最近开始学习基于python核CUDA的GPU编程,发生报错error detected in the compilation of "kernel.cu",网上各种查询有的说是版本匹配问题,有的说。。。,回看自己的核函数发现是这一块的内容报错
Traceback (most recent call last):
File "E:/cuda_program/class4/LEFE1.py", line 7, in <module>
ker = SourceModule("""
File "G:\anaconda\envs\pytorchpy38\lib\site-packages\pycuda\compiler.py", line 356, in __init__
cubin = compile(
File "G:\anaconda\envs\pytorchpy38\lib\site-packages\pycuda\compiler.py", line 305, in compile
return compile_plain(source, options, keep, nvcc, cache_dir, target)
File "G:\anaconda\envs\pytorchpy38\lib\site-packages\pycuda\compiler.py", line 155, in compile_plain
raise CompileError(
pycuda.driver.CompileError: nvcc compilation of C:\Users\19087\AppData\Local\Temp\tmpkl66u61m\kernel.cu failed
[command: nvcc --cubin -arch sm_86 -m64 -IG:\anaconda\envs\pytorchpy38\lib\site-packages\pycuda\cuda kernel.cu]
[stdout:
kernel.cu(42): error: identifier "_syncthreads" is undefined
kernel.cu(26): warning #550-D: variable "cell_value" was set but never used
1 error detected in the compilation of "kernel.cu".
kernel.cu
]