Hard-Talkchess-2020 set, final release

[AlexChess]

Hard-Talkchess-2022 (Amchess) ProteusSF-Piranha 230822 results:

Correct/Total:
ProteusSF-P 230822: 174/260 (67%)

Failed tests (hit *):
1. ProteusSF-P 230822:
1, 6, 8, 9, 14, 17, 19, 20, 21, 23, 28, 31, 32, 33, 34, 39, 40, 43, 48, 49, 51, 55, 58, 64, 65, 68, 74, 75, 76, 86, 92, 98, 100, 108, 117, 118, 122, 123, 127, 137, 140, 152, 153, 155, 156, 157, 159, 160, 161, 165, 166, 174, 178, 181, 182, 184, 185, 191, 195, 196, 201, 209, 213, 214, 215, 216, 221, 222, 223, 225, 226, 227, 229, 230, 232, 235, 236, 237, 238, 239, 243, 244, 245, 246, 247, 248

Successful tests:
1. ProteusSF-P 230822:
0, 2, 3, 4, 5, 7, 10, 11, 12, 13, 15, 16, 18, 22, 24, 25, 26, 27, 29, 30, 35, 36, 37, 38, 41, 42, 44, 45, 46, 47, 50, 52, 53, 54, 56, 57, 59, 60, 61, 62, 63, 66, 67, 69, 70, 71, 72, 73, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 91, 93, 94, 95, 96, 97, 99, 101, 102, 103, 104, 105, 106, 107, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 124, 125, 126, 128, 129, 130, 131, 132, 133, 134, 135, 136, 138, 139, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 158, 162, 163, 164, 167, 168, 169, 170, 171, 172, 173, 175, 176, 177, 179, 180, 183, 186, 187, 188, 189, 190, 192, 193, 194, 197, 198, 199, 200, 202, 203, 204, 205, 206, 207, 208, 210, 211, 212, 217, 218, 219, 220, 224, 228, 231, 233, 234, 240, 241, 242, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259

[peter]

Hard-Talkchess-2022 (Amchess) ProteusSF-Piranha 230822 results:

Correct/Total:
ProteusSF-P 230822: 174/260 (67%)

Failed tests (hit *):
1. ProteusSF-P 230822:
1, 6, 8, 9, 14, 17, 19, 20, 21, 23, 28, 31, 32, 33, 34, 39, 40, 43, 48, 49, 51, 55, 58, 64, 65, 68, 74, 75, 76, 86, 92, 98, 100, 108, 117, 118, 122, 123, 127, 137, 140, 152, 153, 155, 156, 157, 159, 160, 161, 165, 166, 174, 178, 181, 182, 184, 185, 191, 195, 196, 201, 209, 213, 214, 215, 216, 221, 222, 223, 225, 226, 227, 229, 230, 232, 235, 236, 237, 238, 239, 243, 244, 245, 246, 247, 248

Successful tests:
1. ProteusSF-P 230822:
0, 2, 3, 4, 5, 7, 10, 11, 12, 13, 15, 16, 18, 22, 24, 25, 26, 27, 29, 30, 35, 36, 37, 38, 41, 42, 44, 45, 46, 47, 50, 52, 53, 54, 56, 57, 59, 60, 61, 62, 63, 66, 67, 69, 70, 71, 72, 73, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 91, 93, 94, 95, 96, 97, 99, 101, 102, 103, 104, 105, 106, 107, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 124, 125, 126, 128, 129, 130, 131, 132, 133, 134, 135, 136, 138, 139, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 158, 162, 163, 164, 167, 168, 169, 170, 171, 172, 173, 175, 176, 177, 179, 180, 183, 186, 187, 188, 189, 190, 192, 193, 194, 197, 198, 199, 200, 202, 203, 204, 205, 206, 207, 208, 210, 211, 212, 217, 218, 219, 220, 224, 228, 231, 233, 234, 240, 241, 242, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259

Thanks for your results, but I have stopped checking Andrea's suite position by position seeing almost all hard ones already included in at least one of my favorite suites too and there are too many too easy and or too little distinct ones as for single- solution only.

Will rather go on with my trials of forming positional suite of its own with not only single best move positions included and let with different hardware- TC and different ways of evaluation of equivalent solutions run different suites then evaluate by EloStatTS in one result- .pgn to get one ranking- and rating- list out of those different single- suite- results together at the end.

That will take much more time till I have new kind of results once in a while, positional suite of consistent difficulty of positions is even biggger selection- problem and needs much more positions to get same statistical discrimination compared to error bar as with tactical single best move suites only, but I don't see any way to run positional test positions (almost by definition of more then one single best solution to be judged as such most of the times and positions) together with difficult tactical single best move- positions in one suite with one kind for evaluation and one hardware- TC only.
Just my two cents again, regards

[AlexChess]

Hard-Talkchess-2022 (Amchess) ProteusSF-Piranha 230822 results:

Correct/Total:
ProteusSF-P 230822: 174/260 (67%)

Failed tests (hit *):
1. ProteusSF-P 230822:
1, 6, 8, 9, 14, 17, 19, 20, 21, 23, 28, 31, 32, 33, 34, 39, 40, 43, 48, 49, 51, 55, 58, 64, 65, 68, 74, 75, 76, 86, 92, 98, 100, 108, 117, 118, 122, 123, 127, 137, 140, 152, 153, 155, 156, 157, 159, 160, 161, 165, 166, 174, 178, 181, 182, 184, 185, 191, 195, 196, 201, 209, 213, 214, 215, 216, 221, 222, 223, 225, 226, 227, 229, 230, 232, 235, 236, 237, 238, 239, 243, 244, 245, 246, 247, 248

