awarzynski added inline comments.
Comment at: flang/lib/Frontend/FrontendActions.cpp:139-142
+ // Clang does not append all target features to the clang -cc1 invocation.
+ // Some AMDGPU features are passed implicitly by the Clang frontend.
+ // That's why we need to extract i
domada marked an inline comment as done.
domada added inline comments.
Comment at: flang/lib/Frontend/FrontendActions.cpp:139-142
+ // Clang does not append all target features to the clang -cc1 invocation.
+ // Some AMDGPU features are passed implicitly by the Clang frontend.
awarzynski added a comment.
A few more comments, but mostly nits. Btw, is this patch sufficient to generate
code for AMDGPU? Or, put differently, what's the level of support atm?
Comment at: clang/lib/Driver/ToolChains/Flang.cpp:107
break;
+ case llvm::Triple::r600:
+ c
domada marked an inline comment as done.
domada added inline comments.
Comment at: flang/lib/Frontend/FrontendActions.cpp:93
+std::string CodeGenAction::getAllTargetFeatures() {
+ std::string allFeaturesStr;
awarzynski wrote:
> This method could be simplified
domada updated this revision to Diff 507333.
domada retitled this revision from "[Flang][AMDGPU][OpenMP] Save target
features in OpenMP MLIR dialect" to "[Flang][AMDGPU] Add support for AMDGPU to
Flang driver".
domada edited the summary of this revision.
domada added a comment.
Applied remarks.