Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

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

Moderators: hgm, Rebel, chrisw

smatovic
Posts: 2642
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by smatovic »

Zeta v099k does not scale well on Nvidia Pascal and Turing gpus, so I wrote a
patch that could fix this issue, but am not able to test it.

If you have an RTX 2080 TI, or similar, running, you may want to run a benchmark
for me, takes only a couple of minutes.

Binary - Windows 64 and Linux 64 bit:
https://github.com/smatovic/Zeta/tree/v099l/bin

Code:
https://github.com/smatovic/Zeta/tree/v099l

Bench instruction:

1. Make sure you have an working OpenCL Runtime Environment:
Open Terminal resp. Command Line and change directory to Zeta binary, type:

zeta.exe -l --dl

This should list the available OpenCL devices with platform id and devide id.

2. Create a config file
Run in terminal:

zeta.exe -l -p 0 -d 0 --guessconfigx

Where p is the selected platform id and d the selected device id.

Rename the created config file to config.txt

3. Run the smp benchmark
Run in terminal:

zeta.exe -l
new
sd 12
st 2000
benchsmp

The benchsmp command can take several minutes to finish.

4. Send me the created zeta.log file or post the benchsmp results.


################################################################################
### WARNING
################################################################################
GPUs may have an operating system and driver specific timeout for computation.

Nvidia GPUs may have an driver specific timeout of 5 seconds when display is
connected.

AMD GPUs may have an driver specific timeout of about 360 to 3600 seconds.

Therefore it is recommended to run the engine on an discrete GPU, without
display connected.

To increase the Windows OS GPU timeout from 2 to 20 seconds, double-click the
.reg file "SetWindowsGPUTimeoutTo20s.reg" and reboot your OS.
################################################################################


Thanks in advance,
Srdja
Top
pferd
Posts: 134
Joined: Thu Jul 24, 2014 2:49 pm

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by pferd »

I have an 1070 running archlinux with latest nvidia drivers and my display was connected throughout the process. The results are attached to this message.
Top
Joost Buijs
Posts: 1563
Joined: Thu Jul 16, 2009 10:47 am
Location: Almere, The Netherlands

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by Joost Buijs »

All I have a 1080 TI which is Pascal (Asus 1080TI Turbo with default settings).
Maybe these results are useful to you.

Code: Select all

Microsoft Windows [Version 10.0.18362.175]
(c) 2019 Microsoft Corporation. All rights reserved.

C:\WINDOWS\system32>d:

D:\>zeta.exe -l --dl
#> ### Query the OpenCL Devices on Host...
#>
#> Number of OpenCL Platforms found: 1
#>
#> platform_id:0, device_id: 0, Vendor: NVIDIA Corporation, Device name: GeForce GTX 1080 Ti
#>

D:\>zeta.exe -l p 0 d 0 --guessconfig
#>
#> ### Query the OpenCL Platforms on Host...
#>
#> Number of OpenCL Platforms found: 1
#>
#> Platform: 0,  Vendor:  NVIDIA Corporation
#>
#> ### Query the OpenCL Devices on Platform...
#>
#> Number of OpenCL Devices found: 1
#>
#> ### Query and check the OpenCL Device...
#>
#> Device: 0, Device name: GeForce GTX 1080 Ti
#>
#> OK, Device Endianness is little
#> OK, CL_DEVICE_MAX_COMPUTE_UNITS: 28
#> OK, CL_DEVICE_MAX_MEM_ALLOC_SIZE: 2816 MB >= 128 MB
#> OK, CL_DEVICE_GLOBAL_MEM_SIZE: 11264 MB
#> OK, Device extension cl_khr_global_int32_base_atomics is supported.
#> OK, Device extension cl_khr_local_int32_base_atomics is supported.
#> OK: Device extension cl_khr_int64_extended_atomics not supported.
#> OK, CL_DEVICE_MAX_WORK_GROUP_SIZE: 1024 >= 64
#> OK, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS: 3 >= 3
#> OK, CL_DEVICE_MAX_WORK_ITEM_SIZES [3]: 64 >= 64
#> OK, CL_DEVICE_AVAILABLE: CL_TRUE
#
#> ### Running NPS-Benchmark for minimal config on device,
#> ### this can last about 4 seconds...
#> ### threadsX: 1
#> ### threadsY: 1
#> ### total work-groups: 1
#> ### total threads: 64
#
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
depth: 1, nodes 20, nps: 6, time: 3.019000 sec, score: 60  move b1c3
depth: 2, nodes 27, nps: 27000, time: 0.001000 sec, score: 0  move a1a1
depth: 3, nodes 525, nps: 30882, time: 0.017000 sec, score: 60  move b1c3
depth: 4, nodes 618, nps: 38625, time: 0.016000 sec, score: 0  move b1c3
depth: 5, nodes 2432, nps: 38000, time: 0.064000 sec, score: 27  move b1c3
depth: 6, nodes 16432, nps: 47628, time: 0.345000 sec, score: 0  move b1c3
depth: 7, nodes 20992, nps: 47926, time: 0.438000 sec, score: 22  move b1c3
depth: 8, nodes 83139, nps: 46654, time: 1.782000 sec, score: 12  move b1c3
#
#
################################################################################
// Zeta OpenCL Chess config file for device: GeForce GTX 1080 Ti
################################################################################
threadsX: 1;
threadsY: 1;
nodes_per_second: 46654;
tt1_memory: 1024; // in MB
tt2_memory: 768; // in MB
opencl_platform_id: 0;
opencl_device_id: 0;
opencl_gpugen: 2;
################################################################################
##### Above output was saved in file config_0_0_.txt
##### rename it to config.txt to let engine use it
#

D:\>dir *.txt
 Volume in drive D is Backup
 Volume Serial Number is 7C3E-C707

 Directory of D:\

24-Jun-19  04:42 PM             1,089 config.txt
24-Jun-19  04:50 PM             1,092 config_0_0_.txt
               2 File(s)          2,181 bytes
               0 Dir(s)  1,226,287,964,160 bytes free

D:\>del config.txt

D:\>renconfig_0_0_.txt config.txt
'renconfig_0_0_.txt' is not recognized as an internal or external command,
operable program or batch file.

D:\>ren config_0_0_.txt config.txt

D:\>zeta.exe -l
#> Zeta 099l
#> Experimental chess engine written in OpenCL.
#> Copyright (C) 2011-2019 Srdja Matovic, Montenegro
#> This is free software, licensed under GPL >= v2
#> engine is initialising...
feature done=0
#> ...finished basic inits.
new
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
sd 12
st 2000
benchsmp
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 1 ###
depth score time nodes pv
1 60 301 20 b1c3
2 0 301 79 b1c3 b8c6
3 60 303 604 b1c3 b8c6 g1f3
4 0 304 1222 b1c3 b8c6 g1f3 g8f6
5 27 309 3654 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 343 20086 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 387 41098 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3
8 12 565 124148 b1c3 b8c6 d2d4 g8f6 d4d5 c6e5 g1f3 e5f3
9 20 1040 350975 b1c3 b8c6 e2e3 d7d6 g1f3 g7g6 f1d3 g8f6 b2b3
10 18 3982 1767591 b1c3 b8c6 e2e3 d7d6 g1f3 g7g6 f1b5 g8f6 b5c6 b7c6
11 10 8833 4094110 b1c3 b8c6 e2e3 g8f6 g1f3 e7e6 f1b5 f8d6 g2g3 a7a6 b5c4
12 9 21390 10134617 b1c3 b8c6 d2d4 g8f6 d4d5 c6e5 f2f4 e5c4 g1f3 c7c6 e2e4 c4d6
#10134617 searched nodes in 213.908000 seconds, with 197117 ttmovehits, and 134853 ttscorehits, 1445 iidhits, ebf: 3.458663, nps: 47378
### workers     #nps            #nps speedup    #time in s      #ttd speedup    #relative ttd speedup ###
### 1           47378           1.000000        213.908000      1.000000        1.000000
Top
smatovic
Posts: 2642
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by smatovic »

pferd wrote: ↑Mon Jun 24, 2019 4:58 pm I have an 1070 running archlinux with latest nvidia drivers and my display was connected throughout the process. The results are attached to this message.
Cool, thanks, accoring to the results, the patch was successfull on Pascal,
15 Compute Units run each 4 workers, 60 workers in total, thanks.

But it would be nice to see some results with more than 64 workers,
cos then the RMO parallel search steps in...

