Author: Aaron Gordon
Date: 00:38:50 07/03/03
Go up one level in this thread
On July 02, 2003 at 20:59:42, Vincent Diepeveen wrote: >On July 02, 2003 at 17:58:58, Aaron Gordon wrote: > >Gordon you are not answerring what is the >worst case timing path in an Athlon? Generally its 5% under what the absolute maximum of the chips are at a reasonable voltage for that particular cpu. I use 1.75v for Athlons, 1.65v for P4. This comes out to be right at 2.4GHz. >If that path is there at 2.2Ghz and intel gives them a 2.2Ghz sticker or a 1700 >sticker and you clock them to 2.4Ghz that means simply that you *definitely* are >crossing that border. Why don't you email the AMD and/or Intel marketing departments along with the engineering departments. You can't seem to get it through your head that they mark the chips down to whatever they feel like. They're selling a massive ammount of 1700+ chips, what are they going to do? Clock the cores they're making now (which are 2.4GHz+) and sell them as whatever. This doesn't mean the cpus "can't do more", it's "AMD wants to fill the orders of 1700+ chips". >It is trivial that some XP1700s will be in fact stickered XP1700 even when some >of them can go XP2200. > >That doesn't take away the fact that they will not be capable of going to 2.4Ghz >when the fastest processor Produced by AMD is 2.25Ghz in those series. > >At 2.4Ghz you are definitely above specifications and you will lose quality in >the chip. QUALITY you hear. > >You keep answerring here, but it is big nonsense of course to say they run very >well at 2.4Ghz. You're only assuming things. I actually talk to people from AMD, run tests like crazy and etc. >They just don't blow up at 2.4Ghz that's all you proof. > >If i can run 2.4Ghz for a small while at a tournament with a 2.1Ghz chip, i'll >do it perhaps. > >But i know the risks very well. I know that with DIEP which is basically doing >integer math, that the chance is not so big that it will be hitting a path that >is the weak chain. > >Denying that is complete nonsense. > >AMD would cutoff their left arm to get them stable clocked to 2.4Ghz if they >could. They would only sell a small % anyway at 2.4Ghz, so only a few would be >needing to test till 2.4Ghz anyway. AMD is taking the easier route, and just because they can clock the MHz higher doesn't mean they will. They can drop the voltage and end up with cooler chips instead. AMD is more focused on higher cache chips right now and high bus speeds. This lets them put a higher PR rating on the chip without having to clock it way up. This is the reason Tbred-B chips are being phased out for the "Thorton". AMD only likes having *1* fab line, as I said 500,000 ^ 498308389 times. A Thorton is a Barton (512K L2) with half of the L2 disabled. Its easier to make all of the chips off a single line and disable the cache than have multiple fab lines. Now, this brings me back to what I was talking about before. They also make the fastest chips possible. They speed vary, of course, but as far as absolute minimum speed they're doing right about 2.4GHz. Sure people may be able to hit 2.5-2.6GHz, but I think thats too "on edge" for anything important you may want to do. 2.4 would be fine, however. If you ever used a P4-3.06GHz there are good chances it is closer to running at the very edge of stability than any of my chips. They can do quite a bit more, some P4-3.06's can't even overclock 5%. How can you consider my CPU "unstable" running 10-15% under the edge and consider a P4-3.06GHz stable running 5% under the edge? I see it won't do any good to talk to you any further, since you don't believe me. Wasting time is not something I like so, go email some knowledgeable people at AMD (people who design the chips, and some of the top marketing people) and see for yourself. Heck, go email Intel. They do the same thing. >They didn't however. That means trivally that they concluded that something goes >wrong in a certain combination of instructions fired at the cpu when clocked to >2.4Ghz > > > >>On July 02, 2003 at 13:08:36, Keith Evans wrote: >> >>>On July 02, 2003 at 12:38:42, Aaron Gordon wrote: >>> >>>>You keep thinking the chips are technically overclocked, which infact they are >>>>not. These chips have 2.4GHz cores and run those speeds with ease, at low >>>>voltages. If you push up to 2.5-2.6GHz, you have to raise the voltage and then, >>>>and only then are you actually 'overclocking'. I'm sure you would still consider >>>>an Athlon XP 1700+ (1.46GHz) with a 2.4GHz core set to 1.53GHz "overclocking". >>>> >> >>Read this post of mine about how I figure out the core type: >>http://www.talkchess.com/forums/1/message.html?304354 >> >> >>>Please explain how you know that these are 2.4GHz cores. Do you have something >>>in writing from AMD? I personally define overclocking as operating parts at a >>>frequency higher than the manufacturer guarantees. >>> >>>When I looked at your website I saw the disclaimer that overclocking chips may >>>result in a shorter lifespan. I agree with this. >> >>I have to put this, at least until I get my "Overclocking Myths" page finished. >> >>>>Well, feel free to buy one of my chips and run it for a year doing your matrix >>>>calculations. :) BurnK7 is much more intense than anything like that. If any >>>>program was unstable enough to produce errors then BurnK7 would be able to find >>>>the problem faster, as it would heat the chip up MUCH more and thus cause the >>>>instabilities to be seen quicker. The chips I test go through harsh testing OVER >>>>the speed I rate them at, at higher temperatures than anyone would normally run, >>>>THEN I clock it back. This results in a completely stable chip. >>> >>>What is the worst case timing path in an Athlon? Just because BurnK7 heats up >>>the chip does not mean that it exercises the worst case path. >> >>Hard for me to explain, but the extra heat BurnK7 generates would push your chip >>'over the edge' if it was that unstable. Running your 'important' code for >>example would result in a cpu temperature a few degrees cooler. Read my post for >>proof how the cpu would be stable in those few degrees. Also, I don't run the >>chips on the very edge. Don't knock my pretested chips before you try them. >>I've taken a lot of time/effort to make sure they work flawlessly. >> >>>It looks like BurnK7 doesn't even attempt to test every instruction: >>> >>>"N E W burnK7 for AMD Athlon/Duron/Thurderbird has been released. >>> >>> >>> >>>These programs are designed to load x86 CPUs as heavily as possible for >>>the purposes of system testing. They have been optimized for different >>>processors. FPU and ALU instructions are coded an assembler endless loop. >>>They do not test every instruction. The goal has been to maximize heat >>>production from the CPU, putting stress on the CPU itself, cooling >>>system, motherboard (especially voltage regulators) and power supply >>>(likely cause of burnBX/MMX errors)."
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.