Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To Matt Taylor: Processor Dogma

Author: Charles Worthington

Date: 02:30:10 02/21/03

Go up one level in this thread


On February 20, 2003 at 15:02:12, Matt Taylor wrote:

>On February 20, 2003 at 08:18:08, Andrew R. Case wrote:
>
>>On February 20, 2003 at 00:04:43, Matt Taylor wrote:
>>
>>>On February 19, 2003 at 21:43:10, Charles Worthington wrote:
>>>
>>>>On February 19, 2003 at 20:59:19, Charles Worthington wrote:
>>>>
>>>>>On February 19, 2003 at 20:58:20, Charles Worthington wrote:
>>>>>
>>>>>>Yes Matt many people here do exagerate their figures. If the Intel guy says
>>>>>>Intel will produce a 20 GHz cpu next week then the AMD guy will tell you that he
>>>>>>can still clock his 2600 to 20.1GHz....and vice versa. The truth is that they
>>>>>>are both good processors. Is the AMD on par with the Xeon in quality or speed?
>>>>>>No, it isn't. The 2.8 Xeon was tested alongside the AMD 2400MP and clearly
>>>>>>out-scored it. The tests were performed by an independant company who had no
>>>>>>reason to lie. Yes as far as speed goes they can both be clocked up. But people
>>>>>>should try to bear in mind that the AMD and the Xeon are very different products
>>>>>>designed for very different markets. The Zeon is primarily designed for
>>>>>>workstation and server platforms where both speed ans stability are required.
>>>>>>They are high end processors and are priced accordingly. The Athlon is designed
>>>>>>more for use in PC's both single cpu (xp) and dual (Mp). It is a fast processor
>>>>>>for the buck but was not designed to be overclocked to try to match a Xeon. If a
>>>>>>2 GHz cpu was stable at 3 GHz then, trust me, AMD would market it as a 3 GHz.
>>>>>>The bottom line is simple: If you want performance and stability and you can
>>>>>>afford to pay for it buy the Xeon. If you want performance but are on a limited
>>>>>>budget buy the AMD. But_don't_ go out and buy a Yugo and then come back here and
>>>>>>tell me that you are building it to catch the neighbors Ferrari. If the
>>>>>>neighbors Ferrari bothers you that much then buy one too. If you can't afford
>>>>>>one then be happy with the yugo. It will still get you where you are going.
>>>>>>   To me trying to compare Intel and AMD is like comparing apples and oranges.
>>>>>>You are comparing a low end home and light business processor against a high end
>>>>>>heavy duty cpu designed specifically for Workstation and server platforms. That
>>>>>>is like comparing a Yugo to a Ferrari.
>>>>
>>>>Charles
>>>>
>>>>
>>>>P.S. The figures for Deep Fritz 7 that I have observed on the chessbase server
>>>>are as follows: 2.2GHz P4 750kNs; 2.8GHz P4 1038 kNs; 3.06GHz P4 w/HT 1,297kNs.
>>>>AMD 2400xp 1030 kNs; Dual AMD 2400mp 2280kNs; Dual AMD 2400xp 2166 kNs.  I have
>>>>played against or observed  most every dual on that site. I did see one machine
>>>>post 2400 kNs but that was only once and may have been during a spike. I am
>>>>unsure of the type cpu which produced that figure. I have seen overclocking
>>>>claims here so wild as to defy belief. I have seen claims of single cpu AMD's
>>>>pulling 1,800 kNs with Deep Fritz 7 (without any way of hyperthreading). They
>>>>need to put that in a dual then come to the chessbase server and let us all see
>>>>the 3300 kNs post. Of course they won't do this because they cant make that
>>>>post. It is easy to post figures on here where no one can prove or disprove them
>>>>but it's not so easy to prove it over the board on the chessbase server. The
>>>>figures i have posted above were with 64 MB Hash tables in 3+2 time controls.
>>>>And yes, some of those AMD dual machines are overclocked the owners have told me
>>>>so themselves. The above figures are how it is in the real world. They are
>>>>unbiased and accurate. And i challenge any AMD here overclocked or not to show
>>>>all of us on the fritz server than he can exceed 2600 kNs with Deep Fritz 7.
>>>>Forget posting the benchmarks here...let the chessbase server do that. And be
>>>>sure to use the Deep Fritz 7 GUI and send evaluations. The kNs will be posted at
>>>>the end of every game.
>>>
>>>AthlonMP 2600 is available. Comparing AthlonMP 2400 to P4 3.06 GHz isn't exactly
>>>fair, is it?
>>>
>>>-Matt
>>
>>          Sure it is, nobody has a dual 2600MP yet....that is the problem
>
>Nobody has a dual 3.06 either. I have not seen the 3.06 sold yet, though I
>suppose Dell has them. I can order an MP 2600 right now if I want to.
>
>-Matt


Matt, I have ordered the dual 3.06 system from Dell. Production was delayed a
few days (evidently Dell was selling the system before they had all the parts in
stock) but I am told that the machine is now in production. With next day
shipping I should have the machine one day next week and will begin testing with
fritz and crafty. I will post the benchmark results here. I have no idea what to
expect with crafty but i have calculated that the results with df7 should be in
the 2550 to 2650 kNs range based on the performance of the 3.06 P4. I suppose
that a small boost from the P4 figures is possible with a workstation board but
I can't be sure.



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.