Successful tests:
1. ProteusSF-P 230822:
0, 2, 3, 4, 5, 7, 10, 11, 12, 13, 15, 16, 18, 22, 24, 25, 26, 27, 29, 30, 35, 36, 37, 38, 41, 42, 44, 45, 46, 47, 50, 52, 53, 54, 56, 57, 59, 60, 61, 62, 63, 66, 67, 69, 70, 71, 72, 73, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 91, 93, 94, 95, 96, 97, 99, 101, 102, 103, 104, 105, 106, 107, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 124, 125, 126, 128, 129, 130, 131, 132, 133, 134, 135, 136, 138, 139, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 158, 162, 163, 164, 167, 168, 169, 170, 171, 172, 173, 175, 176, 177, 179, 180, 183, 186, 187, 188, 189, 190, 192, 193, 194, 197, 198, 199, 200, 202, 203, 204, 205, 206, 207, 208, 210, 211, 212, 217, 218, 219, 220, 224, 228, 231, 233, 234, 240, 241, 242, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259

Thanks for your results, but I have stopped checking Andrea's suite position by position seeing almost all hard ones already included in at least one of my favorite suites too and there are too many too easy and or too little distinct ones as for single- solution only.

Will rather go on with my trials of forming positional suite of its own with not only single best move positions included and let with different hardware- TC and different ways of evaluation of equivalent solutions run different suites then evaluate by EloStatTS in one result- .pgn to get one ranking- and rating- list out of those different single- suite- results together at the end.

That will take much more time till I have new kind of results once in a while, positional suite of consistent difficulty of positions is even biggger selection- problem and needs much more positions to get same statistical discrimination compared to error bar as with tactical single best move suites only, but I don't see any way to run positional test positions (almost by definition of more then one single best solution to be judged as such most of the times and positions) together with difficult tactical single best move- positions in one suite with one kind for evaluation and one hardware- TC only.
Just my two cents again, regards

OK, I repeat the test with your Hard-talkchess 2020 108 + 20 MZ test suite https://banksiagui.com/forums/viewtopic.php?t=65

[peter]

Hard-Talkchess-2022 (Amchess) ProteusSF-Piranha 230822 results:

Correct/Total:
ProteusSF-P 230822: 174/260 (67%)

Failed tests (hit *):
1. ProteusSF-P 230822:
1, 6, 8, 9, 14, 17, 19, 20, 21, 23, 28, 31, 32, 33, 34, 39, 40, 43, 48, 49, 51, 55, 58, 64, 65, 68, 74, 75, 76, 86, 92, 98, 100, 108, 117, 118, 122, 123, 127, 137, 140, 152, 153, 155, 156, 157, 159, 160, 161, 165, 166, 174, 178, 181, 182, 184, 185, 191, 195, 196, 201, 209, 213, 214, 215, 216, 221, 222, 223, 225, 226, 227, 229, 230, 232, 235, 236, 237, 238, 239, 243, 244, 245, 246, 247, 248

Successful tests:
1. ProteusSF-P 230822:
0, 2, 3, 4, 5, 7, 10, 11, 12, 13, 15, 16, 18, 22, 24, 25, 26, 27, 29, 30, 35, 36, 37, 38, 41, 42, 44, 45, 46, 47, 50, 52, 53, 54, 56, 57, 59, 60, 61, 62, 63, 66, 67, 69, 70, 71, 72, 73, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 91, 93, 94, 95, 96, 97, 99, 101, 102, 103, 104, 105, 106, 107, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 124, 125, 126, 128, 129, 130, 131, 132, 133, 134, 135, 136, 138, 139, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 158, 162, 163, 164, 167, 168, 169, 170, 171, 172, 173, 175, 176, 177, 179, 180, 183, 186, 187, 188, 189, 190, 192, 193, 194, 197, 198, 199, 200, 202, 203, 204, 205, 206, 207, 208, 210, 211, 212, 217, 218, 219, 220, 224, 228, 231, 233, 234, 240, 241, 242, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259

Thanks for your results, but I have stopped checking Andrea's suite position by position seeing almost all hard ones already included in at least one of my favorite suites too and there are too many too easy and or too little distinct ones as for single- solution only.

Will rather go on with my trials of forming positional suite of its own with not only single best move positions included and let with different hardware- TC and different ways of evaluation of equivalent solutions run different suites then evaluate by EloStatTS in one result- .pgn to get one ranking- and rating- list out of those different single- suite- results together at the end.

That will take much more time till I have new kind of results once in a while, positional suite of consistent difficulty of positions is even biggger selection- problem and needs much more positions to get same statistical discrimination compared to error bar as with tactical single best move suites only, but I don't see any way to run positional test positions (almost by definition of more then one single best solution to be judged as such most of the times and positions) together with difficult tactical single best move- positions in one suite with one kind for evaluation and one hardware- TC only.
Just my two cents again, regards

OK, I repeat the test with your Hard-talkchess 2020 108 + 20 MZ test suite https://banksiagui.com/forums/viewtopic.php?t=65

The latest 128- collection (more difficulty of positions on average) you find here:


forum3/viewtopic.php?p=932939&sid=b8319 ... 10#p932939

thus you could compare directly to the latest runs and EloStatTS- evaluation given there too.

But you can use the version of the 128 positions MZ used to use then together with me too (from the link given by you to his posting in Banksia- forum), didn't download to check but I'd be rather sure it's the same suite he got from me way back then.

