I’m using a Geforce 8800 GTX with CUDA 1.1, and I would like to understand a problem with the shared memory.
This GPU use 16 Ko of shared memory but when I want to create an array of unsigned char in the shared memory, a CUDA ERROR appears.
cudaError_enum at memory location 0x0012ef08.
I would like to have an array of 72416sizeof (unsigned char) bytes (11584 bytes) but it crashes.
However if I reduce the array size (<4000 bytes) it runs. :dry:
Please help :blink: