Setting up the Environment

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi all,

I’m trying to set up my paths to allow compiling to work. Ideally I would like to be able to compile in both Visual C++ express and at the command line but at present neither is working. I was wondering if someone could tell me if my environment variables are correct. I’ve listed them below:

Visual Studio

I have added the following to the VC++ Directories section in options

Executable Files :

C:\CUDA\Bin

Include Files:

C:\CUDA\include

C:\CUDA\SDK\common\inc

Library Files:

$(CUDA_LIBRARIES) cuda.lib cudart.lib

C:\CUDA\lib

My guess is this is not correct but please let me know!

In the system environment variables I have set the following:

CUDA_BIN_PATH C:\CUDA\bin

CUDA_INC_PATH C:\CUDA\include

CUDA_LIB_PATH C:\CUDA\lib

CUDA_LIBRARIES C:\CUDA\SDK\common\inc

PATH C:\CUDA\bin

VC_INCLUDE C:\Program Files\Microsoft Visual Studio 8\VC\include

These are all variables I have added to try and get compilation to work but I still get errors related to files not being found (when i compile simpleCUFFT it can’t find standard C libraries (stdio.h etc) and cutil.h.

Find below my error messages at a command prompt:

"C:\CUDA\bin/../include/driver_types.h", line 48: error: could not open source

          file "limits.h"

  #include <limits.h>

                     ^

"C:\CUDA\bin/../include/driver_types.h", line 49: error: could not open source

          file "stddef.h"

  #include <stddef.h>

                     ^

"C:\CUDA\bin/../include/common_functions.h", line 49: error: could not open

          source file "time.h"

  #include <time.h>

                   ^

"C:\CUDA\bin/../include/math_functions.h", line 385: error: could not open

          source file "math.h"

  #include <math.h>

                   ^

"C:\CUDA\bin/../include/math_functions.h", line 386: error: could not open

          source file "stdlib.h"

  #include <stdlib.h>

                     ^

"simpleCUFFT.cu", line 33: error: could not open source file "stdlib.h"

  #include <stdlib.h>

                     ^

"simpleCUFFT.cu", line 34: error: could not open source file "stdio.h"

  #include <stdio.h>

                    ^

"simpleCUFFT.cu", line 35: error: could not open source file "string.h"

  #include <string.h>

                     ^

"simpleCUFFT.cu", line 36: error: could not open source file "math.h"

  #include <math.h>

                   ^

"C:\CUDA\bin/../include/cufft.h", line 44: error: could not open source file

          "stdio.h"

  #include <stdio.h>

                    ^

"simpleCUFFT.cu", line 40: error: could not open source file "cutil.h"

  #include <cutil.h>

                    ^

11 errors detected in the compilation of "simpleCUFFT.cu".

Any help would be much appreciated.

Thank you,

Chris Morrison

2

Without the seeing the actual compile lines, it’s hard to say.

Try putting the paths in your environment variables in quotes.

3

Assuming you mean what Visual Studio is executing according to the property pages of the project->Configuration Properties->CUDA->Command line is

nvcc.exe -ccbin “C:\Program Files\Microsoft Visual Studio 8\VC\bin” -c -Xcompiler "/EHsc /W3 /nologo /Wp64 /O2 /Zi /MT " -o Debug\simpleCUFFT.obj simpleCUFFT.vcproj

The directories I have listed above (for environment variables) are pretty straightforward to understand and all the stuff is at these locations.

THanks,

Chris

4

Ok I’m pretty sure my mistake is with my path environments now (i wasn’t using the visual C++ command prompt and using it removed most errors). I’m using it now and i only get the following errors which I still can’t seem to solve

Y:\CUDA\Examples\fft\NVIDIA CUDA SDK\projects\simpleCUFFT>nvcc simpleCUFFT.cu

"simpleCUFFT.cu", line 121: error: identifier "CUFFT_DATA_C2C" is undefined

      cufftPlan1d(&plan, new_size, CUFFT_DATA_C2C, 1);

                                   ^

"simpleCUFFT.cu", line 124: error: identifier "cufftExecute" is undefined

      cufftExecute(plan, d_signal, d_signal, -1);

      ^

2 errors detected in the compilation of "C:\DOCUME~1\CHRISM~1\LOCALS~1\Temp/tmpx

ft_00000260_00000000-3.ii".
5

Those errors come from the API difference from 0.8 and 0.9.

CUFFT_DATA_C2C is now CUFFT_C2C

cufftExecute is now cufftExecC2C

That should work

6

Your right, that did fix these two problems. I can’t get the SDK to install on my machine (Ancient win2k box with no nvidia card) so i had to install it on a colleagues computer and try to manually copy across the files and settings. This caused me to accidentally download 0.8 rather than 0.9. I have now moved to 0.9.

I still think though that my problem is with my environment. I regularly get the error saying could not find cutil.h at the command prompt which I have fixed, for the short term, by copying the file into my project directory. This then compiles fine but then the linker doesnt work

C:\CUDA\SDK\projects\BlackScholes>nvcc BlackScholes.cu

tmpxft_00000120_00000000-8.i

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutCheckCmdLineFlag referenced in function _main

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutDeleteTimer referenced in function _main

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutGetTimerValue referenced in function _main

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutStopTimer referenced in function _main

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutStartTimer referenced in function _main

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutResetTimer referenced in function _main

tmpxft_00000120_00000000-9.obj : error LNK2019: unresolved external symbol __imp

__cutCreateTimer referenced in function _main

a.exe : fatal error LNK1120: 7 unresolved externals

and in visual studio i get the following error.

------ Build started: Project: BlackScholes, Configuration: Debug Win32 ------

Performing Custom Build Step

'C:\CUDA\bin' is not recognized as an internal or external command,

operable program or batch file.

Project : error PRJ0019: A tool returned an error code from "Performing Custom Build Step"

Build log was saved at "file://c:\CUDA\SDK\projects\BlackScholes\Debug\BuildLog.htm"

BlackScholes - 1 error(s), 0 warning(s)

========== Build: 0 succeeded, 1 failed, 0 up-to-

Thank you for your help,

Chris