--
Srdja
Last edited by smatovic on Mon Jun 24, 2019 5:33 pm, edited 2 times in total.
Top
smatovic
Posts: 2642
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by smatovic »

Joost Buijs wrote: ↑Mon Jun 24, 2019 4:59 pm All I have a 1080 TI which is Pascal (Asus 1080TI Turbo with default settings).
Maybe these results are useful to you.

Code: Select all

...
D:\>zeta.exe -l p 0 d 0 --guessconfig
...
Thanks, but a remark, you had an typo, '--guessconfig' instead of '--guessconfigx',
without the x the command tests only for minimal setup.

--
Srdja
Top
Chessqueen
Posts: 5578
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by Chessqueen »

smatovic wrote: ↑Mon Jun 24, 2019 5:31 pm
Joost Buijs wrote: ↑Mon Jun 24, 2019 4:59 pm All I have a 1080 TI which is Pascal (Asus 1080TI Turbo with default settings).
Maybe these results are useful to you.

Code: Select all

...
D:\>zeta.exe -l p 0 d 0 --guessconfig
...
Thanks, but a remark, you had an typo, '--guessconfig' instead of '--guessconfigx',
without the x the command tests only for minimal setup.

--
Srdja
This people might test it for you, if you ask them :mrgreen: :roll:
https://tcec.chessdom.com/
Do NOT worry and be happy, we all live a short life :roll:
Top
Joost Buijs
Posts: 1563
Joined: Thu Jul 16, 2009 10:47 am
Location: Almere, The Netherlands

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by Joost Buijs »

smatovic wrote: ↑Mon Jun 24, 2019 5:31 pm
Joost Buijs wrote: ↑Mon Jun 24, 2019 4:59 pm All I have a 1080 TI which is Pascal (Asus 1080TI Turbo with default settings).
Maybe these results are useful to you.

Code: Select all

...
D:\>zeta.exe -l p 0 d 0 --guessconfig
...
Thanks, but a remark, you had an typo, '--guessconfig' instead of '--guessconfigx',
without the x the command tests only for minimal setup.

--
Srdja
You are right of course, must be the heat, it is 32 deg. C on my attic today and I'm not used to these kind of temperatures.
I reran the test with --guessconfigx

Code: Select all

D:\>zeta.exe -l --dl
#> ### Query the OpenCL Devices on Host...
#>
#> Number of OpenCL Platforms found: 1
#>
#> platform_id:0, device_id: 0, Vendor: NVIDIA Corporation, Device name: GeForce GTX 1080 Ti
#>

