When I run multifindbestfit.py, It reports an error: CUDA error: invalid configuration argument. How can I fix it ？

[laiyingxin2]

No description provided.

[Eaton2022]

python -m btcdet.datasets.multifindbestfit
No CUDA runtime is found, using CUDA_HOME='/usr/local/cuda-11.1:/usr/local/cuda-11.1:'
Traceback (most recent call last):
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/runpy.py", line 194, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/home/scy/project/btcdet/BtcDet/btcdet/datasets/multifindbestfit.py", line 23, in
from ..ops.chamfer_distance import ChamferDistance
File "/home/scy/project/btcdet/BtcDet/btcdet/ops/chamfer_distance/init.py", line 1, in
from .chamfer_distance import ChamferDistance
File "/home/scy/project/btcdet/BtcDet/btcdet/ops/chamfer_distance/chamfer_distance.py", line 11, in
cd = load(name="cd",
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1080, in load
return _jit_compile(
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1293, in _jit_compile
_write_ninja_file_and_build_library(
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1392, in _write_ninja_file_and_build_library
_write_ninja_file_to_build_library(
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1783, in _write_ninja_file_to_build_library
cuda_flags = common_cflags + COMMON_NVCC_FLAGS + _get_cuda_arch_flags()
File "/home/scy/anaconda3/envs/pcdet1/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1562, in _get_cuda_arch_flags
arch_list[-1] += '+PTX'
IndexError: list index out of range

can anybody help me?

[hiepbk]

open bashrc file and add following commands:
export PATH="/usr/local/cuda-11.6/bin:$PATH"
export LD_LIBRARY_PATH="'usr/local/cuda-11.6/lib64:$LD_LIBRARY_PATH"
export CUDA_HOME=/usr/local/cuda-11.6
replace '11.6' corresponding with your cuda version

[Castiel-Lee]

Hello, I meet the same problem. But,

'''
open bashrc file and add following commands:
export PATH="/usr/local/cuda-11.6/bin:$PATH"
export LD_LIBRARY_PATH="'usr/local/cuda-11.6/lib64:$LD_LIBRARY_PATH"
export CUDA_HOME=/usr/local/cuda-11.6
replace '11.6' corresponding with your cuda version
''''

it didn't solve it.

No CUDA runtime is found, using CUDA_HOME='/usr/local/cuda-11.1'
Traceback (most recent call last):
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/runpy.py", line 194, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/home/xxx/shuangzh_workspace/BtcDet_env/BtcDet/btcdet/datasets/multifindbestfit.py", line 21, in
from ..ops.chamfer_distance import ChamferDistance
File "/home/xxx/shuangzh_workspace/BtcDet_env/BtcDet/btcdet/ops/chamfer_distance/init.py", line 1, in
from .chamfer_distance import ChamferDistance
File "/home/xxx/shuangzh_workspace/BtcDet_env/BtcDet/btcdet/ops/chamfer_distance/chamfer_distance.py", line 11, in
cd = load(name="cd",
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1079, in load
return _jit_compile(
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1292, in _jit_compile
_write_ninja_file_and_build_library(
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1391, in _write_ninja_file_and_build_library
_write_ninja_file_to_build_library(
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1782, in _write_ninja_file_to_build_library
cuda_flags = common_cflags + COMMON_NVCC_FLAGS + _get_cuda_arch_flags()
File "/home/xxx/miniconda/envs/Btcdet/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1561, in _get_cuda_arch_flags
arch_list[-1] += '+PTX'
IndexError: list index out of range

[Rumilku]

I meet the same problem. If your gpus<4, this error may be caused by line 10 'os.environ["CUDA_VISIBLE_DEVICES"] = "3"', remove it.

[Castiel-Lee]

Thanks. It solved my problem!

[ElizabethSF]

I solve this probelm using export TORCH_CUDA_ARCH_LIST="3.5;5.0;6.0;6.1;7.0;7.5;8.0;8.6+PTX"