cuFFT not working in release mode From SDK2.0 , All but cuFFT is fine

Thanks NVIDIA for a great card and software.

Everything is fine in EMU mode.

I just compiled the projects files and all the programs are working except the one including CuFFT. the error messages are:

OceanFFT

[CUDA FFT Ocean Simulation]

Left mouse button - rotate

Middle mouse button - pan

Left + middle mouse button - zoom

‘w’ key - toggle wireframe

[CUDA FFT Ocean Simulation]

CUFFT error in file ‘oceanFFT.cpp’ in line 337.

or:

simpleFFT

cufftSafeCall() CUFFT error in file <simpleCUFFT.cu>, line 124.

or:

convolutionFFT2D

Input data size : 1000 x 1000

Convolution kernel size : 7 x 7

Padded image size : 1006 x 1006

Aligned padded image size : 1024 x 1024

Allocating memory…

Generating random input data…

Creating FFT plan for 1024 x 1024…

cufftSafeCall() CUFFT error in file <convolutionFFT2D.cu>, line 231.

I am using a 9600 GSO with 768MB AMD64

Under Linux Slackware64(beta)

I did tried the prevoius versions : they would not compile.

any advice?

I think that I have located the problem in the definition of the Complex functions. There seems to be some memory leaks to prevent the proper transfert of data to the GPU memory. It may have to do with conversion from Complex to float. Does it have something to do with the fact that my GPU is 1.1? Only the projects with Complex functions behave that way.

I know that I am using some beta software, but I would like to know if anyone else had problems using the cuFFT library. Does it only work with double-precision card?

You probably have an incorrect installation of either the driver or the toolkit (wrong LD_LIBRARY_PATH, etc).

This ONLY happen with cuFFT, where Complex is invoved.

Hi,

Just wondering why you’re still using SDK2.0?

N.

It is SDK 2.2, How do I correct this in the mini title?

Now, I found that this driver/card combo does not accept float2 data type.

This happended when cudaMalloc is called.

And what driver and toolkit are you using?

This is my setup:

NVIDIA-Linux-x86_64-180.22-pkg2.run

cudadriver_2.2_linux_64_185.18.08-beta.run

cudagdb_2.2_linux_64_rhel5.3.run

cudasdk_2.2_linux.run

cudatoolkit_2.2_linux_64_rhel5.3.run

uhhh, are you using 180.22 or 185.18.08

This could be the reason.

I’m using NVIDIA-Linux-x86_64-185.18.14-pkg2.run without problems.

N.

You are absolutely right. I had this version in another directory. I installed it and now everything is working fine.

Thank you for your help.