Remove deprecated SYCL usage for newer compilers #1772
Conversation
|
This is good to merge. Please review when you have some time. Thank you. |
include/RAJA/policy/sycl/policy.hpp
Outdated
| @@ -39,7 +39,7 @@ struct uint3 { | |||
| unsigned long x, y, z; | |||
| }; | |||
|
|
|||
| using sycl_dim_t = cl::sycl::range<1>; | |||
| using sycl_dim_t = sycl::range<1>; | |||
There was a problem hiding this comment.
Maybe keep the global scope ::sycl for consistency?
| using sycl_dim_t = sycl::range<1>; | |
| using sycl_dim_t = ::sycl::range<1>; |
There was a problem hiding this comment.
Good catch. Fixed.
Note that I didn't exhaustively search the code for these. I only changed the lines that generated compiler warnings.
| device = reinterpret_cast<T *>(::sycl::malloc_device(sycl::MaxNumTeams * sizeof(T), *(q))); | ||
| host = reinterpret_cast<T *>(::sycl::malloc_host(sycl::MaxNumTeams * sizeof(T), *(q))); |
There was a problem hiding this comment.
The references to MaxNumTeams look slightly confusing; should they be under the RAJA::sycl:: namespace?
| device = reinterpret_cast<T *>(::sycl::malloc_device(sycl::MaxNumTeams * sizeof(T), *(q))); | |
| host = reinterpret_cast<T *>(::sycl::malloc_host(sycl::MaxNumTeams * sizeof(T), *(q))); | |
| device = reinterpret_cast<T *>(::sycl::malloc_device(RAJA::sycl::MaxNumTeams * sizeof(T), *(q))); | |
| host = reinterpret_cast<T *>(::sycl::malloc_host(RAJA::sycl::MaxNumTeams * sizeof(T), *(q))); |
There was a problem hiding this comment.
Note that I did not write this code and was fixing compiler warnings. It looks like this file was started as a copy of the similarly named openmp target file.
This is probably a better was to define the variable: https://github.com/LLNL/RAJA/blob/develop/include/RAJA/policy/openmp_target/reduce.hpp#L70
However, that file also contains similar namespace scoping issues. For example, https://github.com/LLNL/RAJA/blob/develop/include/RAJA/policy/openmp_target/reduce.hpp#L107
I'll make an issue and we can fix those files and others for consistency as needed in a separate PR.
|
I'm having the same build failure (OMP Target) on a different RAJA PR.... |
Try restarting the failed job. One test fails intermittently. Not sure why. |
Summary