Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_lane.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_lane.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_lane.ll
@@ -0,0 +1,301 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half>, <4 x half>)
+declare <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half>, <8 x half>)
+declare <4 x half> @llvm.fma.v4f16(<4 x half>, <4 x half>, <4 x half>)
+declare <8 x half> @llvm.fma.v8f16(<8 x half>, <8 x half>, <8 x half>)
+declare half @llvm.fma.f16(half, half, half) #1
+
+define dso_local <4 x half> @t_vfma_lane_f16(<4 x half> %a, <4 x half> %b, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfma_lane_f16:
+; CHECK:         dup v2.4h, v2.h[0]
+; CHECK-NEXT:    fmla v0.4h, v2.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %lane1 = shufflevector <4 x half> %c, <4 x half> undef, <4 x i32> zeroinitializer
+  %fmla3 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %b, <4 x half> %lane1, <4 x half> %a)
+  ret <4 x half> %fmla3
+}
+
+define dso_local <8 x half> @t_vfmaq_lane_f16(<8 x half> %a, <8 x half> %b, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmaq_lane_f16:
+; CHECK:         dup v2.8h, v2.h[0]
+; CHECK-NEXT:    fmla v0.8h, v2.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %lane1 = shufflevector <4 x half> %c, <4 x half> undef, <8 x i32> zeroinitializer
+  %fmla3 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %lane1, <8 x half> %a)
+  ret <8 x half> %fmla3
+}
+
+define dso_local <4 x half> @t_vfma_laneq_f16(<4 x half> %a, <4 x half> %b, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfma_laneq_f16:
+; CHECK:         dup v2.4h, v2.h[0]
+; CHECK-NEXT:    fmla v0.4h, v1.4h, v2.4h
+; CHECK-NEXT:    ret
+entry:
+  %lane1 = shufflevector <8 x half> %c, <8 x half> undef, <4 x i32> zeroinitializer
+  %0 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %lane1, <4 x half> %b, <4 x half> %a)
+  ret <4 x half> %0
+}
+
+define dso_local <8 x half> @t_vfmaq_laneq_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmaq_laneq_f16:
+; CHECK:         dup v2.8h, v2.h[0]
+; CHECK-NEXT:    fmla v0.8h, v1.8h, v2.8h
+; CHECK-NEXT:    ret
+entry:
+  %lane1 = shufflevector <8 x half> %c, <8 x half> undef, <8 x i32> zeroinitializer
+  %0 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %lane1, <8 x half> %b, <8 x half> %a)
+  ret <8 x half> %0
+}
+
+define dso_local <4 x half> @t_vfma_n_f16(<4 x half> %a, <4 x half> %b, half %c) {
+; CHECK-LABEL: t_vfma_n_f16:
+; CHECK:         dup v2.4h, v2.h[0]
+; CHECK-NEXT:    fmla v0.4h, v2.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %vecinit = insertelement <4 x half> undef, half %c, i32 0
+  %vecinit3 = shufflevector <4 x half> %vecinit, <4 x half> undef, <4 x i32> zeroinitializer
+  %0 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %b, <4 x half> %vecinit3, <4 x half> %a) #4
+  ret <4 x half> %0
+}
+
+define dso_local <8 x half> @t_vfmaq_n_f16(<8 x half> %a, <8 x half> %b, half %c) {
+; CHECK-LABEL: t_vfmaq_n_f16:
+; CHECK:         dup v2.8h, v2.h[0]
+; CHECK-NEXT:    fmla v0.8h, v2.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %vecinit = insertelement <8 x half> undef, half %c, i32 0
+  %vecinit7 = shufflevector <8 x half> %vecinit, <8 x half> undef, <8 x i32> zeroinitializer
+  %0 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %vecinit7, <8 x half> %a) #4
+  ret <8 x half> %0
+}
+
+define dso_local half @t_vfmah_lane_f16(half %a, half %b, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmah_lane_f16:
+; CHECK:         fmadd h0, h1, h2, h0
+; CHECK-NEXT:    ret
+entry:
+  %extract = extractelement <4 x half> %c, i32 0
+  %0 = tail call half @llvm.fma.f16(half %b, half %extract, half %a)
+  ret half %0
+}
+
+define dso_local half @t_vfmah_laneq_f16(half %a, half %b, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmah_laneq_f16:
