[llvm] be44177 - [X86][avx512fp16] Promote fp16 to fp32 for frem.

Fri Jan 28 19:41:48 PST 2022

Author: Luo, Yuanke
Date: 2022-01-29T11:41:27+08:00
New Revision: be44177edea535d93af147cb7f26fae55f8da8ba

URL: https://github.com/llvm/llvm-project/commit/be44177edea535d93af147cb7f26fae55f8da8ba
DIFF: https://github.com/llvm/llvm-project/commit/be44177edea535d93af147cb7f26fae55f8da8ba.diff

LOG: [X86][avx512fp16] Promote fp16 to fp32 for frem.

Promote fp16 to fp32 for frem.

Differential Revision: https://reviews.llvm.org/D118470

Added: 
    llvm/test/CodeGen/X86/avx512fp16-frem.ll

Modified: 
    llvm/lib/Target/X86/X86ISelLowering.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index e08035ee4c1eb..4fc79bc49f52e 100644

--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -805,8 +805,9 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
   setOperationAction(ISD::FMAXNUM, MVT::f80, Expand);
 
   // Some FP actions are always expanded for vector types.
-  for (auto VT : { MVT::v4f32, MVT::v8f32, MVT::v16f32,
-                   MVT::v2f64, MVT::v4f64, MVT::v8f64 }) {
+  for (auto VT : { MVT::v8f16, MVT::v16f16, MVT::v32f16,
+                   MVT::v4f32, MVT::v8f32,  MVT::v16f32,
+                   MVT::v2f64, MVT::v4f64,  MVT::v8f64 }) {
     setOperationAction(ISD::FSIN,      VT, Expand);
     setOperationAction(ISD::FSINCOS,   VT, Expand);
     setOperationAction(ISD::FCOS,      VT, Expand);
@@ -1958,6 +1959,8 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
     // AVX512_FP16 scalar operations
     setGroup(MVT::f16);
     addRegisterClass(MVT::f16,    &X86::FR16XRegClass);
+    setOperationAction(ISD::FREM,                 MVT::f16, Promote);
+    setOperationAction(ISD::STRICT_FREM,          MVT::f16, Promote);
     setOperationAction(ISD::SELECT_CC,            MVT::f16, Expand);
     setOperationAction(ISD::BR_CC,                MVT::f16, Expand);
     setOperationAction(ISD::SETCC,                MVT::f16, Custom);