BTW, Marco used EloStatTS there too, which works with Fritz only, as for my personal pov I also don't like Banksia's position-testing-feature so much, because of it's unusual time management, especially I doubt the bug-free working of "extra plies" to be defined with time control/position. And don't use this "time per suite"- feature, that's just nonsense in my eyes, time management of an engine is programmed for game playing, not for test suites, how can the engine "know" how much time to use more or less for single position always started all over again with functionally empty hash as for changed position? Time per game is something quite different, there's no use in using something like that for test suites).

Just my two cents again

[AlexChess]

Hard-Talkchess-2022 (Amchess) ProteusSF-Piranha 230822 results:

Correct/Total:
ProteusSF-P 230822: 174/260 (67%)

Failed tests (hit *):
1. ProteusSF-P 230822:
1, 6, 8, 9, 14, 17, 19, 20, 21, 23, 28, 31, 32, 33, 34, 39, 40, 43, 48, 49, 51, 55, 58, 64, 65, 68, 74, 75, 76, 86, 92, 98, 100, 108, 117, 118, 122, 123, 127, 137, 140, 152, 153, 155, 156, 157, 159, 160, 161, 165, 166, 174, 178, 181, 182, 184, 185, 191, 195, 196, 201, 209, 213, 214, 215, 216, 221, 222, 223, 225, 226, 227, 229, 230, 232, 235, 236, 237, 238, 239, 243, 244, 245, 246, 247, 248

Successful tests:
1. ProteusSF-P 230822:
0, 2, 3, 4, 5, 7, 10, 11, 12, 13, 15, 16, 18, 22, 24, 25, 26, 27, 29, 30, 35, 36, 37, 38, 41, 42, 44, 45, 46, 47, 50, 52, 53, 54, 56, 57, 59, 60, 61, 62, 63, 66, 67, 69, 70, 71, 72, 73, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 91, 93, 94, 95, 96, 97, 99, 101, 102, 103, 104, 105, 106, 107, 109, 110, 111, 112, 113, 114, 115, 116, 119, 120, 121, 124, 125, 126, 128, 129, 130, 131, 132, 133, 134, 135, 136, 138, 139, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 154, 158, 162, 163, 164, 167, 168, 169, 170, 171, 172, 173, 175, 176, 177, 179, 180, 183, 186, 187, 188, 189, 190, 192, 193, 194, 197, 198, 199, 200, 202, 203, 204, 205, 206, 207, 208, 210, 211, 212, 217, 218, 219, 220, 224, 228, 231, 233, 234, 240, 241, 242, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259

Thanks for your results, but I have stopped checking Andrea's suite position by position seeing almost all hard ones already included in at least one of my favorite suites too and there are too many too easy and or too little distinct ones as for single- solution only.

Will rather go on with my trials of forming positional suite of its own with not only single best move positions included and let with different hardware- TC and different ways of evaluation of equivalent solutions run different suites then evaluate by EloStatTS in one result- .pgn to get one ranking- and rating- list out of those different single- suite- results together at the end.

That will take much more time till I have new kind of results once in a while, positional suite of consistent difficulty of positions is even biggger selection- problem and needs much more positions to get same statistical discrimination compared to error bar as with tactical single best move suites only, but I don't see any way to run positional test positions (almost by definition of more then one single best solution to be judged as such most of the times and positions) together with difficult tactical single best move- positions in one suite with one kind for evaluation and one hardware- TC only.
Just my two cents again, regards

OK, I repeat the test with your Hard-talkchess 2020 108 + 20 MZ test suite https://banksiagui.com/forums/viewtopic.php?t=65

The latest 128- collection (more difficulty of positions on average) you find here:


forum3/viewtopic.php?p=932939&sid=b8319 ... 10#p932939

thus you could compare directly to the latest runs and EloStatTS- evaluation given there too.

But you can use the version of the 128 positions MZ used to use then together with me too (from the link given by you to his posting in Banksia- forum), didn't download to check but I'd be rather sure it's the same suite he got from me way back then.

BTW, Marco used EloStatTS there too, which works with Fritz only, as for my personal pov I also don't like Banksia's position-testing-feature so much, because of it's unusual time management, especially I doubt the bug-free working of "extra plies" to be defined with time control/position. And don't use this "time per suite"- feature, that's just nonsense in my eyes, time management of an engine is programmed for game playing, not for test suites, how can the engine "know" how much time to use more or less for single position always started all over again with functionally empty hash as for changed position? Time per game is something quite different, there's no use in using something like that for test suites).

Just my two cents again

Thank you very much!
I restart the test right now with your revised positions. You could contact Pham and suggest your improvements. it's very kind and interested to users feedbacks. forum3/viewtopic.php?f=2&t=72350&sid=ca ... 827ef55b38
By the way these are MZ test results:

Correct/Total:
ProteusSF-P 230822: 95/114 (83%)


