Skip to content

@milostatarevic milostatarevic released this Nov 7, 2019

v0.6

tc=15+0.15
Score of Xiphos 0.6 SSE vs Xiphos 0.5 SSE: 5264 - 2346 - 12762  [0.572] 20372
Elo difference: 50.11 +/- 2.89
Assets 6

@milostatarevic milostatarevic released this Mar 11, 2019

v0.5

tc=20+0.2, hash=64MB
Score of Xiphos 0.5 SSE vs Xiphos 0.4 SSE: 2943 - 808 - 4661  [0.627] 8412
Elo difference: 90.15 +/- 4.89
Assets 6

@milostatarevic milostatarevic released this Sep 24, 2018

v0.4

tc=20+0.2, hash=64MB
Score of Xiphos 0.4 SSE vs Xiphos 0.3 SSE: 3663 - 1148 - 4055  [0.642] 8866
Elo difference: 101.34 +/- 5.34
Assets 6

@milostatarevic milostatarevic released this Jun 25, 2018

v0.3

tc=40/20, hash=128MB
Score of Xiphos 0.3 SSE vs Xiphos 0.2.2 SSE: 1320 - 423 - 2133  [0.616] 3876
Elo difference: 81.89 +/- 7.26
Assets 6

@milostatarevic milostatarevic released this Apr 20, 2018

About 45 elo stronger than v0.1 in self play (tc=40/90s).

Assets 6
Apr 14, 2018
Mar 22, 2018
v0.1.1
Summary:

- Fix a terrible incremental time control bug.
- Pull counter/killer moves separately (a minor speed boost).
- In quiescence search, use hash only for depth 0 or when in check.
- Simplify threats, use only in endgame.
- If possible, increase the time buffer.
- Don't build BMI2 version by default.

tc=10s+0.1s, hash=128MB
Score of Xiphos 0.1.1 SSE vs Xiphos 0.1 SSE: 3370 - 2198 - 6563  [0.548] 12131
Elo difference: 33.67 +/- 4.18

tc=40/4s, hash=128MB
Score of Xiphos 0.1.1 SSE vs Xiphos 0.1 SSE: 9045 - 8220 - 17191  [0.512] 34456
Elo difference: 8.32 +/- 2.59

tc=40/20s, hash=128MB
Score of Xiphos 0.1.1 SSE vs Xiphos 0.1 SSE: 1472 - 1221 - 3827  [0.519] 6520
Elo difference: 13.38 +/- 5.41