Allie & Stein

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

Re: Allie & Stein

Post by gonzochess75 »

Crem and the Lc0 developers have been nothing but awesome about the introduction and development of Allie. Indeed, crem was kindly answering questions about the Lc0 backend and how it works when I started to create Allie.

As I said, Allie would not be possible without the Lc0 project and developer team. That said, it is original and different than Lc0 (which can be plainly seen in that it is weaker than Lc0 currently given the same network :)

The plan is to make Allie even more different in the future with regards to the search. Although I have not released it, the patch to switch to pure AlphaBeta search exists and plays decent chess although it is weaker (currently) than the mcts search. In fact, if I had not entered or released Allie yet it would probably be much further along, but I have been spending my time making Allie scale.

All this is to say that there exists little to no contentiousness here and I see no reason to introduce it.
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

Re: Allie & Stein

Post by gonzochess75 »

velmarin wrote: Thu Mar 07, 2019 3:06 am I am unfamiliar with QT, I get this error when I try it on Windows.
A link is appreciated if someone has got it.
Image
I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the same thing on my own windows 10 vmware instance it works. Anyway, here is the appveyor.yml that should give you an idea about how to compile this for Windows.



If anyone would like to help me bugfix the script I will hook it up so that Windows executables are automatically generated when I release a new version of Allie in the future.

In the meantime, @borg has also been working on an appveyor.yml that produced the executable, but it is dynamically linked so you'll have to get ahold of Qt dlls for it to work. You can find that here: https://ci.appveyor.com/project/borg323/allie/history

Steppenwolf, to close out the thread... I have been doing opensource development for quite a long time and I've never published pre-built executables for any of my hobby projects. I dare say that the vast, vast, vast majority of open source projects on github are the same way. I hope you will understand that there was no maliciousness in my failure to do so for this project either. Anyway, like I said, it will be rectified shortly hopefully.
supersharp77
Posts: 1250
Joined: Sat Jul 05, 2014 7:54 am
Location: Southwest USA

Re: Allie & Stein

Post by supersharp77 »

gonzochess75 wrote: Thu Mar 07, 2019 3:04 pm
velmarin wrote: Thu Mar 07, 2019 3:06 am I am unfamiliar with QT, I get this error when I try it on Windows.
A link is appreciated if someone has got it.
Image
I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the same thing on my own windows 10 vmware instance it works. Anyway, here is the appveyor.yml that should give you an idea about how to compile this for Windows.



If anyone would like to help me bugfix the script I will hook it up so that Windows executables are automatically generated when I release a new version of Allie in the future.

In the meantime, @borg has also been working on an appveyor.yml that produced the executable, but it is dynamically linked so you'll have to get ahold of Qt dlls for it to work. You can find that here: https://ci.appveyor.com/project/borg323/allie/history

Program needs some exotic dlls before it will run cudart64_100.dll........any ideas? Thx AR 8-)
chrisw
Posts: 4457
Joined: Tue Apr 03, 2012 4:28 pm
Location: Midi-Pyrénées
Full name: Christopher Whittington

Re: Allie & Stein

Post by chrisw »

supersharp77 wrote: Sat Mar 09, 2019 2:55 am
gonzochess75 wrote: Thu Mar 07, 2019 3:04 pm
velmarin wrote: Thu Mar 07, 2019 3:06 am I am unfamiliar with QT, I get this error when I try it on Windows.
A link is appreciated if someone has got it.
Image
I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the same thing on my own windows 10 vmware instance it works. Anyway, here is the appveyor.yml that should give you an idea about how to compile this for Windows.



If anyone would like to help me bugfix the script I will hook it up so that Windows executables are automatically generated when I release a new version of Allie in the future.