Failed tests (hit *):
1. ProteusSF-P 230822:
Hard-Talkchess-2020.007, Hard-Talkchess-2020.008, Hard-Talkchess-2020.011, Hard-Talkchess-2020.019, Hard-Talkchess-2020.034, Hard-Talkchess-2020.038, Hard-Talkchess-2020.039, Hard-Talkchess-2020.053, Hard-Talkchess-2020.054, Hard-Talkchess-2020.069, Hard-Talkchess-2020.070, Hard-Talkchess-2020.074, Hard-Talkchess-2020.087, Hard-Talkchess-2020.091, Hard-Talkchess-2020.095, Hard-Talkchess-2020.099, Hard-Talkchess-2020.108
Successful tests:
1. ProteusSF-P 230822:
Hard-Talkchess-2020.001, Hard-Talkchess-2020.002, Hard-Talkchess-2020.003, Hard-Talkchess-2020.010, Hard-Talkchess-2020.012, Hard-Talkchess-2020.013, Hard-Talkchess-2020.014, Hard-Talkchess-2020.016, Hard-Talkchess-2020.020, Hard-Talkchess-2020.021, Hard-Talkchess-2020.023, Hard-Talkchess-2020.028, Hard-Talkchess-2020.029, Hard-Talkchess-2020.031, Hard-Talkchess-2020.035, Hard-Talkchess-2020.036, Hard-Talkchess-2020.043, Hard-Talkchess-2020.046, Hard-Talkchess-2020.047, Hard-Talkchess-2020.049, Hard-Talkchess-2020.050, Hard-Talkchess-2020.052, Hard-Talkchess-2020.056, Hard-Talkchess-2020.058, Hard-Talkchess-2020.059, Hard-Talkchess-2020.061, Hard-Talkchess-2020.065, Hard-Talkchess-2020.066, Hard-Talkchess-2020.067, Hard-Talkchess-2020.068, Hard-Talkchess-2020.072, Hard-Talkchess-2020.078, Hard-Talkchess-2020.081, Hard-Talkchess-2020.083, Hard-Talkchess-2020.089, Hard-Talkchess-2020.090, Hard-Talkchess-2020.093, Hard-Talkchess-2020.094, Hard-Talkchess-2020.096, Hard-Talkchess-2020.097, Hard-Talkchess-2020.101, Hard-Talkchess-2020.103, Hard-Talkchess-2020.105, Hard-Talkchess-2020.109, Hard-Talkchess-2020.110, Hard-Talkchess-2020.113, Hard-Talkchess-2020.114

[peter]

I restart the test right now with your revised positions. You could contact Pham and suggest your improvements. it's very kind and interested to users feedbacks. forum3/viewtopic.php?f=2&t=72350&sid=ca ... 827ef55b38

Banksia GUI is a great project in very active development.
As for the position- testing- feature, I guess, Pham knows all the problems about especially the point of extra plies already, my problem with the feature of time per suite is a principally one:

If engine is given number of moves to ponder about, it will take as much time for the single one as time to depth and move count suggests, but this one engine's time management is made for game playing, not for the kind of hash aging that's done in test suite- mode.

Thus the positions the engine evaluates most quickly as for move order of candidates will get judged "solved" sooner than the ones with more almost equal candidates, engine will go the "easiest" way with going to next position instead of holding the solution up to at least the number of extraplies given by GUI, that's quite the opposite concept of the one of extraplies and time per position to let GUI adjudicate output as stable or by strictly as upper limit defined time over- rule.
So you can't compare such resuts to each other for one more relevant parameter. You can use the feature for comparisons of results of their own again of course if you like those results more, some engines will profit, others will get punished, it's the same as for chosing different suites, hardware- TCS and opponents to compare to each other. Some ways of testing will give more discrimination but as well as with suite with less solutions per engine and hardware- time, more discrimination by less solutions will mean more statsitical error.

Only way for me to get different results from different suites, hardware- TCs and opponents into one comparable ranking and rating is done by EloStatTS, there you can let the program evaulate some result- .pgns finally togehter out of a conjoint one.
EloStatTS works with Fritz- GUI only anyhow, so at the moment I don't invest much time in working with other GUIs and that's why I can't give much heip and advisory as for development of other GUIs, as for result- documentation I like Shredder much too e.g., very fine table of solutions of a single run together with time- count per position, and time management works very well with "Exact Time"- feature chosen for TC, that's another one point I doubt to be bug- free in Banksia (ver 0.55 beta) so far too, chosing time per position and extra plies (even if not set just to maximum, which would be the most "secure" way to deny too much accidential solutions) seems to me to let engine ponder longer sometimes then upper limit of TC should grant.
The concept of letting engine have more then one "laps" also is one, that for sure would work only for quite uncomparable results of their own as for hardware- TC and engine- pool. The hash- residues from one lap to next one if not fully accidentially between different runs at all, at least would depend mainly on the very different kind of dealing with aging of hash up to full clearing of hash with each new position of single run, even more with repeating the unsolved ones in next lap.
Maybe of some interest for certain engines and their kind of hash clearing done by GUI too, but that's not the interest you have in results of test suites as one's used to have.
E.g. Dragon would have some kind of disadvantage with more laps compared to SF I'd say, LC0 and engines of its kind (NN, puct instead of A-B- search) having disadvantages and advantages of their own with NN- cache instead of hash- usage, not to be judged by user at all without watching engine's output all the time of all postions run anyhow

Just one more single cent of my very own again, don't want to tell users how to get test- results of their own interest to compare to same way of testing of their own too, problem (as much as with game- playing- results) is to compare them with the ones of other users and testers.

[AlexChess]

The latest 128- collection (more difficulty of positions on average) you find here:


forum3/viewtopic.php?p=932939&sid=b8319 ... 10#p932939

thus you could compare directly to the latest runs and EloStatTS- evaluation given there too.

ProteusSF-Piranha (Stockfish derivative) -> http://www.microsmeta.com/DBLOG/articol ... colo=1525
File name : Peter.epd
Total test items : 128
Test for : best moves
Total engines : 1
Expand ply : 1
Laps : 1
Total tests : 128
Total corrects : 81 (63%)
Ave correct elapse : 14:10
Status : completed

2 CPUs Mac mini 2010 (!) Core 2 Duo 400 kN/s (not exactly a Threadripper )
How many positions calculates Vinvin PC with 30 threads of a 16x3.5GHz CPU and a 3070ti GPU?

