Help with Scorpio 3.0.8

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

Moderator: Ras

User avatar
Gabor Szots
Posts: 1441
Joined: Sat Jul 21, 2018 7:43 am
Location: Budapest, Hungary
Full name: Gabor Szots

Re: Help with Scorpio 3.0.8

Post by Gabor Szots »

This is what I get running scorpio.bat go quit:
feature done=0
Number of cores 4 of 4
treeht 83886080 X 320 = 25600.0 MB
processors [16]
ht 67108864 X 16 = 1024.0 MB
eht 262144 X 8 X 16 = 64.0 MB
pht 32768 X 24 X 16 = 12.0 MB
EgbbProbe 4.3 by Daniel Shawul
egbb_cache 4084 X 8216 = 32.0 MB
Loading egbbs....
180 egbbs loaded !
nn_cache 131072 X 1568 = 196.0 MB
Loading neural network : e:\Sakk\Engines\Scorpio\nets-scorpio/ens-net-12x128.pb
With "FLOAT" precision
Loading graph on /cpu:0
0. main_input = -1 -1 32
1. policy/Reshape
0. value/BiasAdd
Neural network loaded !
loading_time = 3s
# rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
# [st = 12502ms, mt = 29250ms , hply = 0 , moves_left 10]
63 25 116 252 63 216 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7
64 29 205 474 64 231 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6
65 32 301 714 65 236 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8
66 33 413 1002 66 242 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4
67 33 502 1249 67 248 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5
68 33 606 1545 68 254 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5
69 32 703 1832 69 260 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4
70 32 807 2154 70 266 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6
71 32 920 2468 71 268 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8
72 32 1032 2781 72 269 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3 Re8xe1 Qd1xe1
73 32 1129 3063 73 271 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3 Re8xe1 Qd1xe1
74 32 1217 3313 74 272 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3 Re8xe1 Qd1xe1 Nd6-e8 Bc1-f4
# nodes 24334 <0% qnodes> time 12305ms nps 1977 eps 0 nneps 272 tbhits 0
# Tree: nodes 4257 depth 24 pps 272 visits 3313 qcalls 0 scalls 0
move e2e4
Bye Bye

But there is something strange. At the first run I modified some parameters under Arena (by using the engine configuration), I set hash to 256 MB, eht to 256 as well, number of cores to 1, etc. Now as I copied the above printout here I noticed that they are the default values. Looking at the ini file I can see that it indeed contains the default values. And I can also see that the date of the ini file is the same as the date of installation. It has not been overwritten.
I have just changed some values under Arena and I can see that the ini file remained the same. Maybe that is the problem: Scorpio want to use all 4 cores and an eht of 25.6 GB when 2 of the cores are already in use by other tournaments and my RAM is only 8 GB.

It seems that Arena writes the modified values to some place but the engine reads them from its install directory e:\Sakk\Engines\Scorpio\bin\Windows\scorpio.ini.

Currently that's all I can provide, I'm going to investigate things further.

Regards,
Gabor
Gabor Szots
CCRL testing group
Top
Daniel Shawul
Posts: 4186
Joined: Tue Mar 14, 2006 11:34 am
Location: Ethiopia

Re: Help with Scorpio 3.0.8

Post by Daniel Shawul »

Hi Gabor,

The installation seems to be correct. I recommend you do not change the default values in scorpio.ini such as eht.
Scorpio by default uses all cores of your CPU, and it is preferable you test it as a "multicore" engine.
So if other engines are running at the same time it might get really slow.
For your CPU with 4 cores, it will launch 4x4=16 threads and will use all 4 cores.
If you have to change settings of scorpio, always edit scorpio.ini instead of doing it via the GUI.
I don't trust what Arena and the like do to engines that have their own ini file. This has caused me severe problems f.i. on engine reinstallation.
8GB ram is fine. So the only issue I can see is you running other games in parallel.

regards,
Daniel
Top
User avatar
Gabor Szots
Posts: 1441
Joined: Sat Jul 21, 2018 7:43 am
Location: Budapest, Hungary
Full name: Gabor Szots

Re: Help with Scorpio 3.0.8

Post by Gabor Szots »

Daniel Shawul wrote: ↑Thu Jun 04, 2020 4:31 pm Hi Gabor,

The installation seems to be correct. I recommend you do not change the default values in scorpio.ini such as eht.
Scorpio by default uses all cores of your CPU, and it is preferable you test it as a "multicore" engine.
So if other engines are running at the same time it might get really slow.
For your CPU with 4 cores, it will launch 4x4=16 threads and will use all 4 cores.
If you have to change settings of scorpio, always edit scorpio.ini instead of doing it via the GUI.
I don't trust what Arena and the like do to engines that have their own ini file. This has caused me severe problems f.i. on engine reinstallation.
8GB ram is fine. So the only issue I can see is you running other games in parallel.