D:\>zeta.exe -l -p 0 -d 0 --guessconfigx
#>
#> ### Query the OpenCL Platforms on Host...
#>
#> Number of OpenCL Platforms found: 1
#>
#> Platform: 0,  Vendor:  NVIDIA Corporation
#>
#> ### Query the OpenCL Devices on Platform...
#>
#> Number of OpenCL Devices found: 1
#>
#> ### Query and check the OpenCL Device...
#>
#> Device: 0, Device name: GeForce GTX 1080 Ti
#>
#> OK, Device Endianness is little
#> OK, CL_DEVICE_MAX_COMPUTE_UNITS: 28
#> OK, CL_DEVICE_MAX_MEM_ALLOC_SIZE: 2816 MB >= 128 MB
#> OK, CL_DEVICE_GLOBAL_MEM_SIZE: 11264 MB
#> OK, Device extension cl_khr_global_int32_base_atomics is supported.
#> OK, Device extension cl_khr_local_int32_base_atomics is supported.
#> OK: Device extension cl_khr_int64_extended_atomics not supported.
#> OK, CL_DEVICE_MAX_WORK_GROUP_SIZE: 1024 >= 64
#> OK, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS: 3 >= 3
#> OK, CL_DEVICE_MAX_WORK_ITEM_SIZES [3]: 64 >= 64
#> OK, CL_DEVICE_AVAILABLE: CL_TRUE
#
#> ### Running NPS-Benchmark for minimal config on device,
#> ### this can last about 4 seconds...
#> ### threadsX: 1
#> ### threadsY: 1
#> ### total work-groups: 1
#> ### total threads: 64
#
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
depth: 1, nodes 20, nps: 6, time: 3.039000 sec, score: 60  move b1c3
depth: 2, nodes 27, nps: 27000, time: 0.001000 sec, score: 0  move a1a1
depth: 3, nodes 525, nps: 30882, time: 0.017000 sec, score: 60  move b1c3
depth: 4, nodes 618, nps: 38625, time: 0.016000 sec, score: 0  move b1c3
depth: 5, nodes 2432, nps: 50666, time: 0.048000 sec, score: 27  move b1c3
depth: 6, nodes 16432, nps: 61543, time: 0.267000 sec, score: 0  move b1c3
depth: 7, nodes 21004, nps: 60881, time: 0.345000 sec, score: 22  move b1c3
depth: 8, nodes 83128, nps: 59718, time: 1.392000 sec, score: 12  move b1c3
#
#> ### Running NPS-Benchmark for best config,
#> ### this can last about some minutes...
#
#
#> ### Running NPS-Benchmark for threadsY on device,
#> ### this can last about 4 seconds...
#> ### threadsX: 28
#> ### threadsY: 1
#> ### total work-groups: 28
#> ### total threads: 1792
#
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
depth: 1, nodes 560, nps: 183, time: 3.056000 sec, score: 60  move b1c3
depth: 2, nodes 1643, nps: 1643000, time: 0.001000 sec, score: 0  move b1c3
depth: 3, nodes 2940, nps: 2940000, time: 0.001000 sec, score: 60  move b1c3
depth: 4, nodes 4395, nps: 4395000, time: 0.001000 sec, score: 0  move b1c3
depth: 5, nodes 9073, nps: 533705, time: 0.017000 sec, score: 27  move b1c3
depth: 6, nodes 49017, nps: 1531781, time: 0.032000 sec, score: 0  move b1c3
depth: 7, nodes 63042, nps: 1970062, time: 0.032000 sec, score: 22  move b1c3
depth: 8, nodes 263873, nps: 2094230, time: 0.126000 sec, score: 21  move b1c3
depth: 9, nodes 776125, nps: 1979910, time: 0.392000 sec, score: 15  move b1c3
depth: 10, nodes 3482884, nps: 2024932, time: 1.720000 sec, score: 10  move b1c3
#
#> ### Running NPS-Benchmark for threadsY on device,
#> ### this can last about 4 seconds...
#> ### threadsX: 28
#> ### threadsY: 2
#> ### total work-groups: 56
#> ### total threads: 3584
#
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
depth: 1, nodes 1120, nps: 367, time: 3.047000 sec, score: 60  move b1c3
depth: 2, nodes 3284, nps: 3284000, time: 0.001000 sec, score: 0  move b1c3
depth: 3, nodes 5599, nps: 5599000, time: 0.001000 sec, score: 60  move b1c3
