problem with cudaMallocHost && concurrent Kernels

hi all,

I’m trying to run 2 kernels (the same) at the same time on GTX 460 (compute cap. 2.1) and CUDA 3.2. But for cudaMallocHost I’m getting the invalid argument error. I know that for Async kernel launch I need cudaMallocHost but still I get the error.

Thanks!! (3.23 KB)

you have to allocate the host memory with cudaMallocHost instead of malloc, not replace the cudaMalloc call with cudaMallocHost

Unfortunately it did not help (but you were right). I altered the code a little bit. Still I get the error, plus these kernels are not working concurrently. (3.87 KB)