regards,
Daniel
I have edited scorpio.ini manually to use my preferred values, among them mt=1 but somehow it still wants to use all cores. Shall I not edit scorpio.bat as well?
Gabor Szots
CCRL testing group
Top
Daniel Shawul
Posts: 4186
Joined: Tue Mar 14, 2006 11:34 am
Location: Ethiopia

Re: Help with Scorpio 3.0.8

Post by Daniel Shawul »

"mt" is more like a batchsize parameter for the NN version of Scorpio. When running Scorpio on GPU, I launch as many as 640 threads on the CPU even when the CPU has only 8 cores.
The only way to restrict this, when running on the GPU, is by binding the Scorpio process to specific cpu cores using e.g. "START /A" on windows.

However, when running on CPU, reducing "mt" to the number of available cpu cores, such as 1, still work reasonably unlike the GPU case.
It will be 2-3x slower than when it was set to "mt 16" but it will still work.
In any case, I prefer you set it as a multicore engine if you can with it using all 4 of your cores and mt still set to 16.
I can still set mt to 16, 64, 128 and it only implies the batchsize, it doesn't matter as long as all threads will run on those 4 cores.

regards,
Daniel
Top
User avatar
Gabor Szots
Posts: 1441
Joined: Sat Jul 21, 2018 7:43 am
Location: Budapest, Hungary
Full name: Gabor Szots

Re: Help with Scorpio 3.0.8

Post by Gabor Szots »

Daniel Shawul wrote: ↑Thu Jun 04, 2020 5:20 pm "mt" is more like a batchsize parameter for the NN version of Scorpio. When running Scorpio on GPU, I launch as many as 640 threads on the CPU even when the CPU has only 8 cores.
The only way to restrict this, when running on the GPU, is by binding the Scorpio process to specific cpu cores using e.g. "START /A" on windows.

However, when running on CPU, reducing "mt" to the number of available cpu cores, such as 1, still work reasonably unlike the GPU case.
It will be 2-3x slower than when it was set to "mt 16" but it will still work.
In any case, I prefer you set it as a multicore engine if you can with it using all 4 of your cores and mt still set to 16.
I can still set mt to 16, 64, 128 and it only implies the batchsize, it doesn't matter as long as all threads will run on those 4 cores.

regards,
Daniel
That makes it clear, thanks.

Regards,
Gabor
Gabor Szots
CCRL testing group
Top
Dann Corbit
Posts: 12777
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Help with Scorpio 3.0.8

Post by Dann Corbit »

Daniel:
I have some nice GPUs and I can give scorpio some calibration time in a trade.
I will give you a week of 2080 Super GPU, 24 hours a day, in exchange for you dividing your depth figure by 4.
My database routines assume that among high strength engines, if all else is equal, then depth trumps.
Hence, I get "real" 70 ply depth searches which ran for 20 hours overwritten by Scorpio 71 ply searches that ran for 2 minutes.
So I have to do a lot of manual edits when I import Scorpio data now.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Top
Daniel Shawul
Posts: 4186
Joined: Tue Mar 14, 2006 11:34 am
Location: Ethiopia

Re: Help with Scorpio 3.0.8

Post by Daniel Shawul »

Dann Corbit wrote: ↑Thu Jun 04, 2020 7:50 pm Daniel:
I have some nice GPUs and I can give scorpio some calibration time in a trade.
I will give you a week of 2080 Super GPU, 24 hours a day, in exchange for you dividing your depth figure by 4.
My database routines assume that among high strength engines, if all else is equal, then depth trumps.
Hence, I get "real" 70 ply depth searches which ran for 20 hours overwritten by Scorpio 71 ply searches that ran for 2 minutes.
So I have to do a lot of manual edits when I import Scorpio data now.
Deal!! :) Set the mcts_strategy_depth to anything you like in scorpio.ini.
Here is what the new binary does when set to 20.

Code: Select all


$ ./scorpio.sh mcts_strategy_depth 20 go quit
feature done=0
Number of cores 4 of 4
treeht 83886080 X 320 = 25600.0 MB
processors [16]
ht 67108864 X 16 = 1024.0 MB
eht 262144 X 8 X 16 = 64.0 MB
pht 32768 X 24 X 16 = 12.0 MB
EgbbProbe 4.3 by Daniel Shawul
egbb_cache 4084 X 8216 = 32.0 MB
180 egbbs loaded !      
nn_cache 131072 X 1568 = 196.0 MB
Loading neural network : ../nets-scorpio/ens-net-12x128.pb
With "HALF" precision
Loading graph on /cpu:0
0. main_input = -1 -1 32 
1. policy/Reshape
0. value/BiasAdd
Neural network loaded !	
loading_time = 0s
# rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
# [st = 12502ms, mt = 29250ms , hply = 0 , moves_left 10]
21 25 90 212 21 234 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1
22 29 196 520 22 264 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3
23 32 310 841 23 271 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3
24 32 413 1149 24 278 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5
25 32 516 1458 25 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5
26 32 614 1781 26 289 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4
27 32 709 2074 27 292 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6
28 31 813 2375 28 291 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8
29 31 905 2590 29 285 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
30 31 996 2809 30 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
31 32 1101 3114 31 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
32 32 1191 3366 32 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
33 32 1191 3367 33 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
# nodes 25629 <0% qnodes> time 12098ms nps 2118 eps 0 nneps 281 tbhits 0
# Tree: nodes 4342 depth 20 pps 281 visits 3382 qcalls 0 scalls 0
move e2e4
Bye Bye
Note that in the endgame (for <= 10 pieces) Scorpio starts with an AB search, so it will add mcts_strategy_depth to the depth the AB search stopped.

Get your GPU cranking Dann!
Let me know if you have issues with using the client.

Daniel
Top
Modern Times
Posts: 3708
Joined: Thu Jun 07, 2012 11:02 pm

Re: Help with Scorpio 3.0.8

Post by Modern Times »

Daniel Shawul wrote: ↑Thu Jun 04, 2020 1:47 pm Could you print what you get when you run "scorpio.bat go quit" from the "bin/Windows" directory?
FYI this is what I get on my GTX1050

Code: Select all


C:\Engines\ScorpioNN\Scorpio\bin\Windows>scorpio.bat go quit
feature done=0
Number of cores 8 of 8
treeht 83886080 X 320 = 25600.0 MB
processors [10]
ht 67108864 X 16 = 1024.0 MB
eht 262144 X 8 X 10 = 40.0 MB
pht 32768 X 24 X 10 = 7.5 MB
EgbbProbe 4.3 by Daniel Shawul
egbb_cache 4084 X 8216 = 32.0 MB
180 egbbs loaded !
nn_cache 131072 X 1568 = 196.0 MB
Loading neural network : C:\Engines\ScorpioNN\Scorpio\nets-scorpio/ens-net-20x256.uff
With "INT8" precision
Loading graph on /gpu:0
0. main_input 2048 = 32 8 8
1. policy/Reshape 1024 = 1024 1 1
2. value/BiasAdd 3 = 3 1 1
Neural network loaded !
nn_cache 131072 X 1568 = 196.0 MB
Loading neural network : C:\Engines\ScorpioNN\Scorpio\nets-maddex/ME.uff
With "INT8" precision
Loading graph on /gpu:0
0. main_input 7168 = 112 8 8
1. value_head 3 = 3 1 1
2. policy_head 1858 = 1858 1 1
Neural network loaded !
loading_time = 2s
# rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
# [st = 12502ms, mt = 29250ms , hply = 0 , moves_left 10]
63 31 88 1693 63 1906 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4
64 31 176 3633 64 2063 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3 Re8xe1 Qd1xe1 Nd6-e8 Bc1-f4 d7-d5
65 30 263 5578 65 2120 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3 Re8xe1 Qd1xe1 Nd6-e8 Bc1-f4 d7-d5 Nb1-d2 c7-c6 Bf1-d3 g7-g6 Nd2-f3
66 30 350 7569 66 2158 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3 Re8xe1 Qd1xe1 Nd6-e8 Bc1-f4 d7-d5 Nb1-d2 c7-c6 Bf1-d3 g7-g6 Nd2-f3
67 29 439 9671 67 2198 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8
68 29 527 11888 68 2254 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8
69 29 614 14028 69 2281 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8
70 29 701 16140 70 2301 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8
71 30 789 18120 71 2294 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8 Bc2-b1
72 30 876 20246 72 2309 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8 Bc2-b1
73 30 961 22236 73 2313 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3 Bc8-b7 d2-d4 Rf8-e8 Nb1-d2 Be7-f8 a2-a4 h7-h6 Bb3-c2 Ra8-b8 Bc2-b1
# nodes 203951 <0% qnodes> time 9673ms nps 21084 eps 0 nneps 2310 tbhits 0
# Tree: nodes 28673 depth 30 pps 2309 visits 22236 qcalls 0 scalls 0
move e2e4
Bye Bye
Top
Daniel Shawul
Posts: 4186
Joined: Tue Mar 14, 2006 11:34 am
Location: Ethiopia

Re: Help with Scorpio 3.0.8

Post by Daniel Shawul »

At first I thought that it was a mistake it was launching only 10 threads, but it isn't.
The GTX 1050 has only 5 SMs, so it allocates 2x the number of threads.

