Closed rhornung67 closed 3 months ago
The way memory is managed in the FIR (Apps) kernel for SYCL is different than for CUDA and HIP. Make SYCL implementation consistent with other GPU variants.
On second look, all GPU variants are doing essentially the same thing.
The way memory is managed in the FIR (Apps) kernel for SYCL is different than for CUDA and HIP. Make SYCL implementation consistent with other GPU variants.