Author: Robert Hyatt
Date: 15:23:22 02/20/03
Go up one level in this thread
On February 20, 2003 at 16:46:50, Aaron Gordon wrote: >On February 20, 2003 at 15:01:09, Robert Hyatt wrote: > >>On February 20, 2003 at 14:16:12, Aaron Gordon wrote: >> >>>On February 20, 2003 at 11:42:49, Robert Hyatt wrote: >>>>That is _not_ the same idea. The idea that a vendor purposefully underclocks a >>>>chip >>>>is ridiculous. The idea that they don't release the next generation at a faster >>>>clock rate >>>>until the current supply of slower chips is exhausted is not contradictory at >>>>all. Two >>>>totally different business practices, one of which makes economic sense, the >>>>other makes >>>>zero sense. >>> >>>They make ALL of the chips off the same line. >> >> >>You are _completely_ missing the point. We are talking about overclocking the >>_high_end_ chips. Not the low-end. The fastest production intel chips are >>3.06ghz. >>I don't give a squat about overclocking the 2.4ghz processors. We are talking >>about >>the _top_ end. >> >>This is about taking the best and overclocking, not about taking something that >>was >>intentionally marked low simply to fill a market niche request... > >I was referring to those as well. An AIUHB 0301 is an AIUHB 0301, whether it's >on a 2800+ or 1700+. These will do 2.3Ghz MINIMUM which means AMD is producing >at the very least 2.3GHz silicon. Same goes for Intel, the absolute MINIMUM >you'll get off a P4-3.06 is 3.2GHz. I consider these the absolute upper limits >for those chips, any more and you really are overclocking. > No, the absolute minimum you will get from a 3.06ghz pentium IV is 3.06 ghz. If you want to go beyond what _really_ bright engineers say works reliably, feel free to do so, but you won't convince me that just because it works for the tests you run it will work for my mission-critical applications. I leave _that_ to the Intel engineers, they will tell me when they have a 3.2ghz chip that is _reliable_ at 3.2ghz, because they will produce one and sell it. > > >>> Why do you think you can run out >>>and buy an AthlonXP 1700+ (1466MHz) with the Thoroughbred-B core for $56 and >>>overclock it to 2.1-2.3GHz? Try that with one of the very first 1700+ chips, you >>>will not get over 1.6GHz. Same thing goes for my old Celeron-2 566MHz. It does >>>1.1GHz (yes, 566 to 1100) air-cooled. This is a cC0 and basically is a P3-1GHz >>>core with some L2 cache disabled. Intel and AMD both make the same stuff and >>>mark it to whatever they feel is needed. If Celeron 566's are selling a lot, >>>they'll start marking them 566 to meet demand. 2100+'s are selling like >>>wild-fire, AMD is putting their latest and greatest silicon in those chips. You >>>can pay $300 or whatever it costs for a 2800+ *OR* you can get a 2100+ with the >>>*EXACT* same core for $97. >>> >>>You may know about programming, Hyatt, but you sure don't know about >>>overclocking. >> >>It isn'what I know about overclocking. It is what _you_ are failing to read in >>this >>discussion. Again, to keep it simple, We are talking about taking the fastest >>chips >>offered and overclocking _those_. Not about taking a 2.4gh part made on a >>2.8ghz >>production line, and then overclocking _that_ to 2.8. Because you _could_ just >>buy >>the 2.8ghz part. We are discussing going _beyond_ the current leading edge... >>Nothing >>more, nothing less... >> >> >> >>>I've been doing this all my life, I even built a liquid cooler >>>when I was a kid (I think I was 13-14 years old or so). I mean completely built >>>from scratch, evaporative radiator and all (the evaporation helps keep the water >>>below-ambient). All my previous overclocks were 100% stable and infact I've >>>still got some of those systems still today, chugging along running fine. >>>Here are a few of my past overclocks: >> >>I don't care what you have done all your life. My dad shaved with a straight >>razor. I >>don't. Its dangerous. I don't care about clocking slower parts. I do care >>about overclocking >>the best parts to go beyond what the engineers think is safe. That is _all_ I >>care about. >> >>Buy all the cheapo parts you want and overclock them all you want... But that >>is not >>the point here.. > >Cheap doesn't mean bad. My entire computer cost less than ONE of your CPU's and >my system is still much faster than you using one cpu with HT. I get 2.0-2.5mn/s >in the endgame, too, just like yours with two cpus. :) You could buy 10 >computers identical to mine for the price you paid for yours. When I said "cheapo" what I meant was, for example, the old pentium pro 200. Intel started selling 180's as well at a cheaper price. And eventually _all_ were coming off the same fab line, and some were stamped 180 and some 200, even though all had passed thru the same QC checks at 200mhz. But there was a demand, and it was cheaper to take 200 parts and stamp/sell them as 180's. And yes, the 180's overclocked to 200 quite well. But that is not the point here. I am talking about taking the top of the line part from either AMD _or_ Intel, which is their most expensive part also, and then overclocking that. And I believe _that_ is a road to trouble if you have to depend on the results to be 100% correct. Anybody can take slower parts and overclock them, once you discover both come off the same line, but that isn't the same as taking a part the engineers decided could run at XXXX mhz, and run it at XXXX+N mhz and expect 100% reliability. Unless you believe the engineers are way conservative or incompetent. > >I copied a position where you got 2109kn/s in crafty during the gguseinov match >on fics, in the same position with crafty 19.3 I got 1661Kn/s. If you figure 50 >elo per doubling in speed this comes out to 13.485 elo more than my computer. >Now, your system was $5000.. mine was $500. This is $333.70 per elo point. Going >over my PC speed-wise. Kind of ridiculous, isn't it? You know whats even better? >My pc is nearly silent (liquid cooled) and my CPU doesn't run blazing hot >either, average cpu temp is 77F/25C. =) My cpus don't run blazing hot, although there is another significant performance difference in our boxes. I can copy 8 gigs of data in 30 seconds. That is important for chess endgames with tables. I don't think you can get 1/10th of that speed with IDE drives of any kind. But that is yet a different subject. And copying a position from a game and comparing NPS doesn't work. You have to compare both machines starting from scratch on the same position. A pre-loaded hash table can lower NPS but make the search go faster. If you want to compare, pick a position, or we can run the bench with default settings and compare nps or time to finish, for a much better comparison...
This page took 0.02 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.