How fast are new computer systems with chess!

[jpqy]

Hi M Ansari,

I follow you for a long time,read everything on the different forums and many people get from you interesting information about computers and overcocking..all perfect..i also do nothing else then testing and overclocking but with a lower budget

Now,to come back to my comparing test..if you check my list well i have a Skulltrail Harpertown 2xE5430 @ 3.0Ghz and the next Skulltrail is a Skulltrail 2xQX9775 @ 4.0Ghz. So,no other Skulltrail at 3.0Ghz!!

To make it it for everybody clear,you have here a difference from 1Ghz!!
You are right when the two systems run at 3.0Ghz they will have almost equal results.

A little compare:
6.286.188 +0.25 Skulltrail 2xQX9775 @ 4.0Ghz
5.278.434 +0.26 Skulltrail Harpertown 2xE5430 @ 3.0Ghz

So,for 1Ghz higher clock speed you get about 1Milj. nodes.s more!!

5.493.977 +0,32 QX9650 @4.0Ghz DDR3: 7-7-7-18
4.173.420 +0,21 QX9650 @3.0Ghz DDR3: 7-7-7-18

Here again,this person did for me 6 tests,always the same but with 200Mhz higher clocks,start from 3.0Ghz till 4.0Ghz (3.0-3.2-3.4-3.6-3.8-4.0Ghz)

The difference is about 1.3Milj. nodes/s. higher on the same system with 1Ghz higher clocked!

Now for the Nehalem or Gainestown..just do the same..we take for 1Ghz higher clocks = +/- 1Milj.Nodes.s.
Then you see where they will come!!

What i like to see from you is,you let it run on the highest speed that for you is possible and one test with clock speed @2.93Ghz!!

Then you will have a nice picture what the speed difference will be.

For the test you just need Arena+TogaII142JD multi core
Start position - Infinite - and let it run till at least Depth 21

Else if you give me your e-mail adress,i send you the files with explenation!

Kind regards,
Jean-Paul.

[Albert Silver]

Hi Albert,

Thanks for your result...if i have your memory timings then your result is complete!

Maybe it's better i put my list also on this forum...

JP

Memory is a fairly standard GSkill PC-8500 at 5-5-5-15

[M ANSARI]

Hi Vael, send me the files and I will run them at 4.8 Ghz on the Skulltrail. It is on air now but I will put it on Vapor for this test. I did a lot of testing at 4.8 Ghz all the way up to 5.2Ghz. But for 24/7 full 100% load I would only consider 4.8 Ghz as stable enough. At 4.8 Ghz Zappa Mexico II was incredibly strong ... probably just as strong as Rybka 2.3.2a. But Rybka 3 changed all that and Rybka 3 gains as much as Zappa Mexico II with the increased hardware and keeps a good 100 to 120 ELO margin. I can't remember what the Fritzmark was at 4.8 Ghz but it was dramatically better than the default 3.2 Ghz.

[jpqy]

Hi,i have just send you a PM..

Fritz-benchmark link: http://www.jens.tauchclub-krems.at/dive ... marks.html

Maybe you can give also your result for this benchmark with your system!

JP.

[smirobth]

A friend of mine wants a new computer for chess. He would prefer 8 64-bit cores, but to also keep the cost under control. What would people recommend for a cost effective motherboard/CPU combo preferably well under $2K for the MB+CPUs?

[M ANSARI]

Intel Skulltrail D5400XS ($600)
2X Intel Xeon E5430 ($400 x 2 = $800)
4 x 1GB FBDIMMS PC6400 = $150
PS 1000 watt+ = $200
2 x Zalman 9700's = $120

This would be the core setup that I would not try to save money on. Everything else is not critical (HDD, Monitor, Case, etc.) this should give you to a very stable 3.2 Ghz on 8 cores. One thing I might also add is a small fan to the MCH since it gets quite hot and even if you are not overclocking the memory it is a good idea to have a fan there. Intel for some reason does not include a fan although it includes the bracket for it and recommends you put a fan there.

[smirobth]

This would be the core setup.....

Thanks!

[Albert Silver]

As an aside, it is worth mentioning the OS is highly relevant. I did 3 trial runs in XP 32-bit SP3, and three in Vista 64 SP1, and XP was consistently faster in all trial runs.

[Zach Wegner]

As an aside, it is worth mentioning the OS is highly relevant.

It is indeed. That's why you shouldn't use Windows at all!

[smirobth]

As an aside, it is worth mentioning the OS is highly relevant. I did 3 trial runs in XP 32-bit SP3, and three in Vista 64 SP1, and XP was consistently faster in all trial runs.

Faster at doing what and by how much? I'm sure 64 bit engines weren't faster on XP 32-bit.