The effective speedup from 1 to 8 cpus for SF and Komodo

[Arsha Mahdavi]

log2(1.7)=0.7655, assuming 1.7x speedup per doubling of cores.

I remember it (1.7x speedup) being mentioned in RybkaForum several times.

[Werewolf]

log2(1.7)=0.7655, assuming 1.7x speedup per doubling of cores.

I remember it (1.7x speedup) being mentioned in RybkaForum several times.

Nice one, yes it was logarithmic. The assumption then is not 0.76, but the 1.7x speedup.

[Adam Hair]

Adam,

Great stuff. I would love to see how this continues with more cores. It would take a lot of computer time.

Mark

I would love to see that data also. I am making more money at my new job, so I have had hopes of buying a server with more cores. Alas, my wife is quite proficient at making use of the extra money. If I could only make her understand that she is standing in the way of science

[Adam Hair]

Good work, thanks Adam!

Thanks Terry

[Adam Hair]

Thanks, very time and CPU consuming test.

The central value for Komodo 9 effective speedup is a bit larger than 4.5, having 9 Elo points difference. The doubling at 40/4' is worth about 70 Elo points, so in your case about 70-80 Elo points, and the effective speedup must be multiplied by 2^(9/70 to 80) for a total of ~4.9 for Komodo 9 on 8 cores. Therefore,
4.0 Stockfish 230315
4.9 Komodo 9
going from 1 to 8 cores, at blitz.
Error margins are harder to compute.

These are difficult to misinterpret, important numbers. As Stockfish shows hardly any widening, maybe in its case time to depth is the easier way to compute the effective speedup. Komodo does widen, so playing games is the only way.

Thanks Kai.

It seems that Vas was well aware of effective speedup before your discovery with Komodo:
http://www.rybkaforum.net/cgi-bin/rybka ... l?tid=3836

[Adam Hair]

8^0.76 = 4.86

very close to theoretical in Komodo's case.

What is the significance of the number 0.76 above? Where does it come from?

It's a number I've always used handed down by the Rybka team, LC in particular.

You didn't come across this when you were working with them?

If so I don't recall it now. It seems too simplistic, since each doubling is worth less than the one before.

Each doubling is worth less Elo, which is due (at least in part) to the higher draw rate associated with higher quality of play. However, the effective speedup/time odds may not diminish as much with increasing number of cores. It is possible that the time odds needed to make Rybka n cores equal to Rybka 2n cores is approximately constant for lower values of n.