[PATCH] D85385: [X86][FPEnv] Teach X86 mask compare intrinsics to respect strict FP semantics.

Mon Aug 10 17:43:47 PDT 2020

craig.topper added inline comments.

================
Comment at: clang/test/CodeGen/avx512f-builtins-constrained-cmp.c:793
   // CHECK-LABEL: test_mm512_mask_cmp_pd_mask_true_us
-  // CHECK: [[CMP:%.*]] = call <8 x i1> @llvm.x86.avx512.cmp.pd.512(<8 x double> %{{.*}}, <8 x double> %{{.*}}, i32 31, i32 4)
-  // CHECK: and <8 x i1> [[CMP]], {{.*}}
+  // CHECK: call <8 x i1> @llvm.x86.avx512.mask.cmp.pd.512(<8 x double> %{{.*}}, <8 x double> %{{.*}}, i32 31, <8 x i1> {{.*}}, i32 4) #2
   return _mm512_mask_cmp_pd_mask(m, a, b, _CMP_TRUE_US);
----------------
I missed this in our internal review. All of these check lines check "#2" at the end, do we need that? 

================
Comment at: llvm/lib/IR/AutoUpgrade.cpp:3764
+
+    if (NumElts < 8) {
+      int Indices[8];
----------------
Can we use getX86MaskVec here?

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D85385/new/

https://reviews.llvm.org/D85385