[SYCL] Emit an error on attempt to compile in less than C++17 mode #6678
Conversation
|
If we can do that, it would be really great! |
|
@aelovikov-intel @pvchupin Friendly ping. |
|
@romanovvlad, this change breaks post-commit on Windows. Please, take a look. |
@romanovvlad, ping on that |
| @@ -10,13 +9,15 @@ | |||
|
|
|||
| #include <sycl/sycl.hpp> | |||
|
|
|||
| // cxx14-warning@* {{DPCPP does not support C++ version earlier than C++17. Some features might not be available.}} | |||
| // clang-format off | |||
| // cxx14-error@* {{static assertion failed due to requirement '201402L >= 201703L'}} | |||
There was a problem hiding this comment.
I don't think it works reliably - any attempt to use C++17 feature without macro guard would result in an extra error reported and the test would fail.
There was a problem hiding this comment.
Hm, the static_assert should fire very early as it's in the most basic header. But to make the test more reliable, probably, we could add -Xclang -verify-ignore-unexpected=error,note and -Xclang -ferror-limit=<Big number> options.
…ce (#7077) CUDA backend implementation using the "unified" matrix extension interface. The same interface will be used for a future Intel backend implementation of the matrix extension. - New "unified" interface uses SYCL_EXT_ONEAPI_MATRIX_VERSION=4 - `joint_matrix_load`, `joint_matrix_store`, `joint_matrix_mad` and `joint_matrix` interfaces match the new spec from #6662 - Separated `joint_matrix_*` functions into new header matrix-unified.hpp: Intel backend implementations can be called from the same functions in the future. - C++17 everywhere in line with #6678 - Updated device code tests to use new interfaces - Completely removed uint16 implementations that are replaced by bfloat16 that is being moved out of the experimental namespace - Updated all CUDA runtime matrix tests here: intel/llvm-test-suite#1183 Signed-off-by: JackAKirk <[email protected]>
No description provided.