search

manyoso / allie

Allie: A UCI compliant chess engine
GNU General Public License v3.0
104 stars 21 forks source link

Allie 0.2 running on Windows 7 64-bit and CPU Intel Core i 3570 Quad instead of GPU? #2

Open yellow-panther opened 5 years ago

yellow-panther commented 5 years ago

Hello manyoso,

I obtained the following error messages during installation of https://developer.nvidia.com/cuda-downloads?target_os=Windows&target_arch=x86_64 ( NVIDIA CUDA 10.1 Toolkit:) GPU Library Advisor - CURAND Development / Runtime - CUDART Runtime- NPP Runtine - CUPTI - CUSOLVER Development - CUFFT Development / Runtime - NVRTC Development / Runtime - CUSOLVER Runtime - CUBLAS Development - Demo Suite - NVGRAPH Development / Runtime - NVCC - NPP Development - Visual Profiler - DISASSEMBLER - CUDA Profiler Tools - Fortran Examples - NVML Development - Nsight Systems / Compute - Samples - CUSPARSE / CUBLAS Runtime - PhysX System Software - CUDA Documentation / Sanitizer API - MEMCHECK - Occupancy Calculator- NOT INSTALLED Graphics Driver (= Intel Standard VGA 1280x1025) could not find compatible hardware

I added cudart64_100.dll and cudnn64_7.dll, but nothing happens. What can be done in such a case? Or is Allie designed only for GPU, and does not support CPU-structure? cublas64_100.dll is still required for download to run the executable file.

Regards,

uwuplant commented 5 years ago

What GPU do you have?

yellow-panther commented 5 years ago

only a quad-core cpu, no gpu

Am Do., 11. Apr. 2019 um 23:01 Uhr schrieb Evan notifications@github.com:

What GPU do you have?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/manyoso/allie/issues/2#issuecomment-482312387, or mute the thread https://github.com/notifications/unsubscribe-auth/AQ5qH9OS6pp98upVsyr8FEoaQXJKha8Iks5vf6KZgaJpZM4cj0H7 .

uwuplant commented 5 years ago

You need a GPU supporting CUDA to run Allie.

yellow-panther commented 5 years ago

LCZero https://github.com/LeelaChessZero/lc0/releases is designed to run on GPU and CPU -systems

Can you create such a blas-version for CPUs and Windows 64-bit version, too?

cf. text below:

LC Zero with ChessBase GUI Since v0.7 no adapter is needed (inBetween, polyglot) for running leela in Fritz. Just do this: 1) Create a Leela-folder somewhere on your PC in a folder, which is not write-protected (normally C:\ProgramFiles\ChessBase\Engines is write-protected!) - I created it in my documents-folder. 2) download the Windows GPU Leela files here: https://github.com/glinscott/leela-chess/releases - https://github.com/LeelaChessZero/lc0/releases 3) Unzip all files in your leela-folder 4) Download the Network-file here (but NOT the first on top of the list (the newest)): http://162.217.248.187/networks 5) unpack the network-file, you get a folder. In it a file. Rename it to weights.txt 6) Copy the weights.txt file in your leela-folder 7) start the lczero.exe -file one time with mouse-double-click. That creates a file called "leelaz_opencl_tuning", which is needed! (If you want to run Leela faster, create a small .bat file with one line in it "lczero.exe -w weights.txt --full-tuner" and start that .bat file with double-click. Then you have to wait around 20-30 minutes, while Leela tries around 5000 GPU-configurations in order to find the fastest.) 8) Start Fritz and create UCI-engine with lczero.exe. It is recommended, to set the number of Threads to the number of CPUs of your system, that makes Leela a little bit faster, although she uses the GPU for main calculations. (Thread is 2 by default) GitHub If you don't have a gpu you need the blas version: https://github.com/LeelaChessZero/lc0/releases/download/v0.18.0/lc0-v0.18.0-windows-blas.zip Note that this is a lot slower than the gpu versions. You will also need a weights file, 11248 is our choice for the upcoming competitions: http://lczero.org/get_network?sha=994de36c13ddf2540bf588817fcca0cb4e279e61c7d118c3bccc9c74419caecd Just copy it in the same folder with lc0.exe, no need to unpack or rename. Also lc0 will not create leelaz_opencl_tuning, so go from step 4 to step 8 in the above procedure.