[PATCH] D21562: [AMDGPU] Wave and register controls

Konstantin Zhuravlyov via llvm-commits llvm-commits at lists.llvm.org
Thu Jul 28 07:39:01 PDT 2016


kzhuravl added inline comments.

================
Comment at: lib/Target/AMDGPU/AMDGPUSubtarget.cpp:269-285
@@ +268,19 @@
+  if (Requested.first > Requested.second) {
+    Ctx.emitError("invalid amdgpu-flat-work-group-size: "
+                  "minimum is greater than maximum");
+    return Default;
+  }
+  // Make sure requested values do not violate subtarget's specifications.
+  if (Requested.first < getMinFlatWorkGroupSize()) {
+    Ctx.emitError("invalid amdgpu-flat-work-group-size: "
+                  "requested minimum flat work group size is not supported by "
+                  "the subtarget");
+    return Default;
+  }
+  if (Requested.second > getMaxFlatWorkGroupSize()) {
+    Ctx.emitError("invalid amdgpu-flat-work-group-size: "
+                  "requested maximum flat work group size is not supported by "
+                  "the subtarget");
+    return Default;
+  }
+
----------------
arsenm wrote:
> More DiagnosticInfos. These should also provide the function so clang will be able to provide a more useful debug location
Changed to clamp


https://reviews.llvm.org/D21562





More information about the llvm-commits mailing list