+; CHECK:         fmadd h0, h1, h2, h0
+; CHECK-NEXT:    ret
+entry:
+  %extract = extractelement <8 x half> %c, i32 0
+  %0 = tail call half @llvm.fma.f16(half %b, half %extract, half %a)
+  ret half %0
+}
+
+define dso_local <4 x half> @t_vfms_lane_f16(<4 x half> %a, <4 x half> %b, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfms_lane_f16:
+; CHECK:         fneg v1.4h, v1.4h
+; CHECK-NEXT:    dup v2.4h, v2.h[0]
+; CHECK-NEXT:    fmla v0.4h, v2.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %sub = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+  %lane1 = shufflevector <4 x half> %c, <4 x half> undef, <4 x i32> zeroinitializer
+  %fmla3 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %sub, <4 x half> %lane1, <4 x half> %a)
+  ret <4 x half> %fmla3
+}
+
+define dso_local <8 x half> @t_vfmsq_lane_f16(<8 x half> %a, <8 x half> %b, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmsq_lane_f16:
+; CHECK:         fneg v1.8h, v1.8h
+; CHECK-NEXT:    dup v2.8h, v2.h[0]
+; CHECK-NEXT:    fmla v0.8h, v2.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %sub = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+  %lane1 = shufflevector <4 x half> %c, <4 x half> undef, <8 x i32> zeroinitializer
+  %fmla3 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %sub, <8 x half> %lane1, <8 x half> %a)
+  ret <8 x half> %fmla3
+}
+
+define dso_local <4 x half> @t_vfms_laneq_f16(<4 x half> %a, <4 x half> %b, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfms_laneq_f16:
+; CHECK:         dup v2.4h, v2.h[0]
+; CHECK-NEXT:    fmls v0.4h, v2.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %sub = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+  %lane1 = shufflevector <8 x half> %c, <8 x half> undef, <4 x i32> zeroinitializer
+  %0 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %lane1, <4 x half> %sub, <4 x half> %a)
+  ret <4 x half> %0
+}
+
+define dso_local <8 x half> @t_vfmsq_laneq_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmsq_laneq_f16:
+; CHECK:         dup v2.8h, v2.h[0]
+; CHECK-NEXT:    fmls v0.8h, v2.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %sub = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+  %lane1 = shufflevector <8 x half> %c, <8 x half> undef, <8 x i32> zeroinitializer
+  %0 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %lane1, <8 x half> %sub, <8 x half> %a)
+  ret <8 x half> %0
+}
+
+define dso_local <4 x half> @t_vfms_n_f16(<4 x half> %a, <4 x half> %b, half %c) {
+; CHECK-LABEL: t_vfms_n_f16:
+; CHECK:         fneg v1.4h, v1.4h
+; CHECK-NEXT:    dup v2.4h, v2.h[0]
+; CHECK-NEXT:    fmla v0.4h, v2.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %sub = fsub <4 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+  %vecinit = insertelement <4 x half> undef, half %c, i32 0
+  %vecinit3 = shufflevector <4 x half> %vecinit, <4 x half> undef, <4 x i32> zeroinitializer
+  %0 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %sub, <4 x half> %vecinit3, <4 x half> %a) #4
+  ret <4 x half> %0
+}
+
+define dso_local <8 x half> @t_vfmsq_n_f16(<8 x half> %a, <8 x half> %b, half %c) {
+; CHECK-LABEL: t_vfmsq_n_f16:
+; CHECK:         fneg v1.8h, v1.8h
+; CHECK-NEXT:    dup v2.8h, v2.h[0]
+; CHECK-NEXT:    fmla v0.8h, v2.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %sub = fsub <8 x half> <half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000, half 0xH8000>, %b
+  %vecinit = insertelement <8 x half> undef, half %c, i32 0
+  %vecinit7 = shufflevector <8 x half> %vecinit, <8 x half> undef, <8 x i32> zeroinitializer
+  %0 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %sub, <8 x half> %vecinit7, <8 x half> %a) #4
+  ret <8 x half> %0
+}
+
+define dso_local half @t_vfmsh_lane_f16(half %a, half %b, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmsh_lane_f16:
