawhammer really does have 512KB L2 cache, 130nm Sledgehammer will probably come in closer to 137mm2.