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: kash johal who wrote (63532)	6/27/1999 6:43:00 PM
From: Bob Drzyzgula	Respond to of 1571281

> Bob, Thanks for the post. You're welcome. > For business2business servers I can see value in high end > servers As well as servers for internal functions, e.g. ERP. Part of the challenge in technology procurement is understanding when you really have one of those applications (ask eBay) and when you don't. If you don't, but you still have intense computational requirements, the challenge is finding equipment that can do the computations without having to take all that RAS overhead that would just be wasted. This isn't always easy. > For lots of other apps. Linux is winning out and the K7 will > be the lowest cost/highest performance Linux system out > there. I agree. That's why I'm watching both things closely. We already use Linux for several applications, and my current plans are to modify our Solaris end-user interface environment so that it looks as much like Linux as possible; moving the GNU tools more to the foreground, for example (we have done some of this for years, but I'm talking about using e.g. the GNU versions of ps and chmod, so that even scripts can be identical cross-platform, and users won't have to be confused whether they should do a "ps aux" or a "ps -ef"), and merging authentication and naming onto software platforms not dependant on Solaris (e.g. LDAP). I want the two operating systems to be as fungible as possible, without damaging the stability of Solaris or its ability to run the critical applications. > I am talking about the version of L7 with larger full speed > cache in the slot B motherboards coming later this year. That also is where my greatest interest lies. > I suspect dual K7 systems running at 800M+ will run well > under $5-6K. For us, again assuming (dangerous, of course) that AMD follows through on all this, such systems would be dynamite competition to the UltraSPARCs we now mostly use. Alphas are also a possibility, but it is more difficult to buy the parts and adding a third instruction architecture to our support efforts is a daunting task. > Their performance should easily exceed a Quad Xeon 600 with > 2Mb cache - not that such a beast is available today. From what I see, the big question mark is the system bus. Good SMP memory busses are non-trivial (even Intel isn't particularly good at them). If AMD can leverage the technology developed for the Alpha into a moderate volume at a decent price point, they could have a path to sell into the high end. But, once they start loading it up with RAS crap, then I start to lose interest, which of course places a limit on how far they can take the ASPs in the scientific market of which I might be considered a part. > Such sales will be significant to AMDs bottom line in Q1 > 2000 methinks. If ("if" -- what am I thinking?) Intel keeps up the price pressure at the midrange and below, this seems about the only place they could pick up anything in the way of a margin. As many others have said, it seems that the only way they can win this game is to pass Intel. Matching them simply isn't enough. Right now, I'm paying about $3K for a 300MHz UltraSPARC II (the ones with the real bus interface and the cache, not the IIis, which cost about $1300 with a motherboard) from distributers. The quad-processor motherboards cost about $6K. If I can buy a dual-Athlon board, two 800MHz Athlons, and 1GB of memory for, say, $6K aggregate ($1K memory, $1K motherboard and $2K each for the Athlons), AND if the Athlons work like they seem that they should, then it will be a mighty attractive platform indeed. The UltraSPARC-III is supposed to trickle out in the same timeframe, but if history is any guide, the first USIII systems will cost a fortune, and it will be a year before you can get them for a reasonable price. Best Regards, --Bob Drzyzgula