In the meantime, @borg has also been working on an appveyor.yml that produced the executable, but it is dynamically linked so you'll have to get ahold of Qt dlls for it to work. You can find that here: https://ci.appveyor.com/project/borg323/allie/history

Program needs some exotic dlls before it will run cudart64_100.dll........any ideas? Thx AR 8-)
unless you can find that lone dll, you need to download several gbytes of cuda 10 or 10.1, which requires that you sign up to nvidia developers with a password of at least eight chars, mixed upper lower case numerical bla bla, and install it. Quite possibly you also need to have downloaded cuda development toolkit beforehand as well.
User avatar
Guenther
Posts: 4712
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: Allie & Stein

Post by Guenther »

chrisw wrote: Sat Mar 09, 2019 11:11 am
supersharp77 wrote: Sat Mar 09, 2019 2:55 am
gonzochess75 wrote: Thu Mar 07, 2019 3:04 pm
velmarin wrote: Thu Mar 07, 2019 3:06 am I am unfamiliar with QT, I get this error when I try it on Windows.
A link is appreciated if someone has got it.
Image
I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the same thing on my own windows 10 vmware instance it works. Anyway, here is the appveyor.yml that should give you an idea about how to compile this for Windows.



If anyone would like to help me bugfix the script I will hook it up so that Windows executables are automatically generated when I release a new version of Allie in the future.

In the meantime, @borg has also been working on an appveyor.yml that produced the executable, but it is dynamically linked so you'll have to get ahold of Qt dlls for it to work. You can find that here: https://ci.appveyor.com/project/borg323/allie/history

Program needs some exotic dlls before it will run cudart64_100.dll........any ideas? Thx AR 8-)
unless you can find that lone dll, you need to download several gbytes of cuda 10 or 10.1, which requires that you sign up to nvidia developers with a password of at least eight chars, mixed upper lower case numerical bla bla, and install it. Quite possibly you also need to have downloaded cuda development toolkit beforehand as well.
The cuda dlls are delivered with each LC0 download since several months at least. He will also need the appropriate qt5core.dll.
But all of this is invain anyway, because he has no available cuda hardware (can be deducted from earlier posts of him).
https://rwbc-chess.de

[Trolls n'existent pas...]
chrisw
Posts: 4457
Joined: Tue Apr 03, 2012 4:28 pm
Location: Midi-Pyrénées
Full name: Christopher Whittington

Re: Allie & Stein

Post by chrisw »

Guenther wrote: Sat Mar 09, 2019 11:49 am
chrisw wrote: Sat Mar 09, 2019 11:11 am
supersharp77 wrote: Sat Mar 09, 2019 2:55 am
gonzochess75 wrote: Thu Mar 07, 2019 3:04 pm
velmarin wrote: Thu Mar 07, 2019 3:06 am I am unfamiliar with QT, I get this error when I try it on Windows.
A link is appreciated if someone has got it.
Image
I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the same thing on my own windows 10 vmware instance it works. Anyway, here is the appveyor.yml that should give you an idea about how to compile this for Windows.



If anyone would like to help me bugfix the script I will hook it up so that Windows executables are automatically generated when I release a new version of Allie in the future.

In the meantime, @borg has also been working on an appveyor.yml that produced the executable, but it is dynamically linked so you'll have to get ahold of Qt dlls for it to work. You can find that here: https://ci.appveyor.com/project/borg323/allie/history

Program needs some exotic dlls before it will run cudart64_100.dll........any ideas? Thx AR 8-)
unless you can find that lone dll, you need to download several gbytes of cuda 10 or 10.1, which requires that you sign up to nvidia developers with a password of at least eight chars, mixed upper lower case numerical bla bla, and install it. Quite possibly you also need to have downloaded cuda development toolkit beforehand as well.
The cuda dlls are delivered with each LC0 download since several months at least. He will also need the appropriate qt5core.dll.
But all of this is invain anyway, because he has no available cuda hardware (can be deducted from earlier posts of him).
It depends. It used to be that one had to go through hoops to get the dll’s, because nvidia only distributed them via the password developer download. Then Crem or someone persuaded nvidia to allow the dll’s to be distributed along with lczero. That was with cuda 9.

But, if developers have upgraded to tensorflow13 plus, they then need to cuda10 toolkit download and a cuda 10 download to get at another nvidia dll, the 100.dll named above.

I don’t think (not sure) and am being too lazy to go look, that lczero requires cuda 10, and therefore only comes with the 9-series dlls.

