Zitat von
Tibono
Sorry I think you messed up something here. If computer A can only compute 333 positions while pondering, this assumes computer B (3x accelerated) pulled after 10 secs - this hands over the turn to play to comp. A and stops prematurely A's pondering phase. Yet over these 10 secs B computed "only" 1000 positions, not 3000.
Another way to say it is: if comp. B 3 times accelerated benefits from 3000 computed positions per move, while comp. A features only 1000 per 30 secs, this means comp. B requires 30s per move as well (to reach 3000 positions). Then comp. A can use 30s pondering while B computes...
Kind regards,
Eric
Hi Eric
Just focusing on Pondering I think I am right.
1) Computer B since Computer A gives it full 30 seconds gets the benefit of 3000 Positions
2) Computer A since it only receives 10 seconds for pondering now gets 333 positions instead of its 1000 positions.
Computer B with 3x setup has a 3:1 advantage if computer a gets its full 30 seconds to ponder which is the same 3:1 advantage when not pondering.
However since computer B only gives it 10 seconds the result is 333 or 1/3rd of what it should get. This means while pondering Computer B gets instead of a 3:1 benefit because the 3x speed. It now has 9:1 in their pondering modes and 3:1 when not pondering.
Computer B always stays at 3000 positions when pondering 30 seconds or calculating 30 seconds.
Computer A is 1000 by calculating 30 seconds and 333 (10 sec) when pondering that is 9:1 under ponder.
Hence bigger ponder mistakes because the search depth with 10 seconds makes bigger Ponder mistakes.
Regards
Nick
Oh wait you are right.. I forgot Computer B moves in 10 seconds so the 3000 becomes 1000 so it is still 3:1.
But still making the other computer do possible ponder moves at 3 ply is obviously worse than 5 ply so there is a bigger handicap for computer A.