Anshuman Goswami

Results 6 comments of Anshuman Goswami

Can a pool memory resource be used to avoid cudaMalloc on each async call?

This is what I have now. This calls `cudaMalloc` on each invocation of `async::sort`. ``` #include #include #include #include #include #include #include #include #define CHECK_CUDA_ERROR(stmt) \ do { \ cudaError_t...

Is --expt-extended-lambda and --expt-relaxed-constexpr not supported?

Thanks for the detailed answer @maddyscientist That helps. For moving *a* kernel from offline to runtime compilation in a code base that has `--expt-relaxed-constexpr` and `--expt-extended-lambda` among the `nvcc` options,...

Is --expt-extended-lambda and --expt-relaxed-constexpr not supported?

Let me see if I can share a representative snippet. Yes, toolkit installation on the deployment host, is something I missed. That is possible. Good to know that otherwise, invoking...

Compilation error with histogram test program

No, it does not. @leofang

mallocMC managing two or more contexts

@slizzered Thanks for your reply ! I have a followup question. CE > But all of the state is included in one chunk of allocated memory (the memPool that contains...

mallocMC managing two or more contexts

@sudkannan Adding Sudarsun (interested in checkpoint/restarts) in the thread...