================
@@ -3574,6 +3574,46 @@ An error will be given if:
attributes;
- The AMDGPU target backend is unable to create machine code that can meet
the
request.
+}];
+}
+
+def AMDGPUKernargPreloadCountDocs : Documentation {
+ let Category = DocCatAMDGPUAttributes;
+ let Content = [{
+Clang supports the
+``__attribute__((amdgpu_kernarg_preload_count(<count>)))`` attribute on AMDGPU
+kernel functions.
+
+Kernarg preloading can reduce kernel launch latency. Without preloading, the
+kernel may need to load explicit kernel arguments from the kernarg segment
+memory. On targets that support kernarg preloading, the hardware can copy part
----------------
AaronBallman wrote:
```suggestion
memory. On targets that support kernel argument preloading, the hardware can
copy part
```
and same change elsewhere too? Or is that a term of art users will already be
familiar with?
https://github.com/llvm/llvm-project/pull/211059
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