depth: 4, nodes 8038, nps: 8038000, time: 0.001000 sec, score: 0  move b1c3
depth: 5, nodes 17516, nps: 1030352, time: 0.017000 sec, score: 27  move b1c3
depth: 6, nodes 66407, nps: 4150437, time: 0.016000 sec, score: 0  move b1c3
depth: 7, nodes 94972, nps: 2877939, time: 0.033000 sec, score: 22  move b1c3
depth: 8, nodes 450470, nps: 3575158, time: 0.126000 sec, score: 0  move b1c3
depth: 9, nodes 1284792, nps: 4104766, time: 0.313000 sec, score: 18  move g1f3
depth: 10, nodes 5831075, nps: 4188990, time: 1.392000 sec, score: 9  move g1f3
#
#> ### Running NPS-Benchmark for threadsY on device,
#> ### this can last about 4 seconds...
#> ### threadsX: 28
#> ### threadsY: 4
#> ### total work-groups: 112
#> ### total threads: 7168
#
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
depth: 1, nodes 2240, nps: 727, time: 3.079000 sec, score: 60  move b1c3
depth: 2, nodes 4581, nps: 4581000, time: 0.001000 sec, score: 0  move b1c3
depth: 3, nodes 9767, nps: 9767000, time: 0.001000 sec, score: 60  move b1c3
depth: 4, nodes 15392, nps: 15392000, time: 0.001000 sec, score: 0  move b1c3
depth: 5, nodes 29582, nps: 1740117, time: 0.017000 sec, score: 27  move b1c3
depth: 6, nodes 104777, nps: 6548562, time: 0.016000 sec, score: 0  move b1c3
depth: 7, nodes 193558, nps: 6048687, time: 0.032000 sec, score: 22  move b1c3
depth: 8, nodes 650407, nps: 8130087, time: 0.080000 sec, score: 12  move b1c3
depth: 9, nodes 1744394, nps: 7965269, time: 0.219000 sec, score: 18  move b1c3
depth: 10, nodes 7018388, nps: 8305784, time: 0.845000 sec, score: 12  move b1c3
depth: 11, nodes 11966859, nps: 8505230, time: 1.407000 sec, score: 12  move b1c3
#
#> ### Running NPS-Benchmark for threadsY on device,
#> ### this can last about 4 seconds...
#> ### threadsX: 28
#> ### threadsY: 8
#> ### total work-groups: 224
#> ### total threads: 14336
#
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
depth: 1, nodes 4480, nps: 1459, time: 3.070000 sec, score: 60  move b1c3
depth: 2, nodes 9190, nps: 540588, time: 0.017000 sec, score: 0  move b1c3
depth: 3, nodes 19380, nps: 19380000, time: 0.001000 sec, score: 60  move b1c3
depth: 4, nodes 29557, nps: 1847312, time: 0.016000 sec, score: 0  move b1c3
depth: 5, nodes 56200, nps: 56200000, time: 0.001000 sec, score: 27  move b1c3
depth: 6, nodes 161859, nps: 10116187, time: 0.016000 sec, score: 0  move b1c3
depth: 7, nodes 307081, nps: 9305484, time: 0.033000 sec, score: 22  move b1c3
depth: 8, nodes 818674, nps: 10362962, time: 0.079000 sec, score: 12  move b1c3
depth: 9, nodes 2801972, nps: 12736236, time: 0.220000 sec, score: 19  move b1c3
depth: 10, nodes 9588683, nps: 12261742, time: 0.782000 sec, score: 14  move b1c3
depth: 11, nodes 16687765, nps: 12709645, time: 1.313000 sec, score: 12  move b1c3
#
#
################################################################################
// Zeta OpenCL Chess config file for device: GeForce GTX 1080 Ti
################################################################################
threadsX: 28;
threadsY: 4;
nodes_per_second: 8505230;
tt1_memory: 1024; // in MB
tt2_memory: 768; // in MB
opencl_platform_id: 0;
opencl_device_id: 0;
opencl_gpugen: 2;
################################################################################
##### Above output was saved in file config_0_0_.txt
##### rename it to config.txt to let engine use it
#