If other developers have compiled with 10 series, then their user products will likely require 10 series nvidia support stuff. which probably “other” developers haven’t yet organised as a free distribution. Guessing a bit here.
User avatar
Guenther
Posts: 4712
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: Allie & Stein

Post by Guenther »

chrisw wrote: Sat Mar 09, 2019 2:54 pm
Guenther wrote: Sat Mar 09, 2019 11:49 am
The cuda dlls are delivered with each LC0 download since several months at least. He will also need the appropriate qt5core.dll.
But all of this is invain anyway, because he has no available cuda hardware (can be deducted from earlier posts of him).
It depends. It used to be that one had to go through hoops to get the dll’s, because nvidia only distributed them via the password developer download. Then Crem or someone persuaded nvidia to allow the dll’s to be distributed along with lczero. That was with cuda 9.

But, if developers have upgraded to tensorflow13 plus, they then need to cuda10 toolkit download and a cuda 10 download to get at another nvidia dll, the 100.dll named above.

I don’t think (not sure) and am being too lazy to go look, that lczero requires cuda 10, and therefore only comes with the 9-series dlls.

If other developers have compiled with 10 series, then their user products will likely require 10 series nvidia support stuff. which probably “other” developers haven’t yet organised as a free distribution. Guessing a bit here.
Since October 2018 the cuda 10 dlls are integrated in the LC0 download and already in September 2018 they were available as
separate download at the LC0 github site. No need for wrong guesswork. Cuda 10 is already 7 months old...
https://rwbc-chess.de

[Trolls n'existent pas...]
User avatar
Guenther
Posts: 4712
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: Allie & Stein

Post by Guenther »

gonzochess75 wrote: Thu Mar 07, 2019 3:04 pm ...

In the meantime, @borg has also been working on an appveyor.yml that produced the executable, but it is dynamically linked so you'll have to get ahold of Qt dlls for it to work. You can find that here: https://ci.appveyor.com/project/borg323/allie/history

...
Well, the last working compilation called 1.0.6 there issues illegal moves not only in the pv, but also while playing? Any idea?
The excerpt below is pasted from running it at cmd. I tried several nets, because I first thought it can't read older or newer nets,
but the result was always the same.