diff  --git a/llvm/test/CodeGen/X86/avx512fp16-frem.ll b/llvm/test/CodeGen/X86/avx512fp16-frem.ll
new file mode 100644
index 0000000000000..887deee6d75da
--- /dev/null
+++ b/llvm/test/CodeGen/X86/avx512fp16-frem.ll
@@ -0,0 +1,1514 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16 | FileCheck %s
+
+define half @frem(half %x, half %y) nounwind {
+; CHECK-LABEL: frem:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    pushq %rax
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    popq %rax
+; CHECK-NEXT:    retq
+  %r = frem half %x, %y
+  ret half %r
+}
+
+define <2 x half> @frem_vec2(<2 x half> %x, <2 x half> %y) nounwind {
+; CHECK-LABEL: frem_vec2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $88, %rsp
+; CHECK-NEXT:    vmovapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm0
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    addq $88, %rsp
+; CHECK-NEXT:    retq
+  %r = frem <2 x half> %x, %y
+  ret <2 x half> %r
+}
+
+define <4 x half> @frem_vec4(<4 x half> %x, <4 x half> %y) nounwind {
+; CHECK-LABEL: frem_vec4:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $88, %rsp
+; CHECK-NEXT:    vmovapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm0
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    addq $88, %rsp
+; CHECK-NEXT:    retq
+  %r = frem <4 x half> %x, %y
+  ret <4 x half> %r
+}
+
+define <8 x half> @frem_vec8(<8 x half> %x, <8 x half> %y) nounwind {
+; CHECK-LABEL: frem_vec8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $88, %rsp
+; CHECK-NEXT:    vmovapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm0
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    addq $88, %rsp
+; CHECK-NEXT:    retq
+  %r = frem <8 x half> %x, %y
+  ret <8 x half> %r
+}
+
+define <16 x half> @frem_vec16(<16 x half> %x, <16 x half> %y) nounwind {
+; CHECK-LABEL: frem_vec16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $152, %rsp
+; CHECK-NEXT:    vmovupd %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; CHECK-NEXT:    vmovupd %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; CHECK-NEXT:    vmovapd %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    addq $152, %rsp
+; CHECK-NEXT:    retq
+  %r = frem <16 x half> %x, %y
+  ret <16 x half> %r
+}
+
+define <32 x half> @frem_vec32(<32 x half> %x, <32 x half> %y) nounwind {
+; CHECK-LABEL: frem_vec32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $248, %rsp
+; CHECK-NEXT:    vmovupd %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
+; CHECK-NEXT:    vmovupd %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
+; CHECK-NEXT:    vextractf32x4 $3, %zmm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vextractf32x4 $3, %zmm1, %xmm1
+; CHECK-NEXT:    vmovapd %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
+; CHECK-NEXT:    vextractf32x4 $2, %zmm1, %xmm1
+; CHECK-NEXT:    vmovapd %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
+; CHECK-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; CHECK-NEXT:    vmovapd %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm1 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vinserti64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
+; CHECK-NEXT:    addq $248, %rsp
+; CHECK-NEXT:    retq
+  %r = frem <32 x half> %x, %y
+  ret <32 x half> %r
+}
+
+define half @frem_strict(half %x, half %y) nounwind #0 {
+; CHECK-LABEL: frem_strict:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    pushq %rax
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    popq %rax
+; CHECK-NEXT:    retq
+  %result = call half @llvm.experimental.constrained.frem.f16(half %x, half %y, metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret half %result
+}
+
+define <2 x half> @frem_strict_vec2(<2 x half> %x, <2 x half> %y) nounwind #0 {
+; CHECK-LABEL: frem_strict_vec2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $56, %rsp
+; CHECK-NEXT:    vmovaps %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    addq $56, %rsp
+; CHECK-NEXT:    retq
+  %result = call <2 x half> @llvm.experimental.constrained.frem.v2f16(<2 x half> %x, <2 x half> %y, metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret <2 x half> %result
+}
+
+define <4 x half> @frem_strict_vec4(<4 x half> %x, <4 x half> %y) nounwind #0 {
+; CHECK-LABEL: frem_strict_vec4:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $72, %rsp
+; CHECK-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, %xmm1, %xmm2
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm1
+; CHECK-NEXT:    vpsrlq $48, %xmm0, %xmm2
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vinsertps $28, (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],zero,zero
+; CHECK-NEXT:    addq $72, %rsp
+; CHECK-NEXT:    retq
+  %result = call <4 x half> @llvm.experimental.constrained.frem.v4f16(<4 x half> %x, <4 x half> %y, metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret <4 x half> %result
+}
+
+define <8 x half> @frem_strict_vec8(<8 x half> %x, <8 x half> %y) nounwind #0 {
+; CHECK-LABEL: frem_strict_vec8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $88, %rsp
+; CHECK-NEXT:    vmovapd %xmm1, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm1
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm2, %xmm2, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    addq $88, %rsp
+; CHECK-NEXT:    retq
+  %result = call <8 x half> @llvm.experimental.constrained.frem.v8f16(<8 x half> %x, <8 x half> %y, metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret <8 x half> %result
+}
+
+define <16 x half> @frem_strict_vec16(<16 x half> %x, <16 x half> %y) nounwind #0 {
+; CHECK-LABEL: frem_strict_vec16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $152, %rsp
+; CHECK-NEXT:    vmovupd %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; CHECK-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; CHECK-NEXT:    vmovapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    addq $152, %rsp
+; CHECK-NEXT:    retq
+  %result = call <16 x half> @llvm.experimental.constrained.frem.v16f16(<16 x half> %x, <16 x half> %y, metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret <16 x half> %result
+}
+
+define <32 x half> @frem_strict_vec32(<32 x half> %x, <32 x half> %y) nounwind #0 {
+; CHECK-LABEL: frem_strict_vec32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    subq $248, %rsp
+; CHECK-NEXT:    vmovupd %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
+; CHECK-NEXT:    vmovups %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
+; CHECK-NEXT:    vextractf32x4 $3, %zmm1, %xmm1
+; CHECK-NEXT:    vmovapd %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm1, %xmm1, %xmm1
+; CHECK-NEXT:    vextractf32x4 $3, %zmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovupd {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; CHECK-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $14, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrldq $10, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,0]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
+; CHECK-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vzeroupper
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm1
+; CHECK-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vcvtsh2ss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    callq fmodf at PLT
+; CHECK-NEXT:    vcvtss2sh %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; CHECK-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
+; CHECK-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
+; CHECK-NEXT:    vpunpcklqdq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-NEXT:    # xmm0 = xmm0[0],mem[0]
+; CHECK-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
+; CHECK-NEXT:    vinserti64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
+; CHECK-NEXT:    addq $248, %rsp
+; CHECK-NEXT:    retq
+  %result = call <32 x half> @llvm.experimental.constrained.frem.v32f16(<32 x half> %x, <32 x half> %y, metadata !"round.dynamic", metadata !"fpexcept.strict") #0
+  ret <32 x half> %result
+}
+
+attributes #0 = { strictfp }
+declare half @llvm.experimental.constrained.frem.f16(half, half, metadata, metadata)
+declare <2 x half> @llvm.experimental.constrained.frem.v2f16(<2 x half>, <2 x half>, metadata, metadata)
+declare <4 x half> @llvm.experimental.constrained.frem.v4f16(<4 x half>, <4 x half>, metadata, metadata)
+declare <8 x half> @llvm.experimental.constrained.frem.v8f16(<8 x half>, <8 x half>, metadata, metadata)
+declare <16 x half> @llvm.experimental.constrained.frem.v16f16(<16 x half>, <16 x half>, metadata, metadata)
+declare <32 x half> @llvm.experimental.constrained.frem.v32f16(<32 x half>, <32 x half>, metadata, metadata)


        