D:\>zeta.exe -l
#> Zeta 099l
#> Experimental chess engine written in OpenCL.
#> Copyright (C) 2011-2019 Srdja Matovic, Montenegro
#> This is free software, licensed under GPL >= v2
#> engine is initialising...
feature done=0
#> ...finished basic inits.
new
###ABCDEFGH###
#8 rnbqkbnr
#7 pppppppp
#6 --------
#5 --------
#4 --------
#3 --------
#2 PPPPPPPP
#1 RNBQKBNR
###ABCDEFGH###
#fen: rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
sd 12
st 2000
benchsmp
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 1 ###
depth score time nodes pv
1 60 303 20 b1c3
2 0 303 79 b1c3 b8c6
3 60 304 604 b1c3 b8c6 g1f3
4 0 306 1222 b1c3 b8c6 g1f3 g8f6
5 27 309 3654 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 337 20086 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 372 41070 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3
8 12 512 124113 b1c3 b8c6 d2d4 g8f6 d4d5 c6e5 g1f3 e5f3
9 20 890 350838 b1c3 b8c6 e2e3 d7d6 g1f3 g7g6 f1d3 g8f6 b2b3
10 18 3206 1767397 b1c3 b8c6 e2e3 d7d6 g1f3 g7g6 f1b5 g8f6 b5c6 b7c6
11 10 7059 4093982 b1c3 b8c6 e2e3 g8f6 g1f3 e7e6 f1b5 f8d6 g2g3 a7a6 b5c4
12 9 17246 10217940 b1c3 b8c6 d2d4 g8f6 d4d5 c6e5 f2f4 e5c4 g1f3 c7c6 e2e4 c4d6
#10217940 searched nodes in 172.467000 seconds, with 198441 ttmovehits, and 136013 ttscorehits, 1446 iidhits, ebf: 3.460842, nps: 59245
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 2 ###
depth score time nodes pv
1 60 304 40 b1c3
2 0 304 158 b1c3 b8c6
3 60 306 798 b1c3 b8c6 g1f3
4 0 306 1580 b1c3 b8c6 g1f3 g8f6
5 27 309 4238 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 323 21269 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 342 43177 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3
8 10 465 187006 b1c3 g8f6 g1f3 b8c6 d2d3 d7d6 b2b3 b7b6
9 19 603 351481 b1c3 g8f6 g1f3 d7d5 g2g3 d5d4 c3b5 b8c6 d2d3
10 11 2200 2250030 b1c3 b8c6 e2e4 g8f6 g1f3 e7e6 d2d4 b7b6 e4e5 f6g4
11 12 3740 4086381 b1c3 b8c6 g1f3 g8f6 e2e3 g7g6 b2b3 e7e6 f1d3 f8b4 g2g3
12 11 8048 9259384 b1c3 b8c6 g1f3 g8f6 e2e3 g7g6 d2d4 d7d5 f1b5 e7e6 b5c6 b7c6
#9259384 searched nodes in 80.484000 seconds, with 203948 ttmovehits, and 127848 ttscorehits, 1233 iidhits, ebf: 3.434717, nps: 115046
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 4 ###
depth score time nodes pv
1 60 304 80 b1c3
2 0 304 316 b1c3 b8c6
3 60 304 1131 b1c3 b8c6 g1f3
4 0 306 2192 b1c3 b8c6 g1f3 g8f6
5 27 307 5289 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 315 24951 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 334 67763 d2d3 b8c6 g1f3 g8f6 b1c3 d7d6 b2b3
8 4 411 243079 b1c3 g8f6 g1f3 b8c6 d2d3 d7d6 b2b3 b7b6
9 18 515 488529 b1c3 g8f6 e2e3 d7d5 b2b3 d5d4 c3b5 b8c6 g1f3
10 15 1243 2216461 b1c3 g8f6 e2e3 e7e6 g1f3 a7a6 f1c4 b8c6 g2g3 g7g6
11 12 2432 5048207 b1c3 g8f6 e2e3 e7e6 g1f3 a7a6 f1d3 b8c6 g2g3 f8b4 a2a3
12 10 5184 11636429 b1c3 g8f6 g1f3 b8c6 e2e4 e7e5 g2g3 d7d6 f1b5 c8g4 b5c6 b7c6
#11636429 searched nodes in 51.844000 seconds, with 255204 ttmovehits, and 175591 ttscorehits, 1359 iidhits, ebf: 3.495623, nps: 224450
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 8 ###
depth score time nodes pv
1 60 306 160 b1c3
2 0 306 625 b1c3 b8c6
3 60 306 1785 b1c3 b8c6 g1f3
4 0 307 3378 b1c3 b8c6 g1f3 g8f6
5 27 307 7449 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 314 29628 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 320 63330 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3
8 12 348 193219 b1c3 b8c6 d2d4 g8f6 d4d5 c6e5 g1f3 e5f3
9 18 445 663305 g1f3 d7d5 e2e3 b8c6 b1c3 g8f6 f1d3 e7e6 b2b3
10 15 806 2448627 g1f3 g8f6 b1c3 e7e6 d2d4 b8c6 e2e4 b7b6 e4e5 f6g4
11 12 1670 6745470 b1c3 b8c6 e2e4 g8f6 g1f3 d7d5 e4d5 f6d5 c3d5 d8d5 c2c3
12 10 3040 13592096 b1c3 b8c6 e2e4 e7e6 g1e2 g8f6 d2d4 d7d5 e4e5 f6d7 g2g3 b7b6
#13592096 searched nodes in 30.406000 seconds, with 296562 ttmovehits, and 209542 ttscorehits, 1523 iidhits, ebf: 3.537646, nps: 447020
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 16 ###
depth score time nodes pv
1 60 304 320 b1c3
2 0 304 1262 b1c3 b8c6
3 60 306 3171 b1c3 b8c6 g1f3
4 0 306 5790 b1c3 b8c6 g1f3 g8f6