15"/position versus 15 minutes /position

Correct/Total:
ProteusSF-P 230822: 81/128

Failed tests (hit *):
1. ProteusSF-P 230822:
TacticalInsanityNr.9_CorbitD_HTC108-5 S, 2, Carlsen, M._Caruana, F._HTC108-7 WM2018, 3, Gusia_Barlov_Schweiz 1980 HTC108-28, 12, Rudolph, W.E._?_Studie 1912 HTC108-30 S, 14, Van Essen, M._2004, #22._HTC108-48 (-9ZNr41) S, 16, Neghina, M._Chase, Derived_HTC108-54, 19, Morozevich, A._Vitiugov, N._Reggio 2011 HTC108-100, 22, ACT3, ??._?_?, 36, ACT3, ??._Festung_?, 37, ACT3, ??._Festung_S, 38, ACT3, ??._Studie?_?, 39, ACT3, ??._Studie?_S, 40, Neghina, M._WhieTo, MoveAndWin_2018? S, 43, Popov, Study_Glarean 2006, 1996?._HTC108-58, 44, Sikorsky, Horst_Novak, Joze_Sideline, 45, Krug, P._CCC, Paloma_S, 46, Sikorsky, Horst_Vinchev, Simeon_corr GER, Hiarcs, 49, Study, ?._?_IQ 100, 51, Study?, ?._?_ACT3, 52, Study?, ?._?_TTT2 31, 53, Study, ?._CCC_ACT5 11, 56, Study, ?._?_ACT5 29, 57, ERET 032_Zugzwang_Fahrni 1922, 59, ERET 046_Endspiel T&S vs T&S_Grandelius-Raznikov, Albena 2011, 61, ERET 090_Endspiel L vs L_Salai 2011, 63, ERET 093_Endspiel T&S&B vs T&T&L_Simkhovich 1923, 64, Fortress, Study?_?_ACT3-51 S, 66, Cyclic, ZZ?._?_ACT5-25 2, 68, Kos, V._?_ACT5-34 2, 73, Zinar, M._HHdbVI.0210037.002a., 09076?._ACT5-44 2, 76, Bernhardt, G._?_ACT5-58 2, 81, Unknows, Author_?_ACT5-61 2, 82, SilhanJ, ZizkaL_?_ACT5-59 3, 86, TTT2.053_?_TTT2, 91, Smyslov, V._?_ACT4-41 1, 95, Wotawa, A._?_ACT4-66 1, 98, Blathy, O._?_ACT4-94 1, 101, Cortlever, N._HHdbVI_ACT3-29 1, 102, Korolkov, V._?_ACT3-38, 103, Fortress, ?._?_ACT3-41 1, 104, Kraemer, A._?_ACT3-49, 106, Sivkov, N._#10, ?._ACT3-88, 107, Bazlo, Yuri_10moves, less_2019 HHdbVI 1, 112, TCEC, rj329_?_CCC Nr.5, 115, Yarmonov, I._Glarean_2020, 117, Krug, P.._Glarean, EngineCrackers2_2018, 120, Illescas, Miguel_Glarean, EngineCrackers_?, 122