I am curious though how much pps it will get with mt set to 20 or 40.
Could you set mt to these values in scoprio.ini and run scorpio.bat go quit again, and post the results here?
My estimate of using 2x number of SMs of GPU could be wrong on low end GPUs.

Thanks.
Daniel
Top
Dann Corbit
Posts: 12777
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Help with Scorpio 3.0.8

Post by Dann Corbit »

Daniel Shawul wrote: ↑Thu Jun 04, 2020 9:45 pm
Dann Corbit wrote: ↑Thu Jun 04, 2020 7:50 pm Daniel:
I have some nice GPUs and I can give scorpio some calibration time in a trade.
I will give you a week of 2080 Super GPU, 24 hours a day, in exchange for you dividing your depth figure by 4.
My database routines assume that among high strength engines, if all else is equal, then depth trumps.
Hence, I get "real" 70 ply depth searches which ran for 20 hours overwritten by Scorpio 71 ply searches that ran for 2 minutes.
So I have to do a lot of manual edits when I import Scorpio data now.
Deal!! :) Set the mcts_strategy_depth to anything you like in scorpio.ini.
Here is what the new binary does when set to 20.

Code: Select all


$ ./scorpio.sh mcts_strategy_depth 20 go quit
feature done=0
Number of cores 4 of 4
treeht 83886080 X 320 = 25600.0 MB
processors [16]
ht 67108864 X 16 = 1024.0 MB
eht 262144 X 8 X 16 = 64.0 MB
pht 32768 X 24 X 16 = 12.0 MB
EgbbProbe 4.3 by Daniel Shawul
egbb_cache 4084 X 8216 = 32.0 MB
180 egbbs loaded !      
nn_cache 131072 X 1568 = 196.0 MB
Loading neural network : ../nets-scorpio/ens-net-12x128.pb
With "HALF" precision
Loading graph on /cpu:0
0. main_input = -1 -1 32 
1. policy/Reshape
0. value/BiasAdd
Neural network loaded !	
loading_time = 0s
# rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
# [st = 12502ms, mt = 29250ms , hply = 0 , moves_left 10]
21 25 90 212 21 234 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1
22 29 196 520 22 264 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3
23 32 310 841 23 271 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 a7-a6 Bb5-a4 Ng8-f6 Ke1-g1 Bf8-e7 Rf1-e1 b7-b5 Ba4-b3 d7-d6 c2-c3 Ke8-g8 h2-h3
24 32 413 1149 24 278 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5
25 32 516 1458 25 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5
26 32 614 1781 26 289 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4
27 32 709 2074 27 292 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6
28 31 813 2375 28 291 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8
29 31 905 2590 29 285 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
30 31 996 2809 30 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
31 32 1101 3114 31 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
32 32 1191 3366 32 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
33 32 1191 3367 33 282 0 e2-e4 e7-e5 Ng1-f3 Nb8-c6 Bf1-b5 Ng8-f6 Ke1-g1 Nf6xe4 Rf1-e1 Ne4-d6 Nf3xe5 Bf8-e7 Bb5-f1 Nc6xe5 Re1xe5 Ke8-g8 d2-d4 Be7-f6 Re5-e1 Rf8-e8 c2-c3
# nodes 25629 <0% qnodes> time 12098ms nps 2118 eps 0 nneps 281 tbhits 0
# Tree: nodes 4342 depth 20 pps 281 visits 3382 qcalls 0 scalls 0
move e2e4
Bye Bye
Note that in the endgame (for <= 10 pieces) Scorpio starts with an AB search, so it will add mcts_strategy_depth to the depth the AB search stopped.

Get your GPU cranking Dann!
Let me know if you have issues with using the client.

Daniel
I will help even if it does not get fixed, but I am not talking about fixing it for me, I am talking about fixing it everywhere (e.g. I download a file from TCEC and the scorpio depths are in the 70s.) Maybe it just needs to have a default changed in the file?
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Top

Return to “Computer Chess Club: General Topics”