SMP at fixed depth revisited

[Laskos]

After testing SF TTD on 4 and 8 threads with Joona's patch today, and getting bad results compared to earlier SF, I tested on widening, and surprise, the new SF widens a bit from 1 to 4 threads on fixed depth:


Not widening:

SF 11042014 (earlier SF):
H0=0
H1=10
Score of SF 4 threads vs SF 1 thread: 437 - 442 - 1455 [0.499] 2334
ELO difference: -1
SPRT: H0 was accepted
Finished match



Widening:

SF 07052014 (Joona patch):
H0=0
H1=10
Score of SF 4 threads vs SF 1 thread: 317 - 256 - 1035 [0.52] 1608
ELO difference: 13
SPRT: H1 was accepted
Finished match

So, now TTD is not completely relevant when testing the new SF SMP efficiency.