beagle-dev / beagle-lib

general purpose library for evaluating the likelihood of sequence evolution on trees
MIT License
125 stars 57 forks source link

CUDA error: "Unknown error" (222) when using beagle #193

Closed yashaektefaie closed 1 year ago

yashaektefaie commented 1 year ago

I installed beagle via the tutorial for linux and ran the make test successfully.

When I run "beast -beagle-info" it can see the gpu. The output is:

--- BEAGLE RESOURCES ---

0 : CPU (x86_64) Flags: PRECISION_SINGLE PRECISION_DOUBLE COMPUTATION_SYNCH EIGEN_REAL EIGEN_COMPLEX SCALING_MANUAL SCALING_AUTO SCALING_ALWAYS SCALERS_RAW SCALERS_LOG VECTOR_SSE VECTOR_NONE THREADING_NONE PROCESSOR_CPU FRAMEWORK_CPU

1 : Tesla V100-SXM2-16GB Global memory (MB): 16161 Clock speed (Ghz): 1.53 Number of cores: 10240 Flags: PRECISION_SINGLE PRECISION_DOUBLE COMPUTATION_SYNCH COMPUTATION_ASYNCH EIGEN_REAL EIGEN_COMPLEX SCALING_MANUAL SCALING_AUTO SCALING_ALWAYS SCALERS_RAW SCALERS_LOG VECTOR_NONE THREADING_NONE PROCESSOR_GPU FRAMEWORK_CUDA

2 : Tesla V100-SXM2-16GB (OpenCL 1.2 CUDA) Global memory (MB): 16160 Clock speed (Ghz): 1.53 Number of compute units: 80 Flags: PRECISION_SINGLE PRECISION_DOUBLE COMPUTATION_SYNCH COMPUTATION_ASYNCH EIGEN_REAL EIGEN_COMPLEX SCALING_MANUAL SCALING_AUTO SCALING_ALWAYS SCALERS_RAW SCALERS_LOG VECTOR_NONE THREADING_NONE PROCESSOR_GPU FRAMEWORK_OPENCL

However, when I go to run the actual command "beast/bin/beast -beagle_gpu -beagle_order 1 input.xml" I get a weird error:

"CUDA error: "Unknown error" (222) from file </home/beagle-lib/libhmsbeagle/GPU/GPUInterfaceCUDA.cpp>, line 284."

This error happens after the alignment is loaded:

....

Random number seed: 1676347048680

File: hcv_coal.xml seed: 1676347048680 threads: 1 Loading package BEAST v2.6.0 Loading package BEAST v2.6.0 Alignment(hcv) 63 taxa 411 sites 246 patterns

CUDA error: "Unknown error" (222) from file </home/ye12/tools/beagle-lib/libhmsbeagle/GPU/GPUInterfaceCUDA.cpp>, line 284.

Has anyone ever run into an error like this? I don't understand since everything else works up to this point.

yashaektefaie commented 1 year ago

Never mind this was because I was using the wrong device in "-beagle_order" I should have used "-beagle_order 2" since this is the instance with the cuda.