Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: " D e e p - I m p a c t " Swiss 30 rounds, round 18

Author: Wilhelm Hudetz

Date: 08:52:23 11/04/05

Go up one level in this thread


On November 04, 2005 at 09:59:05, Per Jørgensen wrote:

>Hi Wilhelm!

HI Per!

>
>Thanks for your results. They are very interesting. The reason I'm writing you
>is that I am considering to buy an AMD Dual Core CPU (if you want to you can
>check what I wrote earlier today under the Subject "Which CPU should I buy?"). I
>can see that you are testing with an AMD Dual Core and I would be delighted if
>you would tell about your experience with this CPU :-) Here are my questions:
>
>1) What's your general opinion of your CPU? Are you satisfied?

best CPU I ever had! Makes Windows faster then ever when using standard-programs
(Winword, Excel, Corel...)because the second cpu makes background tasks.
By chess, as you say, you can use ponder on, what you give more precision in
test-series. And at least, I can run it at 2500mhz without any problem,
but the differenz to 2300mhz is small, so I let it there (49°C by eng-engmatch
with ponder on after 2 weeks permanent run).


>
>2) I can see you are testing with Pondern OFF. Why? You have two cores in your
>CPU and therefore you should be able to test with permanent brain ON.

If you read, the goal of this tournament was, how much can the Deep-programs
(using two Cpu´s)benefit from dual-power against the single one´s.
So if the Deep-versions use 2cpu you can not use ponder on, therefor you need
3cpu´s :-)


>
>3) I have another question concerning the AMD Dual Core CPU: Let's assume that I
>am having two engines (and not Deep-Versions!) playing against each other. Does
>it then make a difference for each engine if I am playing Pondern ON or OFF
>concerning the CPU Power?? I would say no because I'd guess that each engine is
>using one Core (again under the condition that it is not a Deep Version) and
>this is independent of Pondern ON/OFF.
>
>In order to make a detailed answer on question 3 it would be very Great if you
>would do the following test (I know that not only me but many Computerchessfans
>would be very interested in knowing the result of this test):
>
>Let's do the test with Shredder 9 and Fritz 9. Pick out a random position from
>the middlegame. We start with the Pondern OFF option. The time level doesn't
>matter a lot but choose something like 10 min. for the whole game. Now, start
>the game from this middlegameposition and observe the average number of
>positions the two engines are doing when they are thinking. It's hard to this
>precisely but it's not important. Then write down these average numbers after a
>couple of moves (just approximately, maybe 600.000 for Shredder 9 and more than
>a million for Fritz 9). Now, repeat the experiment with Pondern ON. How are the
>average numbers now? Are they approximately the same or are they lower? Exactly
>this question I find very, very interesting and important. I will be very, very
>delighted if you would do this small test and write me the results here in this
>Forum :-)

That is simple, the number of positions is only higher if there is a ponder-hit.
If the program ponders the false move it must also begin from 0pos. :-)

>
>Best regards and a very nice weekend for you
>
>Per



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.