Re: I have heard that SUNW's ULtrasparc III is not nearly as good as SUNW would have the world believe and that others like IBM (according to Chic) and the Intel's of the world will cut into SUNW's business.
ted,
This is what WatsonYouth wrote:
Are you saying these new Wildfire systems will ship with .25um Silicon technology? Seems a bit late to still be introducing .25um systems. That's how IBM screwed up vis-a-vis SUN. Only 60% faster than high end S80?? Remember, the silicon technology in the S80 is all ready 23 months old and they don't even aggressively sort to real short channel lengths. Also, the IBM R6000 line is using an old chip design. IBM will update the S80 line before year end with much more aggressive .18um silicon technology and then next year again with a new (Power 4) design. I suspect Wildfire will leapfrog over IBM S80 and Sun E10000 for the time being and them be leapfrogged itself by Sun with Serengeti and IBM with Power 4 by next year. So, I don't see anything earth shattering here. What is most apparent is the inability of all the competitors (SUN/COMPAQ/IBM/INTEL/HP) to put out competitive CPUs/systems on any kind of a reasonable timely schedule. It makes AMD execution look superb.
ted,
Everyone has plans, no one has delivered on time. I believe IBM is in the best position of anyone because they have already moved to copper. This is a major hurdle to pass. If your Al process is taped out, you're stuck until copper can yield. Because IBM already has stable copper processors running I have a lot more faith that they will hit thier targets. Like Watson said, the SP 80 (btw, SP means Scalable Parallel) is set for 2 aggressive upgrades. SUNW is 2 years late on the Ultra Sparc III. Keep in mind, the SP 80 that blows away the E10000 is only a 24-way system compared to SUNW's 64-way. This gap will get even worse with IBM's 2 aggressive upgrades. IMHO, SUNW has taken a half-assed approach with its processors and they will pay severely in the next few years.
If anything can be taken from this, look at AMD's efforts with copper. They will have working copper out in a month. INTC is still living a pipe dream that they can go to .13um on Al. If there are problems going to .13um, or if copper doesn't yield, INTC will be stuck in neutral. Meanwhile, AMD will have copper yielding and should be able to perform process shrinks far easier than INTC. I expect AMD and IBM to lead the way in copper for some time to come.
I think it is very dangerous and extremely risky for INTC to be going to copper at the last minute. If things don't go as planned, they will have nothing to fall back on. BTW, this is very good for AMD.
chic |
