Question

操作系统：win10

VS：visual stadio2015 64bit

CUDA：CUDA8.0

python：python2.7.12 64bit（pycuda）

我关注了这个网站， https://documen.tician.de/pycuda/tutorial.html#getting-started

import  pycuda.driver as cuda
import pycuda.autoinit
from pycuda.compiler import SourceModule

import numpy

a = numpy.random.randn(4,4)
a = a.astype(numpy.float32)
a_gpu = cuda.mem_alloc(a.nbytes)
cuda.memcpy_htod(a_gpu,a)#transfer the data to the GPU

#executing a kernel
#function: write code to double each entry in a_gpu.
#we write the corresponding CUDA C code, and feed it into the constructor of pycuda.compiler.SourceModule
mod = SourceModule("""
    __global__ void doublify(float *a)
    {
        int idx = threadIdx.x + threadIdx.y*4;
        a[idx] *= 2;
    }
    """)

#If there aren’t any errors, the code is now compiled and loaded onto the device. We find a reference to our pycuda.driver.Function and call it, specifying a_gpu as the argument, and a block size of 4x4:
func = mod.get_function("doublify")
func(a_gpu, block=(4,4,1))

#Finally, we fetch the data back from the GPU and display it, together with the original a:
a_doubled = numpy.empty_like(a)
cuda.memcpy_dtoh(a_doubled, a_gpu)
print a_doubled
print a

但是，失败了错误：

Traceback (most recent call last):
  File "G:/myworkspace/python2.7/cuda/test.py", line 24, in <module>
    """)
  File "D:\python2.7\lib\site-packages\pycuda\compiler.py", line 265, in __init__
    arch, code, cache_dir, include_dirs)
  File "D:\python2.7\lib\site-packages\pycuda\compiler.py", line 255, in compile
    return compile_plain(source, options, keep, nvcc, cache_dir, target)
  File "D:\python2.7\lib\site-packages\pycuda\compiler.py", line 137, in compile_plain
    stderr=stderr.decode("utf-8", "replace"))
CompileError: nvcc compilation of c:\users\gl\appdata\local\temp\tmp8poxqp\kernel.cu failed
[command: nvcc --cubin -arch sm_50 -m64 -Id:\python2.7\lib\site-packages\pycuda\cuda kernel.cu]
[stdout:
nvcc fatal   : Cannot find compiler 'cl.exe' in PATH

]

有人说要将cl.exe的目录添加到环境中。我做了，错误是一样的。我是CUDA的新手。我怎么能解决这个问题？一些建议？

我做了@citizenSNIPS建议：

添加cl.exe的路径，D：\ vs2015 \ VC \ bin。

INCLUDE = C：\ Program Files（x86）\ Windows Kits \ 10 \ Include \ 10.0.10240.0 \ ucrt。

LIB = C：\ Program Files（x86）\ Windows Kits \ 10 \ Lib \ 10.0.10240.0 \ ucrt \ x64（我找不到C：\ Program Files（x86）\ Windows Kits \ 10 \ Lib \我计算机中的10.0.10240.0 \ um \ x64。

出现了以下新错误：

raceback (most recent call last):
  File "G:\myworkspace\python2.7\cuda\test.py", line 24, in <module>
    """)
  File "D:\python2.7\lib\site-packages\pycuda\compiler.py", line 265, in __init__
    arch, code, cache_dir, include_dirs)
  File "D:\python2.7\lib\site-packages\pycuda\compiler.py", line 255, in compile
    return compile_plain(source, options, keep, nvcc, cache_dir, target)
  File "D:\python2.7\lib\site-packages\pycuda\compiler.py", line 147, in compile_plain
    + (stdout+stderr).decode("utf-8", "replace"), stacklevel=4)
  File "D:\python2.7\lib\idlelib\run.py", line 36, in idle_showwarning_subproc
    message, category, filename, lineno, line))
  File "D:\python2.7\lib\idlelib\PyShell.py", line 65, in idle_formatwarning
    s += "%s: %s\n" % (category.__name__, message)
UnicodeEncodeError: 'ascii' codec can't encode characters in position 147-168: ordinal not in range(128)

现在我正在解决这个问题，也许是因为我没有添加C：\ Program Files（x86）\ Windows Kits \ 10 \ Lib \ 10.0.10240.0 \ um \ x64？

Answer 1

您需要指定cl.exe的路径。

转到“控制面板\所有控制面板项目\系统”并选择“高级系统设置”
选择'环境价值'。
在系统变量下，找到PATH，单击编辑，然后添加cl.exe的路径。它应该是：

C：\ Program Files（x86）\ Microsoft Visual Studio 14.0 \ VC \ bin \

确保在安装visual studio时，您选择安装'c ++'编译器。它默认不安装。如果没有，请重新运行visual studio安装程序并选择安装c ++编译器。

完成后，您可能需要添加以下系统变量

INCLUDE = C：\ Program Files（x86）\ Windows Kits \ 10 \ Include \ 10.0.10240.0 \ ucrt

LIB = C：\ Program Files（x86）\ Windows Kits \ 10 \ Lib \ 10.0.10240.0 \ um \ x64; C：\ Program Files（x86）\ Windows Kits \ 10 \ Lib \ 10.0.10240.0 \ ucrt \ 64

请参阅此帖子here

Answer 2

我在.py

中添加这些代码

import sys
reload(sys)
sys.setdefaultencoding('utf8')

并运行，没有错误

Answer 3

您还可以在python文件中添加cl.exe的路径。如果您的MSVS版本发生变化，则必须更改它。

示例：

import os
if (os.system("cl.exe")):
    os.environ['PATH'] += ';'+r"C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Tools\MSVC\14.11.25503\bin\HostX64\x64"
if (os.system("cl.exe")):
    raise RuntimeError("cl.exe still not found, path probably incorrect")

编辑：您需要运行与CUDA兼容的MSVS版本。即CUDA v9.0不支持MSVS2017，CUDA v9.1仅支持15.4版本，不支持更高版本。通过从Visual Studio的Native Tools命令提示符运行nvcc.exe来尝试它是否有效。

pycuda“nvcc fatal：在PATH中找不到编译器'cl.exe'”

3 个答案: