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: milo_morai who wrote (105911)	4/15/2000 2:22:00 PM
From: Dan3	Respond to of 1575845

Hi Milo, I expect - for no good reason :-) that Willamette will be a competitive chip with good binsplits. But it may not be easy to yield. In the same time frame we will have Mustang which will, at the very least, be competitive, and will probably be easier to manufacture. (again, that's my conclusion from reading the same things that everyone else has read). On the darker side, has anyone else noticed this: ibm.com Featuring "buy it today, we'll ship it today" for a 1GHZ PIII? The fine print below says: Orders subject to product availability. Delivery date and costs will vary depending on delivery method chosen. IBM reserves the right to modify or withdraw this offer at any time without notice. But it looks like they do have a few to sell. Regards, Dan

To: milo_morai who wrote (105911)	4/15/2000 3:22:00 PM
From: Joe NYC	Respond to of 1575845

milo, Who can tell me if 'Willamette' at 1.3 GHz is even going to be faster than 'Coppermine' at 1 GHz? There are enough people who seriously doubt that, including yours truly This will make for an interesting comparison with 1.25+ GHz Thunderbird, not to mention Mustang. Joe

To: milo_morai who wrote (105911)	4/15/2000 5:31:00 PM
From: minnow68	Read Replies (5) \| Respond to of 1575845

Milo, This is one possibility that no one talks about. I am referring to is a situation where Mustang might have a significantly higher IPC than Willy AND a higher clock rate. Not just at introduction, but for all of time. Sledgehammer may actually do nothing more than increase AMD's lead over Intel. Everyone assumes that Willy will clock faster because of the deep pipe. But there is more to this than the deep pipe. The PIII has a deeper pipe than Athlon does already. And somehow Athlon is faster. And that is true despite PIII currently being made on a signficantly better process than Athlon. Scumbria is right about deep pipes allowing higher clock rates in general. The empirical evidence is indeed strong that deeper pipes produce higher clock rates. But it is not a perfect correlation. Also, Willy is moving into uncharted waters. I know of no production processor, ever, that has a pipe as long as Willy's. Therefore, there is no empirical evidence to guide us about Willy. Additionally, there are diminishing returns. As you add pipeline stages, the additional benefit you get is reduced. The chip becomes bigger, and the probability of dividing work between stages correctly is reduced. So what is the magic number? We really don't know. I agree with Scumbria that the place to look is actual projects. Empirical evidence is the best. But we seem to have a shortage of production CPUs with 15, 20, 25, 30, etc. stage pipelines. Scumbria has stated that he believes that 20 stages (less than Willy I believe) is where the benefits start being less than the penalties. Maybe he is right. But one thing to consider is that the benefits of each additional pipeline stage decrease as you approach the breakeven point. Those additional transistors could have been used in other ways to speed clock rate. My point is that Willy's ultra deep pipeline is not on well explored territory. Yes, it might be the fastest x86 design the world has ever seen. But Willy might, just might, be completely destroyed by Mustang. Looking at this a different way, it always seemed to me that no one could design a CPU faster than Dirk Meyer. IMHO, no one ever came up with a RISC processor faster than the Alpha. Now Dirk and his buddies show up in the x86 world, and we see the fastest x86 processor the world has seen. The fact that for over a decade, the fastest CPU of the catagory is whatever Dirk has done is very impressive. To me, that places the burden on Intel, not AMD, to prove that their processor will be faster. To date, Intel has convinced me that Intel has a chance of having the fastest processor. IMHO, one chance is all that Intel has now. Mike