Author: Eugene Nalimov
Date: 12:57:55 02/28/02
Go up one level in this thread
On February 28, 2002 at 14:38:01, Dann Corbit wrote: >On February 28, 2002 at 14:22:18, Eugene Nalimov wrote: > >>Sorry, I don't understand. We are not talking about "performance per MHz". We >>are talking about *absolute* performance. And here Alpha is 3% better athan >>contemporary P4 Xeon. > >True enough. Certainly Intel does have better fabs and better technology than >those who manufacture Alpha chips. Now that they have the Alpha technology, >they could build one at .1 micron if they so desired. Once again: we we not talking about "what if", we were talking about *shipping products*. Do not forget that Intel can make PIII/P4 *significally* faster, too, if CPU price is not a concern -- e.g. it can add more cache on die, more pipelines, etc. I agree with you that "ultimate Alpha" should be faster than "ultimate x86" built using the same process, but I do not believe it'll be 2x faster. More like 1.2x-1.3x on integer code. >>Or probably we should talk about price/performance? :-) > >A 2 GHz Xeon system will also be "too expensive." Just go to http://www.dell.com/us/en/bsd/products/minicat_precn_530.htm and price one yourself. I'd say that $2.7k is reasonable price for cutting edge technology (CPU is faster than one we were talking about, it has 512K L2 cache). You can save ~$500 by installing simpler graphics card, declining 3 years 4-hours on-site support, etc. If that's still too much you can buy ~10-15% slower system for ~$1.1-1.3k. Eugene >It's amazing what they can charge you for a teaspoon of sand. > >>Eugene >> >>On February 28, 2002 at 13:42:34, Dann Corbit wrote: >> >>>On February 27, 2002 at 21:55:19, Eugene Nalimov wrote: >>> >>>>Alpha is not "2X as fast as anything around". >>>> >>>>Fastest submitted SpecInt2k result for Alpha is 621(base)/679(peak). AlphaServer >>>>ES45 Model 68/1000, hardware available October 2001. >>>> >>>>Dell Precision WorkStation 530 (2.0GHz P4 Xeon) has SpecInt2k >>>>642(base)/663(peak), hardware available September 2001. >>>> >>>>679 is greater than 663, but not 2x. Not even 1.1x. >>> >>>Interesting. At twice the clock speed, the Xeon was almost as fast! >>> >>>Thanks for proving my point.
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.