Eigenvalue Test Failed

The eigenvalue test that comes with the CUDA SDK reproducibly fails on my machine. It generates the following output:

Matrix size: 2048 x 2048
Precision: 0.000010
Iterations to be timed: 100
Result filename: ‘eigenvalues.dat’
Gerschgorin interval: -2.894310 / 2.923304
Average time step 1: 41.569971 ms
Average time step 2, one intervals: 6.247866 ms
Average time step 2, mult intervals: 0.016520 ms
Average time TOTAL: 47.887832 ms

I am using Red Hat Linux, with gcc4, on an Intel Core2 Quad processor and the NVIDIA Tesla C870. Does anyone have any ideas on why the test always fails?

Thank you.


Where did you obtain this version of Eigenvalues?

In the CUDA SDK itself, as I have mentioned earlier


I downloaded from NVIDIA site program to count the eigenvalues ​​of the tridiagonal matrix, I wanted to rewrite it so that it is running in double precision. To the dimension of the matrix 512x512 everything works smoothly, but with larger matrix I get an error bisect_large.cu (240): cutilCheckMsg cudaThreadSynchronize error: bisectKernelLarge_MultIntervals () FAILED. : Unknown error. Can anyone give me a hint what is the problem.