SI
SI
discoversearch

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: chic_hearne who wrote (109822)5/8/2000 7:46:00 PM
From: enzyme  Respond to of 1578539
 
interesting post from TMF:

What has to be the most interesting bit of news for this month is the reports
that Intel will begin shipping Katmai processors in volume once again. Yes, I
am talking about Slot 1 Pentium III 550, 600 and (perhaps) 650MHz processors
in quantity. One manufacturer told me that Intel notified them last week that
as much as 60% of all shipments will be comprised of these processors until
at least September."

I'm not sure if this is a big deal or not. See the table I constructed below:

introduction shrinks...
arch date microns microns (months from intro) repeated for some
P5 May 93 .8 .6 (10) .35 (21)
P5MMX Jan 97 .35 .25 (6)
P6 Nov 95 .6 .35 (2)
PII May 97 .35 .25 (8) .18 (25)
PIII Feb 99 .25 .18 (8)

To get a better understanding you probably would have to plot these out since
they included overlapping chip introductions. Anyway, here's my premise:
Intel always has 3 process technologies for CPUs (old, current and future). Old is
the volume. Current is the latest thing and Future isn't yet in production.
As the future technology is about ready to come on-line, the current technology
crosses-over from the old and becomes the high volume. This cross-over happens
sometime in-between the first and second shrink around 16-20 months after a new
architecture was introduced. For PIII this means that it would be sometime around
June to Sep. Note that this means Katmai to Coppermine in this case.

I think that Intel is saying that this will be closer to Sep than June. But if
total demand is very high they may have to keep older FABs running longer.
Once the future technology is running, they try and phase out the old technology
as quickly as possible, but it generally takes another year. IOW, they don't startup
a FAB to use just for 1 year, it generally runs for 18-30 months on the same CPU.

If anyone has any additional data like this for Intel or AMD please share it.

Mike

boards.fool.com



To: chic_hearne who wrote (109822)5/8/2000 9:18:00 PM
From: THE WATSONYOUTH  Read Replies (2) | Respond to of 1578539
 
Re: "From all I've read and heard, I can't stress enough how much of a competitive advantage AMD is gaining by going to copper now. Look at how the new POWER3-II is helping IBM servers destroy SUNW servers."

Sorry, but I think you are ill informed. The IBM R6000 UNIX servers (S80 series) are still in .25um technology. At .25um, there is NO benefit of going to copper. The silicon technology in these S80 systems is all ready 23 months old. This illustrates how antiquated SUN's Ultrasparc II systems and technology really is. Had the entire R6000 line been up dated soon after the .25um technology qualified (spring of 98) Sun would no way have dominated the Unix server business in that time frame. The R6000 line should have been entirely updated by Sept 98 with .25um Cu technology. Instead, the first systems did not appear until Sept 99 and the rest of line will only this month be updated. What is even worse is that an entire generation of technology was skipped prior to the .25um Cu technology. So SUN was virtually handed this business via another multi billion dollar blunder by IBM. This should show how vulnerable SUN is on the performance front. However, Ultrasparc III low end systems will start to appear late this summer with the high end Serengeti servers scheduled for the end of the year. So, it appears SUN will be OK despite IBM's screw ups.

THE WATSONYOUTH