how to measure kn/s?

Not logged inRybka Chess Community Forum

Forum RybkaChess Help Search Register Login

Topic Rybka Support & Discussion / Rybka Discussion / how to measure kn/s?

By dsgilchrist (*)

Date 2008-03-05 05:13

What's the proper way to measure kn/s with rybka 2.3? My stronger computer is at home, I'm on the road now on a laptop celeron M 1.3 504 meg ram. With rybka under infinite analysis I see 13 kn/s that seems low. I have a feeling I'm not testing it the same way you all are.

By Uly

Date 2008-03-05 06:29

My Rybka 2.3.2a on my 2.1Ghz computer is running at 30-40kn/s, I wonder if the correlation seems right (As Rybka is a low node counter.)

By BB

Date 2008-03-05 06:55

> I'm on the road now on a laptop celeron M 1.3 504 meg ram.

There is a Fritz benchmark of 938 Kn/s for a 1.4Ghz Celeron M360. For the Athlon XP 2600+ (2083Mhz it says), the benchmark is 989 Kn/s. There might be some 32/64-bit differencing for Rybka.

By dsgilchrist (*)

Date 2008-03-05 14:25

So on fritz 9 for example, I go to the engine pulldown menu then to chess benchmark and hit start? Its as easy as that?

By dsgilchrist (*)

Date 2008-03-05 14:34

I just did that and it siad relative speed 1.73 at 828 kn/s. How interesting it will be to compare the laptop (which I assumed was inferior) to my desktop P4 3.0 ghz with a meg of ram. If I read correctly in other posts, a p4 is getting under 500 kn/s. That would mean my laptop is better for analyzing games, is that correct?

Thanks,
Dave

By BB

Date 2008-03-05 14:46

> How interesting it will be to compare the laptop (which I assumed was inferior) to my desktop P4 3.0 ghz with a meg of ram.

The closest I can find on the Fritzmark page is a 3.2Ghz P4 at 1056 Kn/s for Fritz. Of course, the Fritzmark need not be directly applicable to relative Rybka speed/strength, but in this case I think it is reasonably true. So I'd think that the desktop P4 is slightly superior.

By JhorAVi (***)

Date 2008-03-05 15:58

But rybkas KNPS is not stable. I mean at varies with big difference in my Core2Duo WinXp-64bit. It's speed with infinite analysis of the starting position varies from 200+ knps to 100+ knps as depth progresses.

By Permanent Brain (*****)

Date 2008-03-05 17:24

No engine has always the same kN/s. Simply compare the total number of nodes after the same time in the same position(s).

By premraj_n (**)

Date 2008-03-06 06:49

There is a simple way, after each depth of analysis u will see time taken and total kN so with that convert the time to seconds and then divide total kN by time in seconds u will know actual kN/s (average).

By BigBen

Date 2008-03-06 16:39

Hi, My Quads KNS is always jumping up and down, I can look at the same position twice and get different KNS

http://valid.x86-secret.com/show_oc.php?id=318374

By Phil Harris (***)

Date 2008-03-06 16:59

Brave man putting that much voltage through a 45nm CPU! I am assuming you have at least 1.55V set in BIOS, or 1.6V if it's an unmodded Asus MB.

I decided to limit core volts to 1.45V for now, I seem to be stable at 4gHz and though I can boot at 4.2 it's a little flaky, which is slightly disappointing. I have a second QX arriving next week, so maybe that will be better.

As Nehalem approaches I will take a few more risks perhaps.

Phil

By BigBen

Date 2008-03-06 17:33

Hi,
Not brave at all lol, it works out even less with V-droop ... Its just a cheapy £60 Gigabyte board (wont touch Asus) ... The chip will do 5Ghz if I up the voltage to 1.65\1.7V ... fully loaded its around -40C at the evaperator .... I am going through a bored with computer chess stage lol preferring to go to the local chess club, you cant beat playing real human beings ... What is funny is the amount of TN I have accumulated and enjoy springing on players in the league :)

Regards

Topic Rybka Support & Discussion / Rybka Discussion / how to measure kn/s?