Discussion of computer chess matches and engine tournaments.
Moderator: Ras
-
xr_a_y
- Posts: 1872
- Joined: Sat Nov 25, 2017 2:28 pm
- Location: France
Post
by xr_a_y »
Thanks for testing Minic !
(BTW, if your hardware is not AVX then 3.16 will add some more Elo. )
-
Rebel
- Posts: 7520
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Post
by Rebel »
xr_a_y wrote: ↑Mon Oct 18, 2021 6:16 pm
Thanks for testing Minic !
(BTW, if your hardware is not AVX then 3.16 will add some more Elo. )
I don't see 3.16 binaries on Github.
90% of coding is debugging, the other 10% is writing bugs.
-
Rebel
- Posts: 7520
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Post
by Rebel »
Thanks.
90% of coding is debugging, the other 10% is writing bugs.
-
Rebel
- Posts: 7520
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Post
by Rebel »
xr_a_y wrote: ↑Mon Oct 18, 2021 6:16 pm
Thanks for testing Minic !
(BTW, if your hardware is not AVX then 3.16 will add some more Elo. )
Okay, let's try.
1200 games.
Elo pool : 3168
http://rebel13.nl/b/grl.htm
90% of coding is debugging, the other 10% is writing bugs.
-
Rebel
- Posts: 7520
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Post
by Rebel »
Estimated elo gain for Minic 3.16 : +32
Code: Select all
Gambit Rating List
Running : Gauntlet Minic 3.16
Time Control : Time control : 40/120
Games : 1200
Results from file gauntlet-minic-316.pgn:
No. Name Win Draw Loss Unf. Score Games %
-------------------------------------------------------
1 Minic 3.16 +451 =421 -328 *0 661.5 1200 55.1%
2 Clover 2.4 +68 =71 -61 *0 103.5 200 51.8%
3 Seer 2.1.0 +54 =70 -76 *0 89.0 200 44.5%
4 Stash 31.0 +59 =59 -82 *0 88.5 200 44.2%
5 Wasp 4.50 +51 =74 -75 *0 88.0 200 44.0%
6 Weiss 1.4 +52 =72 -76 *0 88.0 200 44.0%
7 Halogen 10 +44 =75 -81 *0 81.5 200 40.8%
Total Games: 1200
White Wins: 398 (33.2%)
Black Wins: 381 (31.8%)
Draws: 421 (35.1%)
Unfinished: 0 (0.0%)
Estimated elo gain for Minic_3.16
Elo pool : 3168
Minic 3.14 : 3167.0
Minic_3.16 : 3199.0
Difference : 32.0
Out of curiosity,
1. Did you have to write special code for non AVX2 systems?
2. Or is (was) it a matter of setting the right compiler options?
Anyway, congrats with the progress, almost 3200 elo

90% of coding is debugging, the other 10% is writing bugs.
-
xr_a_y
- Posts: 1872
- Joined: Sat Nov 25, 2017 2:28 pm
- Location: France
Post
by xr_a_y »
Rebel wrote: ↑Wed Oct 20, 2021 9:57 am
Estimated elo gain for Minic 3.16 : +32
Code: Select all
Gambit Rating List
Running : Gauntlet Minic 3.16
Time Control : Time control : 40/120
Games : 1200
Results from file gauntlet-minic-316.pgn:
No. Name Win Draw Loss Unf. Score Games %
-------------------------------------------------------
1 Minic 3.16 +451 =421 -328 *0 661.5 1200 55.1%
2 Clover 2.4 +68 =71 -61 *0 103.5 200 51.8%
3 Seer 2.1.0 +54 =70 -76 *0 89.0 200 44.5%
4 Stash 31.0 +59 =59 -82 *0 88.5 200 44.2%
5 Wasp 4.50 +51 =74 -75 *0 88.0 200 44.0%
6 Weiss 1.4 +52 =72 -76 *0 88.0 200 44.0%
7 Halogen 10 +44 =75 -81 *0 81.5 200 40.8%
Total Games: 1200
White Wins: 398 (33.2%)
Black Wins: 381 (31.8%)
Draws: 421 (35.1%)
Unfinished: 0 (0.0%)
Estimated elo gain for Minic_3.16
Elo pool : 3168
Minic 3.14 : 3167.0
Minic_3.16 : 3199.0
Difference : 32.0
Out of curiosity,
1. Did you have to write special code for non AVX2 systems?
2. Or is (was) it a matter of setting the right compiler options?
Anyway, congrats with the progress, almost 3200 elo
This is new special simd code.
Relying on compiler was already introduced in 3.13 (remember the quite huge gap at the time).