Will Saunders comments

Results 5 comments of


                                            Will Saunders

omp.library-only with Cray OpenMP

I agree that `omp_get_max_threads` should be callable outside a parallel region. I did some more digging with CCE - it strongly objects to OpenMP calls in additional threads: ``` #include...

omp.library-only with Cray OpenMP

@tomdeakin Don't suppose there was any news on this?

Atomic fetch_min and fetch_max with llvm cuda backend.

I had a look though acpp internals and realised my CAS loop alternative was wrong as I was missing the atomic load. It looks like acpp is using a CAS...

Atomic fetch_min and fetch_max with llvm cuda backend.

I'm probably missing something, this should work and avoid the atomic load? ``` sycl::atomic_ref element_atomic(d_y[0]); REAL expected = std::numeric_limits::max(); REAL desired; do { desired = sycl::min(value, expected); } while (...

Atomic fetch_min and fetch_max with llvm cuda backend.

Thanks for your input. I can patch the older backends on our end with a CAS loop like above. Introducing the AdaptiveCpp generic backend into the workflow is on my...