Jumbo 0.6.96 64-bit Gauntlet for CCRL 40/40

[Michael Sherwin]

So Sven if you want me to add one more test I can since this is running on a 6 core processor. Just let me know.

Sounds interesting, thanks for your tests! If you actually have some spare resources left for another test then I would be interested in something that would allow to compare Jumbo 0.6.96 to 0.6.66 under identical conditions, e.g. "Jumbo 96 bb versus RomiXNoRL at 2+6 using the nooman 3 move 500 pgn 1000 games" (that is your test 1 but with 96).

Underway!

[Michael Sherwin]

Some preliminary results. After 166 games 2+6 Jumbo 66 vs RomiXNoRl 56% in Jumbo's favor. After 124 games 2+6 Jumbo 96 63% in Jumbo's favor. After 78 games 40/10 Jumbo 66 also 56%.

[Michael Sherwin]

The 100 game control match is finished. Jumbo 66 scored 56%.

[Michael Sherwin]

After 235 games 66 is still at 56% it climbs to 57% at times then falls back to 56% and sometimes to 55% but since it is at 56% most of the time I'm ending the match. Also after 195 games 96 is still at 63%. So I think that shows at least that 96 is stronger than 66 against Romi.

[Graham Banks]

Tirsa ran some further games for the 40/40 list.

Code: Select all

Jumbo 0.6.96 64-bit   - Anatoli 0.35k               14.0 - 10.0   +11/=6/-7    58.33%
Jumbo 0.6.96 64-bit   - Cerebro 3.03d               11.0 - 13.0    +7/=8/-9    45.83%
Jumbo 0.6.96 64-bit   - Coiled 0.6 64-bit            7.0 - 13.0    +4/=6/-10   35.00%
Jumbo 0.6.96 64-bit   - Drosophila 1.5.1 64-bit      8.0 - 20.0    +5/=6/-17   28.57%
Jumbo 0.6.96 64-bit   - Ifrit m1.8 64-bit           16.0 - 8.0    +14/=4/-6    66.67%
Jumbo 0.6.96 64-bit   - Kiwi 0.6d 64-bit             9.5 - 14.5    +8/=3/-13   39.58%
Jumbo 0.6.96 64-bit   - Lozza 1.18 64-bit           13.5 - 14.5    +7/=13/-8   48.21%
Jumbo 0.6.96 64-bit   - MadChess 2.2 64-bit         11.5 - 8.5    +11/=1/-8    57.50%
Jumbo 0.6.96 64-bit   - Minic 0.40 64-bit           12.0 - 16.0    +8/=8/-12   42.86%
Jumbo 0.6.96 64-bit   - Nawito 1812                 10.0 - 14.0    +5/=10/-9   41.67%
Jumbo 0.6.96 64-bit   - Nemeton 1.8                 11.0 - 9.0     +8/=6/-6    55.00%
Jumbo 0.6.96 64-bit   - Orion 0.5 64-bit            10.5 - 17.5    +6/=9/-13   37.50%
Jumbo 0.6.96 64-bit   - Popochin 4.1 64-bit         15.0 - 9.0    +12/=6/-6    62.50%
Jumbo 0.6.96 64-bit   - RomiChess P3n 64-bit        22.0 - 10.0   +16/=12/-4   68.75%
Jumbo 0.6.96 64-bit   - Taltos rev118 64-bit        13.0 - 11.0   +11/=4/-9    54.17%
Jumbo 0.6.96 64-bit   - Teki 2 64-bit               16.0 - 8.0    +14/=4/-6    66.67%
Jumbo 0.6.96 64-bit   - WChess 1.06                 31.5 - 8.5    +26/=11/-3   78.75%
Jumbo 0.6.96 64-bit   - Xpdnt 091007                14.0 - 10.0   +11/=6/-7    58.33%

[Michael Sherwin]

I was just about to start a test between Jbo 96 and Romi P3n to see what the difference is. So if P3n still comes in at about 31.75% that would mean RomiX is about 40 elo stronger. Well at least against Jbo

[tpoppins]

Tirsa ran some further games for the 40/40 list.

and the only reason they didn't make it into last update on Saturday is that at the last moment I discovered that three games vs. Romi had to be replayed due to abnormal terminations. It was the only engine to crash during the above test, repeatedly, apart from the old WChess that relies on WB2UCI to communicate with the GUI.

If Michael is accepting bug reports Romi could benefit from this thread as much as Jumbo. :>