5 27 306 11536 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 311 43748 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 314 86210 b1c3 g8f6 e2e3 b8c6 g1f3 d7d6 b2b3
8 3 331 253471 g1f3 b8c6 d2d4 g8f6 d4d5 c6b4 b1c3 e7e6
9 18 379 749366 g1f3 b8c6 b1c3 g7g6 e2e3 d7d6 f1c4 g8f6 b2b3
10 9 672 3860042 g1f3 g8f6 b1c3 d7d5 e2e3 e7e6 f1b5 b8c6 b5c6 b7c6
11 14 1023 7579606 g1f3 g8f6 b1c3 d7d5 d2d4 c8e6 c1f4 c7c6 f3g5 b8d7 g5e6
12 10 1862 16609584 g1f3 g8f6 b1c3 d7d5 d2d4 c7c6 c1f4 e7e6 f3e5 b8d7 e2e3 g7g6
#16609584 searched nodes in 18.626000 seconds, with 416639 ttmovehits, and 325715 ttscorehits, 1613 iidhits, ebf: 3.592628, nps: 891741
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 28 ###
depth score time nodes pv
1 60 306 560 b1c3
2 0 306 2207 b1c3 b8c6
3 60 306 5161 b1c3 b8c6 g1f3
4 0 306 9581 b1c3 b8c6 g1f3 g8f6
5 27 307 18768 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 311 65337 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 314 123421 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3
8 3 326 349201 b1c3 b8c6 d2d4 d7d5 g1f3 g8f6 e2e3 e7e6
9 19 368 1159194 g1f3 g8f6 b1c3 d7d5 b2b3 d5d4 c3b5 b8c6 d2d3
10 15 498 3663069 g1f3 g8f6 b1c3 d7d5 e2e3 g7g6 f1b5 b8c6 g2g3 c8g4
11 12 800 9551523 g1f3 g8f6 b1c3 d7d5 e2e3 b8c6 d2d4 e7e6 g2g3 g7g6 b2b3
12 11 1523 23866083 g1f3 g8f6 b1c3 d7d5 e2e3 a7a6 d2d4 b8c6 f1d3 e7e6 g2g3 g7g6
#23866083 searched nodes in 15.235000 seconds, with 645344 ttmovehits, and 485979 ttscorehits, 2411 iidhits, ebf: 3.694210, nps: 1566529
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 56 ###
depth score time nodes pv
1 60 306 1120 b1c3
2 0 306 4404 b1c3 b8c6
3 60 306 9860 b1c3 b8c6 g1f3
4 0 306 17296 b1c3 b8c6 g1f3 g8f6
5 27 307 34994 b1c3 b8c6 g1f3 g8f6 d2d3
6 0 309 105430 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6
7 22 312 191223 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3
8 12 323 630170 b1c3 b8c6 d2d4 d7d5 g1f3 g8f6 e2e3 e7e6
9 20 356 1881391 b1c3 b8c6 g1f3 g8f6 d2d3 d7d6 b2b3 b7b6 g2g3
10 14 526 8817609 b1c3 b8c6 g1f3 g8f6 e2e4 d7d5 f1d3 e7e6 b2b3 g7g6
11 12 759 18428768 d2d4 d7d5 b1c3 g8f6 g1f3 e7e6 g2g3 b8c6 e2e3 g7g6 b2b3
12 12 1617 53837682 e2e4 b8c6 b1c3 e7e6 d2d4 g8e7 d4d5 e6d5 e4d5 c6e5 g1f3 d7d6
#53837682 searched nodes in 16.173000 seconds, with 1357767 ttmovehits, and 1276483 ttscorehits, 4809 iidhits, ebf: 3.932774, nps: 3328861
### doing inits for benchsmp depth 12: ###
### computing benchsmp depth 12: ###
### work-groups: 112 ###
depth score time nodes pv
1 60 306 2240 b1c3
2 0 306 6839 b1c3 b8c6
3 60 306 16729 b1c3 b8c6 g1f3
4 0 307 32151 b1c3 b8c6 g1f3 g8f6
5 27 307 61121 b1c3 b8c6 g1f3 g8f6 e2e3
6 0 309 174494 b1c3 b8c6 g1f3 g8f6 e2e3 d7d6
7 22 312 355388 b1c3 b8c6 g1f3 g8f6 e2e3 d7d6 g2g3
8 0 322 968567 b1c3 b8c6 d2d3 g8f6 g2g3 b7b6 g1f3 e7e6
9 14 348 3145129 g1f3 g8f6 b1c3 d7d6 g2g3 b8c6 d2d3 c8g4 b2b3
10 11 440 10494952 d2d4 g8f6 b1c3 d7d6 g1f3 c8g4 e2e4 b8d7 h2h3 g4e6
11 12 578 21800047 b1c3 g8f6 g1f3 d7d5 d2d4 b8c6 e2e3 e7e6 g2g3 g7g6 b2b3
12 13 1020 58420612 e2e4 b8c6 b1c3 g8f6 d2d4 d7d5 e4e5 f6e4 c3e4 d5e4 g1e2 c8e6
#58420612 searched nodes in 10.204000 seconds, with 1609452 ttmovehits, and 1287163 ttscorehits, 5175 iidhits, ebf: 3.957566, nps: 5725265
### workers     #nps            #nps speedup    #time in s      #ttd speedup    #relative ttd speedup ###
### 1           59245           1.000000        172.467000      1.000000        1.000000
### 2           115046          1.941869        80.484000       2.142873        2.142873
### 4           224450          3.788505        51.844000       3.326653        1.552427
### 8           446785          7.541312        30.422000       5.669154        1.704161
### 16          891741          15.051751       18.626000       9.259476        1.633308
### 28          1564886         26.413807       15.251000       11.308570       1.221297
### 56          3325777         56.135995       16.188000       10.654003       0.942118
### 112         5716302         96.485813       10.220000       16.875440       1.583953
Top
smatovic
Posts: 2642
Joined: Wed Mar 10, 2010 10:18 pm
Location: Hamburg, Germany
Full name: Srdja Matovic

