[PATCH] D113107: Support of expression granularity for _Float16.

[Phoebe Wang via Phabricator via cfe-commits]

pengfei added a comment.

> Then we need to add the option -fexcess-precision. I am not sure for now where and what values the _FLT_EVAL_METHOD should have when excess precision is enabled/disabled.

I'm fine with a follow up patch to enable this option. Please notice LLVM15 will branch on July 26. I wish we can enable the option before that and make it usable in LLVM15.



================
Comment at: clang/docs/ReleaseNotes.rst:537
 
+- Support for ``_Float16`` type has been added.
+
----------------
We don't need it anymore. It's in line 523 now.


================
Comment at: clang/lib/Basic/Targets/X86.cpp:374-375
   }
+  // Turn on _float16 for x86 (feature sse2)
+  HasFloat16 = SSELevel >= SSE2;
 
----------------
It's in line 358.


================
Comment at: clang/test/CodeGen/X86/Float16-arithmetic.c:5-14
+  // CHECK-LABEL: @add1
+  // CHECK: [[A:%.*]] = alloca half
+  // CHECK-NEXT: [[B:%.*]] = alloca half
+  // CHECK: [[A_LOAD:%.*]] = load half, ptr [[A]]
+  // CHECK-NEXT: [[A_EXT:%.*]] = fpext half [[A_LOAD]] to float
+  // CHECK-NEXT: [[B_LOAD:%.*]] = load half, ptr [[B]]
+  // CHECK-NEXT: [[B_EXT:%.*]] = fpext half [[B_LOAD]] to float
----------------
Are these code generated manually? It can be updated by command `llvm/utils/update_cc_test_checks.py clang/test/CodeGen/X86/Float16-arithmetic.c`.


================
Comment at: clang/test/CodeGen/X86/Float16-complex.c:5-19
+  // CHECK-LABEL: @add_half_rr(
+  // CHECK: [[A:%.*]] = alloca half
+  // CHECK-NEXT: [[B:%.*]] = alloca half
+  // CHECK: [[A_LOAD:%.*]] = load half, ptr [[A]]
+
+  // AVX-NEXT: [[B_LOAD:%.*]] = load half, ptr [[B]]
+  // AVX-NEXT: [[AB_ADD:%.*]] = fadd half [[A_LOAD]], [[B_LOAD]]
----------------
It can be generated by script too.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D113107/new/

https://reviews.llvm.org/D113107