Author: Vincent Diepeveen
Date: 04:54:15 10/01/02
Go up one level in this thread
On September 30, 2002 at 09:09:12, Aaron Gordon wrote: >On September 30, 2002 at 04:11:57, Vincent Diepeveen wrote: > >>On September 30, 2002 at 01:43:52, Aaron Gordon wrote: >> >>>On September 30, 2002 at 00:21:12, Jeremiah Penery wrote: >>> >>>>On September 30, 2002 at 00:05:40, Slater Wold wrote: >>>> >>>>>I got the annoucement in August that the 2600s would be coming out soon. It is >>>>>now September and they are barely rolling out. (As in, several vendors who are >>>>>selling them don't actually have them yet.) AMD is horrible about this. >>>>>Announce a chip, and 3 months later it hits the streets. >>>> >>>>Well, it's a paper launch, and most every company does it at some point or >>>>another. Does AMD do it more than most? I can't answer, because I don't keep >>>>track of those things. >>> >>>Intel does most of the paper launches. I the 2600+ is the only time I've seen >>>AMD try that. >> >>I remember a lot of other launches. The K7 was announced 2.5 years before >>it was on the market. specs were posted 1.5 years before it was on the >>market and then it still wasn't there. Now you can cry about intel delaying >>the K7 by some 'blackmail' (i do not know how to call it otherwise as >>my english is not too well; they didn't want to deliver their intel chipsets >>to mainboard manufacturers who wanted to make K7 motherboards and 90+% of >>their sales were intel boards at that moment, so they actually needed them >>bigtime). You can cry about other things too, but both manufacturers are >>already drumming years ahead about new things. >> >>My own drumming about a diep windows version is completely amateuristic >>compared to their drumming up to 7 years ahead. > >Just because they post specs doesn't mean they're going to release it soon >after. Only when they announce that they're going to release it should you start >the clock. :) Without jokes, i only get serious about a new processor when i receive the first benchmark of DIEP at it. That's usually 6 months to a year ahead of market introduction. And that's pretty well needed, because if such a processor is going to be the new standard, then it's not really stupid to take loads of time to rewrite a part of the engine such that it works better at it :) >They posted the specs and THEN announced it. Not the other way >around. Hammer specs have been around for a LONG time now. Yeah, they did say >they were going to support up to 8 cpus for the K7's and yes we haven't seen >those. Most likely because of market demand. Most likely because they didn't get it to work. I have a dual k7 and i'm very happy about it, but i always have the impression it's a very instable system compared to for example a dual P3. You simply don't want to know how many reboots i need, killing of explorer, reinstalls of windows, etcetera. My dual P3 i installed the same software and same OS/servicepack like at this machine and the dual P3 is completely working without problems. Then there is other problems for AMD to get processors stable to run SMP. We have already up to 300Mhz higher clocked new XPs, than there is MPs. I'm not saying intel is a hair better here with the P4. Not at all. By keeping L1 caches small at it, they manage of course to clock it that high. But even if you clock a P4 to 3.5Ghz, then still the same 'unstable' AMD is way faster. I wonder how 'stable' that hammer is. Producing 1 such a cpu doesn't mean they can produce it cheap for everyone. So i wait till i have more reliable results on the hammer for DIEP. Basically i lack having a cross compiler for it. >I have seen dual/quad opterons >though so at least there is SOME hope. Slate is probably right, we probably >won't see the 2800+ for another month. Same for Intel's P4-3GHz. Probably will >see that released in December (but it's being announced today or sometime soon). 1 month is easy waiting for. but 1 year...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.