Code: Select all

       _ _ _
  __ _| | (_) ___
 / _` | | | |/ _ \
| (_| | | | |  __/
 \__,_|_|_|_|\___| 0.1 uci chess engine
uci
id name Allie 0.1
id author Adam Treat
option name GPUCores type spin default 1 min 0 max 256
option name Hash type spin default 20 min 0 max 65536
option name MaxBatchSize type spin default 256 min 0 max 65536
option name MoveOverhead type spin default 100 min 0 max 5000
option name Ponder type check default false
option name Threads type spin default 2 min 0 max 256
option name UseFP16 type check default false
uciok
ucinewgame
Using leela net: ID32930go depth 10
info depth 1 seldepth 1 nodes 1 nps 3 batchSize 1 score cp 45 time 323 hashfull0 tbhits -1 pv e2e4
info depth 1 seldepth 2 nodes 3 nps 6 batchSize 1 score cp -65 time 524 hashfull 0 tbhits -1 pv c2c4 a8b8
info depth 2 seldepth 3 nodes 64 nps 48 batchSize 21 score cp 108 time 1344 hashfull 0 tbhits -1 pv c2c4 a8b8
info depth 3 seldepth 4 nodes 85 nps 53 batchSize 21 score cp 108 time 1607 hashfull 0 tbhits -1 pv c2c4 a8b8
info depth 3 seldepth 5 nodes 524 nps 82 batchSize 74 score cp -105 time 6416 hashfull 1 tbhits -1 pv c2c4 a8b8 b1c3
info depth 3 seldepth 6 nodes 596 nps 91 batchSize 71 score cp -105 time 6515 hashfull 1 tbhits -1 pv c2c4 a8b8 b1c3
info depth 4 seldepth 7 nodes 827 nps 116 batchSize 89 score cp 114 time 7155 hashfull 1 tbhits -1 pv e2e4 e7d5 d2d4 h8g8
info depth 4 seldepth 8 nodes 1218 nps 159 batchSize 103 score cp 20 time 7671 hashfull 1 tbhits -1 pv e2e4 b8a6 d2d4 f8g7 g1f3 d8c7 g2g3
info depth 4 seldepth 8 nodes 1735 nps 169 batchSize 127 score cp 71 time 10267hashfull 1 tbhits -1 pv e2e4 b8a6 d2d4 f8g7 g1f3 d8c7 g2g3
info depth 4 seldepth 8 nodes 1995 nps 193 batchSize 136 score cp 38 time 10320hashfull 1 tbhits -1 pv d2d4 b8a6 e2e4 a8a7 g1f3 d8c7 b2b3
info depth 5 seldepth 8 nodes 2163 nps 169 batchSize 136 score cp 52 time 12782hashfull 1 tbhits -1 pv e2e4 b8a6 d2d4 f8g7 g1f3 d8c7 g2g3
info depth 5 seldepth 8 nodes 2437 nps 188 batchSize 144 score cp -38 time 12963 hashfull 1 tbhits -1 pv d2d4 g7f5 e2e4 a8a7 g1f3 d8c7 d1e2
info depth 5 seldepth 9 nodes 2760 nps 212 batchSize 150 score cp -38 time 13002 hashfull 1 tbhits -1 pv d2d4 g7f5 e2e4 a8a7 g1f3 d8c7 d1e2
info depth 5 seldepth 9 nodes 3031 nps 208 batchSize 156 score cp 33 time 14539hashfull 3 tbhits -1 pv e2e4 f7g5 d2d4 h8g8 h2h4 d8c8 b2b4
info depth 5 seldepth 12 nodes 3605 nps 236 batchSize 164 score cp 22 time 15246 hashfull 3 tbhits -1 pv e2e4 f7g5 h2h3 d8c7 b1c3 d8c7 c3b1 d8c7 b1c3 d8c7 c3b1
info depth 6 seldepth 14 nodes 4273 nps 239 batchSize 172 score cp 8 time 17897hashfull 3 tbhits -1 pv e2e4 f7g5 h2h3 d8c7 b1c3 d8c7 c3b1 d8c7 b1c3 d8c7 c3b1
info depth 6 seldepth 14 nodes 6535 nps 318 batchSize 172 score cp -1 time 20542 hashfull 7 tbhits -1 pv e2e4 f7g5 d2d4 h8g8 h2h4 d8c8 b2b4
info depth 6 seldepth 14 nodes 7163 nps 309 batchSize 173 score cp -1 time 23193 hashfull 7 tbhits -1 pv e2e4 h7f6 d2d4 h8g8 g2g3 a8b8 h2h3 d8c7 d1d2
info depth 6 seldepth 14 nodes 8222 nps 318 batchSize 179 score cp -10 time 25876 hashfull 9 tbhits -1 pv e2e4 h7f6 b2b3 d8c8 a2a3 d8c7 d2d4
info depth 6 seldepth 15 nodes 8512 nps 328 batchSize 181 score cp -8 time 25927 hashfull 11 tbhits -1 pv e2e4 h7f6 b2b3 d8c8 a2a3 d8c7 d2d4
info depth 6 seldepth 15 nodes 9874 nps 321 batchSize 187 score cp -8 time 30730 hashfull 14 tbhits -1 pv e2e4 h7f6 b2b3 d8c8 a2a3 d8c7 d2d4
info depth 6 seldepth 15 nodes 11399 nps 329 batchSize 186 score cp -9 time 34606 hashfull 16 tbhits -1 pv e2e4 h7f6 b2b3 d8c8 a2a3 d8c7 d2d4
info depth 7 seldepth 15 nodes 12155 nps 328 batchSize 188 score cp -9 time 37004 hashfull 16 tbhits -1 pv e2e4 h7f6 b2b3 d8c8 a2a3 d8c7 d2d4
info depth 7 seldepth 15 nodes 12748 nps 323 batchSize 163 score cp -9 time 39511 hashfull 17 tbhits -1 pv e2e4 h7f6 b2b3 d8c8 a2a3 d8c7 d2d4
info depth 7 seldepth 15 nodes 12775 nps 323 batchSize 158 score cp 33 time 39532 hashfull 17 tbhits -1 pv g1f3 d8c7 a2a4 a8b8 h2h3 d8c7 f3g1 d8c7 g1f3 d8c7 f3g1 d8c7 g1f3
info depth 7 seldepth 15 nodes 13614 nps 320 batchSize 150 score cp 27 time 42595 hashfull 18 tbhits -1 pv g1f3 d8c7 d2d4 h8g8 a2a3 h8h7 e2e4 h8g8 b1c3 d8c7 h2h3 h8h7
info depth 7 seldepth 15 nodes 14402 nps 319 batchSize 149 score cp 24 time 45099 hashfull 19 tbhits -1 pv g1f3 d8c7 d2d4 h8g8 a2a3 h8h7 e2e4 h8g8 b1c3 d8c7 h2h3 h8h7
info depth 7 seldepth 16 nodes 14677 nps 318 batchSize 150 score cp 25 time 46163 hashfull 20 tbhits -1 pv g1f3 d8c7 d2d4 h8g8 a2a3 h8h7 e2e4 h8g8 b1c3 d8c7 h2h3 h8h7