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by smatovic »

Cool, so the patch is confirmed to work on Pascal gpus,
maybe there is some volunteer with an RTX Turing card?

--
Srdja
Top
j.korhonen
Posts: 19
Joined: Tue Feb 26, 2019 12:34 am
Full name: Juhani Korhonen

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by j.korhonen »

Here: viewtopic.php?f=6&t=71021
Guy from CCRL curious about finalising of engine
Top
User avatar
Guenther
Posts: 4605
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: Looking for someone to test Zeta v099l on RTX 2080 TI, or similar, gpu

Post by Guenther »

smatovic wrote: ↑Mon Jun 24, 2019 3:17 pm Zeta v099k does not scale well on Nvidia Pascal and Turing gpus, so I wrote a
patch that could fix this issue, but am not able to test it.

If you have an RTX 2080 TI, or similar, running, you may want to run a benchmark
for me, takes only a couple of minutes.

Binary - Windows 64 and Linux 64 bit:
https://github.com/smatovic/Zeta/tree/v099l/bin

Code:
https://github.com/smatovic/Zeta/tree/v099l

...

Thanks in advance,
Srdja
Hello Srdja, I have finally added Zeta to the XB/UCI chronology. (Zeta Dva was always there ofc)
Note that it originally was designed for cpu only programs (gpu programs did not exist in the beginning anyway).
Finally I have added another column for 'H Type' (Hardware Type = CPU/GPU).

When was the first release of a Zeta version (binary) working on some ordinary GPU?
I am asking because you have hidden/deleted all previous versions before 0.99k and it was probably during my long hiatus between 2011-2015.
(I can still find some of them with working downloads from the source commits directly though,
but I don't know if they would run or if you want them to be acknowledged in the chronology.
Edit: It seems even github starts just in 2016 with 0.98g?)

BTW is there a faster way to scroll through older/newer commits in github, it has bothered me forever to be unable to see them all at once
in a list and instead have to go through dozens of pages with max 34 entries.

Thanks in advance.

A forum search for 'Zeta release' goes back until 2011:
search.php?st=0&sk=t&sd=d&sr=posts&keyw ... ta+release
https://rwbc-chess.de

trollwatch:
Chessqueen + chessica + AlexChess + Eduard + Sylwy
Top

Return to “Computer Chess Club: General Topics”