New version of Allie v0.6

[corres]

Allie uses only one PV line. Why?
Allie does not display TBhits. Why?
More PV lines are important to analysis.
Number of TBhits would give more info about working of Allie.

[gonzochess75]

Multipv is coming and Allie does display TB hits already and has for some time. If you don’t see them, then something wrong with your TB setup or Allie option... come to allie discord and I can try and help

[corres]

Multipv is coming and Allie does display TB hits already and has for some time. If you don’t see them, then something wrong with your TB setup or Allie option... come to allie discord and I can try and help

Sometimes Allie shows TBhits, so I think Allie rare use TB, even in the case of few figures are on the table.
In general it lacks the info about behavior of Allie.
Discord is a tool for making connection between developers, so at least on Github.com we need a more detailed write down on the behavior of Allie.

[corres]

I run a shortened test between Allie ver.0.6 and Leela ver.0.25.
During the test Allie stopped to work with the note: "Allie exe caused an exception".
The result:
Leela ver.0.25 : Allie ver.0.6 = 12 : 3 (37 draw) 52 games.
Allie 0.6 was in default settings
Leela was in the old default settings (= not Kiudee!), excluding NNCachesize = 10000000.
Net was Jhorthos J13B.2-188.
TC was 1 min + 2 sec / move.
Start position were 2 x 26 with altered colors.
GPU was RTX 2080 Ti OC.

[MikeB]

Hi,

I just released a new version of Allie: https://github.com/manyoso/allie/releases/tag/v0.6

You can find windows exe on Allie discord channel under #releases. This version will be playing in the upcoming TCEC Cup and is substantially stronger and faster than v0.5. Major parts of the code were rewritten and perft tests implemented along with NPS increases due to lockless threading redesign and lots of changes to TM.

Cheers!

Thanks for letting us know. Now if there were only some instructions on how to build it, how to download it, how to install ...?

[MikeB]

Hi,

I just released a new version of Allie: https://github.com/manyoso/allie/releases/tag/v0.6

You can find windows exe on Allie discord channel under #releases. This version will be playing in the upcoming TCEC Cup and is substantially stronger and faster than v0.5. Major parts of the code were rewritten and perft tests implemented along with NPS increases due to lockless threading redesign and lots of changes to TM.

Cheers!

Thanks for letting us know. Now if there were only some instructions on how to build it, how to download it, how to install ...?

This is how far I get when trying to build it ...

Code: Select all

MichaelB7@VM-894787 MINGW64 ~/home/Github/Allie/src (master)
$ qmake
Info: creating stash file C:\Users\MichaelB7\home\Github\Allie\src\.qmake.stash

MichaelB7@VM-894787 MINGW64 ~/home/Github/Allie/src (master)
$ make
C:/Program Files/Git/mingw64/bin/make.exe -f Makefile.Release
make[1]: Entering directory 'C:/Users/MichaelB7/home/Github/Allie/src'
"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin\nvcc" -c -Xcompiler -fno-keep-inline-dllexport,-MD -I "C:/Users/MichaelB7/home/Github/Allie/src/../lib" -I "C:/cache/zlib_build/include" -I "C:\cache\protobuf_build\include" -I "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include" -I "C:\cache\cuda\include" -I "C:/Users/MichaelB7/home/Github/Allie/lib" ../lib/neural/cuda/common_kernels.cu -o common_kernels_cuda.obj
cl : Command line warning D9002 : ignoring unknown option '-fno-keep-inline-dllexport'
common_kernels.cu
cl : Command line warning D9002 : ignoring unknown option '-fno-keep-inline-dllexport'
common_kernels.cu
common_kernels.cu
cl : Command line warning D9002 : ignoring unknown option '-fno-keep-inline-dllexport'
"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\bin\nvcc" -arch=compute_70 -code=sm_70 -c -Xcompiler -fno-keep-inline-dllexport,-MD -I "C:/Users/MichaelB7/home/Github/Allie/src/../lib" -I "C:/cache/zlib_build/include" -I "C:\cache\protobuf_build\include" -I "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include" -I "C:\cache\cuda\include" -I "C:/Users/MichaelB7/home/Github/Allie/lib" ../lib/neural/cuda/fp16_kernels.cu -o fp16_kernels_cuda.obj
cl : Command line warning D9002 : ignoring unknown option '-fno-keep-inline-dllexport'
fp16_kernels.cu
cl : Command line warning D9002 : ignoring unknown option '-fno-keep-inline-dllexport'
fp16_kernels.cu
fp16_kernels.cu
cl : Command line warning D9002 : ignoring unknown option '-fno-keep-inline-dllexport'
g++ -Wl,-s -Wl,-subsystem,console -mthreads -o ../bin/allie.exe common_kernels_cuda.obj fp16_kernels_cuda.obj release/main.o  -LC:/cache/protobuf_build/lib -L"C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.2/lib/x64" -LC:/cache/cuda/lib/x64 -LC:/Users/MichaelB7/home/Github/Allie/bin -lmargean -LC:/cache/zlib_build/lib -lzlibstatic -llibprotobuf -lcudart -lcudnn -lcublas C:/msys64/mingw64/lib/libQt5Core.dll.a
C:/msys64/mingw64/bin/../lib/gcc/x86_64-w64-mingw32/9.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find -lmargean
C:/msys64/mingw64/bin/../lib/gcc/x86_64-w64-mingw32/9.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find -lzlibstatic
C:/msys64/mingw64/bin/../lib/gcc/x86_64-w64-mingw32/9.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find -llibprotobuf
collect2.exe: error: ld returned 1 exit status
make[1]: *** [Makefile.Release:70: ../bin/allie.exe] Error 1
make[1]: Leaving directory 'C:/Users/MichaelB7/home/Github/Allie/src'
make: *** [Makefile:45: release] Error 2