Author: Robert Hyatt
Date: 12:20:10 12/02/02
Go up one level in this thread
On December 02, 2002 at 14:33:40, Aaron Gordon wrote: >On December 02, 2002 at 12:17:18, Robert Hyatt wrote: > >>On December 02, 2002 at 10:18:42, Aaron Gordon wrote: >> >>>You can find out where the chip was made, when it was made and what stepping the >>>CPU is. For example.. if someone walked up to you with three Thunderbird cpus. >>>One was an AFFA stepping, one was a BXHA stepping and the other was an AXIA.. >>>which would you choose? No idea? Only by LOOKING at the top of the cpu can you >>>figure out which particular chip it is (I'm not talking about just the cpuid >>>stepping info). Someone that knows what they're doing would choose the AXIA. It >>>has a 1.5GHz core (doesn't matter what it's marked) and can run 1.5GHz without >>>problems. BXHA can't do over 1.2GHz and an AFFA has problems hitting 1GHz. How >>>about on the line below the stepping? There are two possibilities.. one will >>>start with a number, the other with a letter. Any idea which one to choose? The >>>CPU w/ a letter (usually a 'Y') can run a good 50MHz faster. If you don't like >>>overclocking you can always drop the voltage and leave the chip at it's current >>>clock speed. An AFFA Tbird at 1GHz would need 1.75v at the least. A 1GHz AXIA >>>could run 1.3v. Thats going from 54.3 watts to 29.96 watts. If you get an AYHJA >>>you can just do watts * 0.80. >> >>You are answering your own query. I look at the specs. I order the product. >>If it >>doesn't meet its specifications, back it goes. Regardless of stepping, >>revision, manufacture >>date, fab plant id, or anything else... >> >>Numbers on the top of the chip don't mean a thing without an encyclopedia from >>the >>vendor explaining the differences between various steppings. The write-up >>generally >>explains what the chip being discussed does, which is the important bit of >>data... >> >>your VIN on your auto contains a _lot_ of info. Do you use it? I don't. My >>insurance >>company does of course... but not me... >> >>> >>>How about this one.. an AYHJA core. These run 20% cooler than standard Tbird >>>cores. How do you know if you've got one? LOOK AT THE CHIP. >> >>Totally wrong example. We are not looking "at the chip" since we don't have it >>in our hands. We are looking at a picture that goes along with a long article >>that >>discusses the specific chip. That's a _big_ difference... >> >> >>> >>>Look at the AMD and Intel datasheets on their CPUs. Look at the information >>>on where it was made, the area where it shows the date (what week of the year) >>>it was made, etc. Another example.. Week 25 and newer Celeron 366's came with a >>>550MHz core. Every single one could do 550 at default voltage without problems >>>at all. I wonder how you could tell? Hrm.. perhaps by LOOKING AT THE CPU. :) So, >>>lets go back over this. By looking at the CPU you can figure out where it was >>>made (has an effect sometimes on overclockability), what stepping the CPU is and >>>you can possibly get a cooler running CPU. Yeah, looking at the CPU is useless. >>>Good one Hyatt, very amusing. :) >> >>Didn't say "looking at the CPU is useless". So your comment _is_ very amusing. >>We >>are talking about a _photograph_ of a chip we won't every have in our hands, >>along with >>a review of the chip that we can read to make conclusions... >> >>I don't see why that is so hard to understand... > >Didn't say looking at the CPU is useless? I quote you from your previous >message... "Except that the picture of a microprocessor chip carries absolutely >no technical information of any kind...". This was said by you. A bit >contradictory don't you think? Not at all. Look at _context_. A new chip. A review of the new chip. A picture of what the new chip actually looks like vs a picture of what the new chip will look like but a real picture was unavailable. > >As far as you not using stepping codes and all that stuff.. it's better to do >so. You can specify in the comments tab when buying something. Lets say you >order an AthlonXP 2200+. It's a big tossup right there whether you'll get the >older Thoroughbred-B core or the newer Thoroughbred-A core (runs cooler, clocks >MUCH higher). If you knew what you were doing you could just specify in the >comments tab that you'll want one with the markings, "AXDA2200DUV3C" and NOT >"AXDA2200DKV3C". Again, you are _missing_ the point. The text of the article covers those details. That is the _entire_ point here. The picture was "window dressing" to show what it looks like, because some folks want to see the pictures. you can remove the picture and get _just_ as much info from the text of the article. For a 3.3ghz cpu, that is not even released yet, "stepping" is a moot point. It will be the latest thing out when it finally hits the street. Other details will only be released as the vendor chooses, so the top of the chip doesn't offer a thing until it is released and available, and newer versions are out behind it so that stepping is important. I _do_ pay attention to stepping. in quads, Intel specifically requires that the four processors either have the _same_ stepping, or else that a compatibility table shows that two different steppings can work together in a quad. But for a yet-unreleased chip, that is a moot point... > >Before they ship your order most companies will spend a few minutes looking >around for one of those for you. Now you get a cpu with a superb core. You could >run it at default speed with a much lower voltage or push it from 1.8 up into >the 2.4-2.5GHz range. If you had not bothered specifying the DUVC3 then you >could very well have received the Tbred-A and have a hot running cpu that will >have problems making it up to 2GHz (just 200MHz from your current clock). > >So perhaps instead of generalizing you should instead say, "I haven't a clue >what all that stuff means on the CPU so it has no value to me". Saying, and I >quote again, "Except that the picture of a microprocessor chip carries >absolutely no technical information of any kind..." is pretty ridiculous. Again, for a _new_ and not-yet-released chip that is only distributed in a very few copies, for evaluation, the top of the cpu carries _no_ useful information. It won't be available via a vendor's web site until after it is available. I _still_ don't get your point. We are not talking about a chip that was released _last_ year. We are talking about an evaluation copy of a chip to be released _next_ year. There is a _big_ difference. > >Also.. Yes, I do understand what you were saying previously. You're aparently >not caring that Tom faked the picture and the review. The facts are there, >choosing to believe them is entirely up to you. Personally I think looking at >faked reviews is a waste of time but.. thats just me. ;) I would not say "faked the review". No information on that. As far as the photo goes, I can certainly see why a "mock-up" might be done if the real processor had to be shipped off to someone else quickly... Wouldn't bother me for them to do that. I ordered a ducted-fan radio-controlled model jet from Bob Violett a few years ago. He sent me a picture of a "mock up" of the finished product, which differed a good bit from the final version he sent me to build. Didn't bother me one bit as I understood the concept of "pre-production part"...
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.