info depth 7 seldepth 16 nodes 16389 nps 326 batchSize 153 score cp 26 time 50228 hashfull 21 tbhits -1 pv g1f3 d8c7 d2d4 h8g8 a2a3 h8h7 e2e4 h8g8 b1c3 d8c7 h2h3 h8h7
info depth 8 seldepth 16 nodes 17366 nps 329 batchSize 156 score cp 29 time 52788 hashfull 21 tbhits -1 pv g1f3 d8c7 d2d4 h8g8 a2a3 h8h7 e2e4 h8g8 b1c3 d8c7 h2h3 h8h7
info depth 8 seldepth 16 nodes 18769 nps 335 batchSize 157 score cp 15 time 55991 hashfull 22 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 18 nodes 19262 nps 344 batchSize 158 score cp 14 time 56028 hashfull 23 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 19687 nps 344 batchSize 159 score cp 13 time 57271 hashfull 24 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 21187 nps 353 batchSize 161 score cp 11 time 60025 hashfull 27 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 22419 nps 355 batchSize 162 score cp 6 time 63096 hashfull 29 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 23663 nps 360 batchSize 165 score cp 6 time 65805 hashfull 29 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 24599 nps 359 batchSize 167 score cp 5 time 68516 hashfull 31 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 26784 nps 371 batchSize 170 score cp 3 time 72175 hashfull 31 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 27816 nps 370 batchSize 172 score cp 5 time 75140 hashfull 32 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 28745 nps 370 batchSize 174 score cp 9 time 77757 hashfull 32 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 29800 nps 368 batchSize 176 score cp 9 time 80922 hashfull 32 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 8 seldepth 20 nodes 31158 nps 371 batchSize 177 score cp 5 time 83912 hashfull 33 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 9 seldepth 20 nodes 31406 nps 283 batchSize 177 score cp -1 time 110824 hashfull 42 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 9 seldepth 24 nodes 44280 nps 396 batchSize 181 score cp -2 time 111701 hashfull 43 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 9 seldepth 24 nodes 45290 nps 396 batchSize 182 score cp -3 time 114508 hashfull 43 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 9 seldepth 24 nodes 46396 nps 396 batchSize 182 score cp -3 time 117171 hashfull 46 tbhits -1 pv g1f3 d8d7 e2e4 h8g8 d2d4 f8e7 h2h3 d8c7 b1c3 h7g5 g2g3 h8g8 a2a4
info depth 9 seldepth 24 nodes 47417 nps 396 batchSize 183 score cp -3 time 119828 hashfull 49 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 f3g1 d8c7 d2d4 d8e7 h2h3 d8c7 a2a3 d8c7 g1f3 d8c7 g2g3 d8c7 b2b3 d8c7 h3h4 d8c7 b3b4 d8c7 f1e2
info depth 9 seldepth 24 nodes 48952 nps 398 batchSize 184 score cp -3 time 123131 hashfull 56 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 f3g1 d8c7 d2d4 d8e7 h2h3 d8c7 a2a3 d8c7 g1f3 d8c7 g2g3 d8c7 b2b3 d8c7 h3h4 d8c7 b3b4 d8c7 f1e2
info depth 9 seldepth 24 nodes 49951 nps 397 batchSize 184 score cp -3 time 125805 hashfull 57 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 d2d4 d8c8 h2h3 d8c7 b1c3 a8b8 a2a4 h8g8 g2g3 h8g8 a1b1 d8c7 d1d2
info depth 9 seldepth 24 nodes 51065 nps 397 batchSize 185 score cp -4 time 128504 hashfull 57 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 d2d4 d8c8 h2h3 d8c7 b1c3 a8b8 a2a4 h8g8 g2g3 h8g8 a1b1 d8c7 d1d2
info depth 9 seldepth 24 nodes 52007 nps 396 batchSize 186 score cp -4 time 131205 hashfull 57 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 53112 nps 397 batchSize 187 score cp -4 time 133910 hashfull 58 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 54168 nps 397 batchSize 188 score cp -4 time 136601 hashfull 59 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 54866 nps 394 batchSize 188 score cp -3 time 139258 hashfull 59 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 56154 nps 395 batchSize 189 score cp -2 time 141998 hashfull 59 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 57128 nps 395 batchSize 190 score cp -3 time 144692 hashfull 59 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 58042 nps 393 batchSize 189 score cp -4 time 147566 hashfull 61 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 59468 nps 395 batchSize 187 score cp -3 time 150497 hashfull 61 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 9 seldepth 24 nodes 59994 nps 392 batchSize 187 score cp -1 time 153213 hashfull 61 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
info depth 10 seldepth 24 nodes 62242 nps 397 batchSize 186 score cp -2 time 156627 hashfull 62 tbhits -1 pv g1f3 h7f6 e2e4 e7c6 b1c3 d8c7 d2d4 a8a7 h2h3 c8d7 a2a3 h8g8
bestmove g1f3 ponder h7f6
https://rwbc-chess.de

