search

ccsb-scripps / AutoDock-GPU

AutoDock for GPUs and other accelerators
https://ccsb.scripps.edu/autodock
GNU General Public License v2.0
391 stars 107 forks source link

about the error #249

Open c00jsw00 opened 10 months ago

c00jsw00 commented 10 months ago

Dear Sir, I got the error as shown below, but I can succeed with the adgpu-v1.5.3_linux_ocl_128wi. Could you tell me how to fix it ? Thanks YT

c00jsw00@localhost:~/autodock/x86_64Linux2/src/bin/content> ../autodock_gpu_128wi -C 1 --import_dpf receptor.reactive_config --flexres receptor_flex.pdbqt -L Nirmatrelvir.pdbqt AutoDock-GPU version: 00e3113dd11a6bd12bab4cf8a7ee3f88e3f197c1-dirty

Running 1 docking calculation (contact analysis cutoffs: R=2.1 Å, H=3.7 Å, V=4.0 Å)

Cuda device: NVIDIA GeForce RTX 4090 Available memory on device: 23827 MB (total: 24214 MB)

CUDA Setup time 0.582589s

Running Job #1 Using heuristics: (capped) number of evaluations set to 5210753 Local-search chosen method is: ADADELTA (ad) SetKernelsGpuData copy to cData failed invalid device symbol autodock_gpu_128wi: ./cuda/kernels.cu:130: void SetKernelsGpuData(GpuData*): Assertion `0' failed. Aborted (core dumped)

atillack commented 8 months ago

@c00jsw00 When compiling for Cuda you need to specify the GPU target architecture, in the case of the 4090 this is compute architecture 8.9 - so compiling with make DEVICE=GPU NUMWI=128 OVERLAP=ON TARGETS="89" should fix the above error.