CUFFT_INTERNAL_ERROR while running cufftPlan1d

Hi,

I’m having trouble running the below code:

#include "cufft.h"

#include "cuda_runtime.h"

#define NX 256

#define BATCH 10

int main()

{

  cufftHandle plan;

  cufftComplex *data;

  cudaMalloc ((void**)&data, sizeof(cufftComplex)*(NX)*BATCH);

  cufftPlan1d(&plan, NX, CUFFT_C2C, BATCH);

  cufftDestroy(plan);

  cudaFree(data);

}

The error message I’m seeing:

cufft: ERROR: /root/cuda-stuff/sw/gpgpu_rel1.1/cufft/src/config.cu, line 106

cufft: ERROR: CUFFT_INTERNAL_ERROR

cufft: ERROR: /root/cuda-stuff/sw/gpgpu_rel1.1/cufft/src/cufft.cu, line 94

cufft: ERROR: CUFFT_INVALID_PLAN

Does anyone have any idea why this is happening, and how can I fix it?

Thanks!

Your code is fine, I just tested on Linux with CUDA 1.1:

gcc fft.c -I/usr/local/cuda/include -L/usr/local/cuda/lib -lcufft

It runs without errors.
Which OS are you using?

Hi,

I’m using Linux 2.6.18 version.

And, I used the same command but it’s still giving me the same errors.

Thanks.

Do you have a CUDA-capable GPU?
What driver?
What CUDA version?

Yes, it’s Nvidia Quadro 5600 GPU, driver 169.04, CUDA 1.1.

Thanks.

What is the output of the deviceQuery example from the SDK?

Thanks for the tip on the deviceQuery example. I solved the problem by using that information.

Thanks!

I ran into the same problem.How did you solve the problem?

Could you explain it in detail?

Thank you!

Same here!!

cufftPlan1d runs fine up to NX=1024, but fails above this size, with:

cufft: ERROR: C:/cygwin/home/cuda0/cuda/sw/gpgpu_rel1.1/cufft/src/plan.cu, line

35

cufft: ERROR: CUFFT_INTERNAL_ERROR

This is with CUDA 1.1, VS2005, XP SP2, 8800 GTS 512.

All of the examples in CUDA-SDK build and run fine.

It’s “just” that I can’t get CUDAFFT to work!

Even this very basic program (InitCUDA source omitted):

int main (int argc, char **argv)

{

	cufftHandle  Plan;

	if(!InitCUDA()) 

	{

  return 0;

	}

	CUDA_SAFE_CALL (cufftPlan1d (&Plan, 4096, CUFFT_C2C, 1));

	return 0;

}

fails with the dreaded CUFFT_INTERNAL_ERROR!

It runs fine with 1024 instead of 4096.

My system has 4 GB of RAM.

Thanks for any help…

Fernando