ptxas fatal : Memory allocation failure compilation error

Accelerated Computing CUDA CUDA Programming and Performance
1

I have a working program. But if I add

#include <curand_kernel.h>

The compiler reports an error:

1>ptxas fatal : Memory allocation failure

I did no other changes in the code. Why is ptxas out of memory? What does it exactly mean? What should I do?

I am using Windows XP 32bit, CUDA SDK v3.2RC. I am trying to use the CURAND from the v3.2RC. The orginial program EstimatorPiInlineP compiles and works without a problem.

Thank you,

2

PTXAS is a component of the CUDA compiler. It would be helpful if you could file a bug, attaching a self-contained repro case, so the toolchain team can have a look. Please also report details of your platform, such as the amount of system memory. Thanks!

3

PTXAS is a component of the CUDA compiler. It would be helpful if you could file a bug, attaching a self-contained repro case, so the toolchain team can have a look. Please also report details of your platform, such as the amount of system memory. Thanks!

4

I’m having the same problem. The kernel I would like to build is fairly large and includes several files. It builds with sm_13 but not sm_20, failing with: ptxas fatal : Memory allocation failure.

5

I’m having the same problem. The kernel I would like to build is fairly large and includes several files. It builds with sm_13 but not sm_20, failing with: ptxas fatal : Memory allocation failure.