Successful tests:
1. ProteusSF-P 230822:
HTC108, 2._Corbit, Dann_Tactical Insanity Nr.1, 1, ACT-Suite, Nr.12_Dorsz, OpenChess_HTC108-8, 4, van Foreest, J._L'Ami, E._HTC108-9 Zug 24...Sideline, 5, Konig?, CCC._Cornforth96?, CSS+Playchess_HTC108-11 Arasan, 6, PoorFish, CostalbaM_Borderline, Dual_HTC108-12, 7, LC0, CCC._SF, CSS._HTC108-15 S, 8, AlphaZero, HTC108-21._SF_Game9 2017, 9, Gashimov, V._Azariov, S._2011 HTC108-23, 10, CorbitD, CCC+Rybkaf_?_HTC108-25, 11, Bondarenko, &Kuznezov_Glarean, PoorStockfish_HTC108-29, 13, Gusev, Y._Auerbach, E.25._HTC108-42, 15, Neghina, M._Stronghold, 2003?._HTC108-49 S, 17, Tunik_Duzhakov_2011 HTC108-52, 18, SolomonKing, ?CCC._2.g4, Hiarcs_HTC108-80 S, 20, Nicolic, E._Fischer, R._HTC108-94 S, 21, M. Garcia - P. Krug, Studie 2015_Glarean_The Engine Crackers, 23, Zepler, E.E._Studie, 1928._?, 24, Formela, Michal_Studie, 2019._Polski Zwiazek Szachowy, 25, Stoofvlees_SF_TCEC 2019, 26, Lasker, Emanuel_Lipke, Paul_Nebenvariante 41.Ke4! MS, 27, Los, Andrii_Kubicki, Tadeusz_corr LIPEAD40 (PER), 28, PurePower_Exxon_Chessbase online, 29, Bauer, Robert_Zhak, Boris Mikhailovich_WC30/final, 30, Mihai Neghina_Desperation_White to move and draw S, 31, Tauber, Hans_Sikorsky, Horst_37.e5!, 32, Bondarenko+Gorgiev, Studie1959_BrillanterSchachzug, 17._?, 33, ACT3, ??._Festung_S, 34, ACT3, ??._Studie?_?, 35, Neghina, M._GoldenerK�fig, 2009._S, 41, Ree, H._Hort, V._Wijk 1986, 42, TCEC, Sammlung_Nr.8, j329_CCC, 47, TCEC, Sammlung_Nr.22, rj329_CCC, 48, Zemljanski, Y.I._1move, cut_Tsjechover-90 S, 50, Semenikov+, D A._Sikorsky, H._corr. USA 2021, 54, Petrov, M._Sikorsky, H._CL/2021, 55, ERET 015_Endspiel D&B vs T&L&B_Becker 2015, 58, ERET 037_Zugzwang_Garcia 2000, 60, ERET 050_Koenigsangriff_Hansen-Barua, Biel 1993, 62, ERET 094_Damenopfer_Sirobaba-Chaika, CorrGame 2011, 65, ACT4, Nr.1_?_ACT4-1, 67, Blass, U.._CCC_ACT5-27 2, 69, Campioli, M._#14, ?._ACT5-28 2, 70, Krug, P._?_ACT5-30 2, 71, KrugP, Garcia_HHdbVI.0044383.001, a.01968?_ACT5-31 1, 72, Sokka, H._HHdbVI.1076075.001, a.53387?_ACT5-36 1, 74, Sehwers, J?._HHdbVI.1643584.002a., 89051?._ACT5-ana, 75, Nestorescu, V._HHdbVI.0708805.002a., 33432?._ACT5-46 1, 77, Timman, J._HHdbVI.0033988.002a., 01480._ACT5-48 2, 78, Unknown, Author_?_ACT5-51 2, 79, Ivanov, V._?_ACT5-53 2, 80, Naef, W._?_ACT5-63 2, 83, #7, ?._?_ACT5-67 2, 84, #9, ?._?_ACT5-68 2, 85, Hiarcs, ?._?_TTT2 2, 87, Hiarcs, ?._?_TTT2 6 2, 88, Derived, from_StavrietskyA, 2017._TTT2 14, 89, Midgame, ?._?._TTT2 23, 90, TTT2.073_?_TTT2, 92, RoxlauM, GrushkoM_HHdbVI_ACT4 4 1, 93, Smyslov, V._?_ACT4-26 1, 94, JostenG_?_?, 96, Study, ?._?_ACT4-59, 97, Study, ?._?_ACT4-90 1, 99, Bethge, F._?_ACT4-91, 100, Unknown, Author_?_ACT3-48, 105, Timman=J_Glarean, Nightmare_?, 108, Boudy Bueno, J.L.._Szabo, L._Varna 1979 Txb2!? Eret 17, 109, Lc0-dev_Stockfish_ECO MegaMatch II! (1|1), 110, Carlsen, M._L'Ami, E._Corus B VLH16 35, 111, Opening, Hiarcs_!, ?!._?, 113, TCEC, rj329_?_CCC, 114, Gleichmann, Matthias_Praznik, Niko_corr CT21/final, 116, Eman_SugaR_Glarean, 118, Hobacle's, VersionOf_Walter Eigenmann's, Study_?, 119, Gniazdowski, Marian_Rau, Johann_corr CL/2019/C8 Glarean, 121, Heinz, J._Glarean_Study 1981, 123, Josten, G.._Glarean, EngineCrackers_Studie 2001, 124, Timman, J.._HHdbVI_?, 125, Gelly, M.._HHdbVI_Glarean, 126, Krug, P._HHdbVI, Glarean_?, 127, Ganguly, S._RuszA, correction_HHdbVI, 128

[peter]

The latest 128- collection (more difficulty of positions on average) you find here:


forum3/viewtopic.php?p=932939&sid=b8319 ... 10#p932939

thus you could compare directly to the latest runs and EloStatTS- evaluation given there too.

ProteusSF-Piranha (Stockfish derivative) -> http://www.microsmeta.com/DBLOG/articol ... colo=1525
File name : Peter.epd
Total test items : 128
Test for : best moves
Total engines : 1
Expand ply : 1
Laps : 1
Total tests : 128
Total corrects : 81 (63%)
Ave correct elapse : 14:10
Status : completed

2 CPUs Mac mini 2010 (!) Core 2 Duo 400 kN/s (not exactly a Threadripper )
How many positions calculates Vinvin PC with 30 threads of a 16x3.5GHz CPU and a 3070ti GPU?

15"/position versus 15 minutes /position

Correct/Total:
ProteusSF-P 230822: 81/128

Failed tests (hit *):
1. ProteusSF-P 230822:
TacticalInsanityNr.9_CorbitD_HTC108-5 S, 2, Carlsen, M._Caruana, F._HTC108-7 WM2018, 3, Gusia_Barlov_Schweiz 1980 HTC108-28, 12, Rudolph, W.E._?_Studie 1912 HTC108-30 S, 14, Van Essen, M._2004, #22._HTC108-48 (-9ZNr41) S, 16, Neghina, M._Chase, Derived_HTC108-54, 19, Morozevich, A._Vitiugov, N._Reggio 2011 HTC108-100, 22, ACT3, ??._?_?, 36, ACT3, ??._Festung_?, 37, ACT3, ??._Festung_S, 38, ACT3, ??._Studie?_?, 39, ACT3, ??._Studie?_S, 40, Neghina, M._WhieTo, MoveAndWin_2018? S, 43, Popov, Study_Glarean 2006, 1996?._HTC108-58, 44, Sikorsky, Horst_Novak, Joze_Sideline, 45, Krug, P._CCC, Paloma_S, 46, Sikorsky, Horst_Vinchev, Simeon_corr GER, Hiarcs, 49, Study, ?._?_IQ 100, 51, Study?, ?._?_ACT3, 52, Study?, ?._?_TTT2 31, 53, Study, ?._CCC_ACT5 11, 56, Study, ?._?_ACT5 29, 57, ERET 032_Zugzwang_Fahrni 1922, 59, ERET 046_Endspiel T&S vs T&S_Grandelius-Raznikov, Albena 2011, 61, ERET 090_Endspiel L vs L_Salai 2011, 63, ERET 093_Endspiel T&S&B vs T&T&L_Simkhovich 1923, 64, Fortress, Study?_?_ACT3-51 S, 66, Cyclic, ZZ?._?_ACT5-25 2, 68, Kos, V._?_ACT5-34 2, 73, Zinar, M._HHdbVI.0210037.002a., 09076?._ACT5-44 2, 76, Bernhardt, G._?_ACT5-58 2, 81, Unknows, Author_?_ACT5-61 2, 82, SilhanJ, ZizkaL_?_ACT5-59 3, 86, TTT2.053_?_TTT2, 91, Smyslov, V._?_ACT4-41 1, 95, Wotawa, A._?_ACT4-66 1, 98, Blathy, O._?_ACT4-94 1, 101, Cortlever, N._HHdbVI_ACT3-29 1, 102, Korolkov, V._?_ACT3-38, 103, Fortress, ?._?_ACT3-41 1, 104, Kraemer, A._?_ACT3-49, 106, Sivkov, N._#10, ?._ACT3-88, 107, Bazlo, Yuri_10moves, less_2019 HHdbVI 1, 112, TCEC, rj329_?_CCC Nr.5, 115, Yarmonov, I._Glarean_2020, 117, Krug, P.._Glarean, EngineCrackers2_2018, 120, Illescas, Miguel_Glarean, EngineCrackers_?, 122

