Identical specs are best, and necessary if you are looking for small changes in Elo, such as Bob is with Crafty. In your case, your engine is young and has much opportunity for gaining Elo. You could run the Crafty 19.17 benchmark ( here is a link to it : http://www.mediafire.com/?m25vbp8iy3qbrxd ) on each laptop. Use your chosen time control on the fastest computer. Then, use the chosen time control times the ratio of the benchmark on each laptop to the fastest laptop.stevemulligan wrote:If I have a bunch of old laptops, can I run a few games on those machines and then merge all the PGN files together to save time? Or do the specs on the machines need to be identical for that?
Just to be clear, let's say the chosen time control is 40/120+1 and the bechmark on the fastest laptop is 40. If the benchmark on another laptop is 50, then use the time control 40/144+1.2 on that laptop. Given the sort of gains you may experience at this point in your engine's development, you should be able to confirm positive changes, provided you play enough games. I guess that would be a function of how many laptops you have. Anyway, the CCRL lists consists of games played on various computers where the time controls are determined by the Crafty benchmark. It is not ideal, but it is better than using too few games to measure changes.
By the way, if your engine can handle it, you could decrease the time control you are using. The gain in the number of games you can accumulate will more than offset anything that might be missed by using the shorter time control. Most (but not all) gains found at short time controls do not vanish at long time controls.