[Trolls n'existent pas...]
chrisw
Posts: 4457
Joined: Tue Apr 03, 2012 4:28 pm
Location: Midi-Pyrénées
Full name: Christopher Whittington

Re: Allie & Stein

Post by chrisw »

Guenther wrote: Sat Mar 09, 2019 3:27 pm
chrisw wrote: Sat Mar 09, 2019 2:54 pm
Guenther wrote: Sat Mar 09, 2019 11:49 am
The cuda dlls are delivered with each LC0 download since several months at least. He will also need the appropriate qt5core.dll.
But all of this is invain anyway, because he has no available cuda hardware (can be deducted from earlier posts of him).
It depends. It used to be that one had to go through hoops to get the dll’s, because nvidia only distributed them via the password developer download. Then Crem or someone persuaded nvidia to allow the dll’s to be distributed along with lczero. That was with cuda 9.

But, if developers have upgraded to tensorflow13 plus, they then need to cuda10 toolkit download and a cuda 10 download to get at another nvidia dll, the 100.dll named above.

I don’t think (not sure) and am being too lazy to go look, that lczero requires cuda 10, and therefore only comes with the 9-series dlls.

If other developers have compiled with 10 series, then their user products will likely require 10 series nvidia support stuff. which probably “other” developers haven’t yet organised as a free distribution. Guessing a bit here.
Since October 2018 the cuda 10 dlls are integrated in the LC0 download and already in September 2018 they were available as
separate download at the LC0 github site. No need for wrong guesswork. Cuda 10 is already 7 months old...
So they are. took me two days and about ten retries downloading all of cuda10 to rural France, when I had the one file all along. Doh!
gonzochess75
Posts: 208
Joined: Mon Dec 10, 2018 3:29 pm
Full name: Adam Treat

Re: Allie & Stein

Post by gonzochess75 »

Guenther, almost certainly the problem you are seeing is because of a mismatch between the cuda libraries I built with and the ones you have installed. I built the binary with cuda 10.0. You can look at the appveyor.yml script which has the exact cuda files to download and install.