Successful tests:
1. ProteusSF-P 230822:
HTC108, 2._Corbit, Dann_Tactical Insanity Nr.1, 1, ACT-Suite, Nr.12_Dorsz, OpenChess_HTC108-8, 4, van Foreest, J._L'Ami, E._HTC108-9 Zug 24...Sideline, 5, Konig?, CCC._Cornforth96?, CSS+Playchess_HTC108-11 Arasan, 6, PoorFish, CostalbaM_Borderline, Dual_HTC108-12, 7, LC0, CCC._SF, CSS._HTC108-15 S, 8, AlphaZero, HTC108-21._SF_Game9 2017, 9, Gashimov, V._Azariov, S._2011 HTC108-23, 10, CorbitD, CCC+Rybkaf_?_HTC108-25, 11, Bondarenko, &Kuznezov_Glarean, PoorStockfish_HTC108-29, 13, Gusev, Y._Auerbach, E.25._HTC108-42, 15, Neghina, M._Stronghold, 2003?._HTC108-49 S, 17, Tunik_Duzhakov_2011 HTC108-52, 18, SolomonKing, ?CCC._2.g4, Hiarcs_HTC108-80 S, 20, Nicolic, E._Fischer, R._HTC108-94 S, 21, M. Garcia - P. Krug, Studie 2015_Glarean_The Engine Crackers, 23, Zepler, E.E._Studie, 1928._?, 24, Formela, Michal_Studie, 2019._Polski Zwiazek Szachowy, 25, Stoofvlees_SF_TCEC 2019, 26, Lasker, Emanuel_Lipke, Paul_Nebenvariante 41.Ke4! MS, 27, Los, Andrii_Kubicki, Tadeusz_corr LIPEAD40 (PER), 28, PurePower_Exxon_Chessbase online, 29, Bauer, Robert_Zhak, Boris Mikhailovich_WC30/final, 30, Mihai Neghina_Desperation_White to move and draw S, 31, Tauber, Hans_Sikorsky, Horst_37.e5!, 32, Bondarenko+Gorgiev, Studie1959_BrillanterSchachzug, 17._?, 33, ACT3, ??._Festung_S, 34, ACT3, ??._Studie?_?, 35, Neghina, M._GoldenerK�fig, 2009._S, 41, Ree, H._Hort, V._Wijk 1986, 42, TCEC, Sammlung_Nr.8, j329_CCC, 47, TCEC, Sammlung_Nr.22, rj329_CCC, 48, Zemljanski, Y.I._1move, cut_Tsjechover-90 S, 50, Semenikov+, D A._Sikorsky, H._corr. USA 2021, 54, Petrov, M._Sikorsky, H._CL/2021, 55, ERET 015_Endspiel D&B vs T&L&B_Becker 2015, 58, ERET 037_Zugzwang_Garcia 2000, 60, ERET 050_Koenigsangriff_Hansen-Barua, Biel 1993, 62, ERET 094_Damenopfer_Sirobaba-Chaika, CorrGame 2011, 65, ACT4, Nr.1_?_ACT4-1, 67, Blass, U.._CCC_ACT5-27 2, 69, Campioli, M._#14, ?._ACT5-28 2, 70, Krug, P._?_ACT5-30 2, 71, KrugP, Garcia_HHdbVI.0044383.001, a.01968?_ACT5-31 1, 72, Sokka, H._HHdbVI.1076075.001, a.53387?_ACT5-36 1, 74, Sehwers, J?._HHdbVI.1643584.002a., 89051?._ACT5-ana, 75, Nestorescu, V._HHdbVI.0708805.002a., 33432?._ACT5-46 1, 77, Timman, J._HHdbVI.0033988.002a., 01480._ACT5-48 2, 78, Unknown, Author_?_ACT5-51 2, 79, Ivanov, V._?_ACT5-53 2, 80, Naef, W._?_ACT5-63 2, 83, #7, ?._?_ACT5-67 2, 84, #9, ?._?_ACT5-68 2, 85, Hiarcs, ?._?_TTT2 2, 87, Hiarcs, ?._?_TTT2 6 2, 88, Derived, from_StavrietskyA, 2017._TTT2 14, 89, Midgame, ?._?._TTT2 23, 90, TTT2.073_?_TTT2, 92, RoxlauM, GrushkoM_HHdbVI_ACT4 4 1, 93, Smyslov, V._?_ACT4-26 1, 94, JostenG_?_?, 96, Study, ?._?_ACT4-59, 97, Study, ?._?_ACT4-90 1, 99, Bethge, F._?_ACT4-91, 100, Unknown, Author_?_ACT3-48, 105, Timman=J_Glarean, Nightmare_?, 108, Boudy Bueno, J.L.._Szabo, L._Varna 1979 Txb2!? Eret 17, 109, Lc0-dev_Stockfish_ECO MegaMatch II! (1|1), 110, Carlsen, M._L'Ami, E._Corus B VLH16 35, 111, Opening, Hiarcs_!, ?!._?, 113, TCEC, rj329_?_CCC, 114, Gleichmann, Matthias_Praznik, Niko_corr CT21/final, 116, Eman_SugaR_Glarean, 118, Hobacle's, VersionOf_Walter Eigenmann's, Study_?, 119, Gniazdowski, Marian_Rau, Johann_corr CL/2019/C8 Glarean, 121, Heinz, J._Glarean_Study 1981, 123, Josten, G.._Glarean, EngineCrackers_Studie 2001, 124, Timman, J.._HHdbVI_?, 125, Gelly, M.._HHdbVI_Glarean, 126, Krug, P._HHdbVI, Glarean_?, 127, Ganguly, S._RuszA, correction_HHdbVI, 128

Thanks for repeating your test.

That's all one run, right? With 15" or 15minutes/position? (Ave correct elapse : 14:10?) With 15 minutes/pos., guess you meant Vincent's TC for single thread-testing, didn't you?
Which setting of GUI did you have as for defined TC?
Strict time/position? 1 exta ply is too little as for my personal pov,, Frank Schubert recommends at least2 in the readme of EloStatTS, Walter Eigenmann advocates for max. possible always, 99 in Fritz- GUI

As a matter of fact I'd say so too as for Banksia too ((max. possible, don't trust the lower extra iterations in this one GUI as written before), in Fritz and Shredder- GUI I have 2 extra plies normally,
regards

[AlexChess]

The latest 128- collection (more difficulty of positions on average) you find here:


forum3/viewtopic.php?p=932939&sid=b8319 ... 10#p932939

thus you could compare directly to the latest runs and EloStatTS- evaluation given there too.

ProteusSF-Piranha (Stockfish derivative) -> http://www.microsmeta.com/DBLOG/articol ... colo=1525
File name : Peter.epd
Total test items : 128
Test for : best moves
Total engines : 1
Expand ply : 1
Laps : 1
Total tests : 128
Total corrects : 81 (63%)
Ave correct elapse : 14:10
Status : completed

2 CPUs Mac mini 2010 (!) Core 2 Duo 400 kN/s (not exactly a Threadripper )
How many positions calculates Vinvin PC with 30 threads of a 16x3.5GHz CPU and a 3070ti GPU?

15"/position versus 15 minutes /position

Correct/Total:
ProteusSF-P 230822: 81/128

Thanks for repeating your test.

regards

Mac mini 2010 calculates only 0.4 Mn/s . If @Vinvin PC reaches 30-40 Mn/s I think that 15 minutes (for each position) instead of 15 seconds Is 0K, but correct me if I'm wrong

[peter]

The latest 128- collection (more difficulty of positions on average) you find here:


forum3/viewtopic.php?p=932939&sid=b8319 ... 10#p932939

thus you could compare directly to the latest runs and EloStatTS- evaluation given there too.

ProteusSF-Piranha (Stockfish derivative) -> http://www.microsmeta.com/DBLOG/articol ... colo=1525
File name : Peter.epd
Total test items : 128
Test for : best moves
Total engines : 1
Expand ply : 1
Laps : 1
Total tests : 128
Total corrects : 81 (63%)
Ave correct elapse : 14:10
Status : completed

2 CPUs Mac mini 2010 (!) Core 2 Duo 400 kN/s (not exactly a Threadripper )
How many positions calculates Vinvin PC with 30 threads of a 16x3.5GHz CPU and a 3070ti GPU?

15"/position versus 15 minutes /position

Correct/Total:
ProteusSF-P 230822: 81/128

Thanks for repeating your test.

regards

Mac mini 2010 calculates only 0.4 Mn/s . If Vinvin PC reaches 40 Mn/s I think that 15 Minutes (for each position) instead of 15 seconds Is 0K, but correct me if I'm wrong

So it yet was 15 minutes, not seconds, ok, I see.
As for nodes/s and time to solution there's no certain correlation of more then one single position and one single engine and one single CPU (even less comparable as for GPUs) , if at all, but I guess you know that already too

Yet as for comparing just nodes to nodes, you're about right. If you'd let two runs, one with your hardware- time of 15'/pos. of your hardware togehter with another one run of 15"/pos. and my hardware, which gives with 30 threads of the 16x3.5GHz Ryzen- CPU and avx2- compile of SF 130722 about 22 Mn/s with full hash at starting position (most of the test- positions are endgames or nearer to endgame than to starting postion and that's a big dfference as for n/s and SF, even on average, even more at single position), and then let both runs be evaluated by EloStatTS in one rating- list, then you'd know the difference between the hardware- time- performance of the same engine in two single runs of it

Anyhow I don't have Proteus, otherwise I'd simply let it have a run with my usual hardware- TC and add it to my EloStatTS- list, if I'll do so sooner or later, I'll let you know the result,
regards