+; CHECK:         fneg h1, h1
+; CHECK:         fmadd h0, h1, h2, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = fsub half 0xH8000, %b
+  %extract = extractelement <4 x half> %c, i32 0
+  %1 = tail call half @llvm.fma.f16(half %0, half %extract, half %a)
+  ret half %1
+}
+
+define dso_local half @t_vfmsh_laneq_f16(half %a, half %b, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vfmsh_laneq_f16:
+; CHECK:         fneg h1, h1
+; CHECK-NEXT:    fmadd h0, h1, h2, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = fsub half 0xH8000, %b
+  %extract = extractelement <8 x half> %c, i32 0
+  %1 = tail call half @llvm.fma.f16(half %0, half %extract, half %a)
+  ret half %1
+}
+
+define dso_local <4 x half> @t_vmul_laneq_f16(<4 x half> %a, <8 x half> %b, i32 %lane) {
+; CHECK-LABEL: t_vmul_laneq_f16:
+; CHECK:         fmul v0.4h, v0.4h, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %shuffle = shufflevector <8 x half> %b, <8 x half> undef, <4 x i32> zeroinitializer
+  %mul = fmul <4 x half> %shuffle, %a
+  ret <4 x half> %mul
+}
+
+define dso_local <8 x half> @t_vmulq_laneq_f16(<8 x half> %a, <8 x half> %b, i32 %lane) {
+; CHECK-LABEL: t_vmulq_laneq_f16:
+; CHECK:         fmul v0.8h, v0.8h, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %shuffle = shufflevector <8 x half> %b, <8 x half> undef, <8 x i32> zeroinitializer
+  %mul = fmul <8 x half> %shuffle, %a
+  ret <8 x half> %mul
+}
+
+define dso_local half @t_vmulh_lane_f16(half %a, <4 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vmulh_lane_f16:
+; CHECK:         fmul h0, h0, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %0 = extractelement <4 x half> %c, i32 0
+  %1 = fmul half %0, %a
+  ret half %1
+}
+
+define dso_local half @t_vmulh_laneq_f16(half %a, <8 x half> %c, i32 %lane) {
+; CHECK-LABEL: t_vmulh_laneq_f16:
+; CHECK:         fmul h0, h0, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %0 = extractelement <8 x half> %c, i32 0
+  %1 = fmul half %0, %a
+  ret half %1
+}
+
+define dso_local <4 x half> @t_vmulx_lane_f16(<4 x half> %a, <4 x half> %b, i32 %lane) {
+; CHECK-LABEL: t_vmulx_lane_f16:
+; CHECK:         fmulx v0.4h, v0.4h, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %shuffle = shufflevector <4 x half> %b, <4 x half> undef, <4 x i32> zeroinitializer
+  %vmulx2.i = tail call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> %shuffle) #4
+  ret <4 x half> %vmulx2.i
+}
+
+define dso_local <8 x half> @t_vmulxq_lane_f16(<8 x half> %a, <4 x half> %b, i32 %lane) {
+; CHECK-LABEL: t_vmulxq_lane_f16:
+; CHECK:         fmulx v0.8h, v0.8h, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %shuffle = shufflevector <4 x half> %b, <4 x half> undef, <8 x i32> zeroinitializer
+  %vmulx2.i = tail call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> %shuffle) #4
+  ret <8 x half> %vmulx2.i
+}
+
+define dso_local <4 x half> @t_vmulx_laneq_f16(<4 x half> %a, <8 x half> %b, i32 %lane) {
+; CHECK-LABEL: t_vmulx_laneq_f16:
+; CHECK:         fmulx v0.4h, v0.4h, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %shuffle = shufflevector <8 x half> %b, <8 x half> undef, <4 x i32> zeroinitializer
+  %vmulx2.i = tail call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> %shuffle) #4
+  ret <4 x half> %vmulx2.i
+}
+
+define dso_local <8 x half> @t_vmulxq_laneq_f16(<8 x half> %a, <8 x half> %b, i32 %lane) {
+; CHECK-LABEL: t_vmulxq_laneq_f16:
+; CHECK:         fmulx v0.8h, v0.8h, v1.h[0]
+; CHECK-NEXT:    ret
+entry:
+  %shuffle = shufflevector <8 x half> %b, <8 x half> undef, <8 x i32> zeroinitializer
+  %vmulx2.i = tail call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> %shuffle) #4
+  ret <8 x half> %vmulx2.i
+}
+
+define dso_local <4 x half> @t_vmulx_n_f16(<4 x half> %a, half %c) {
+; CHECK-LABEL: t_vmulx_n_f16:
+; CHECK:         dup v1.4h, v1.h[0]
+; CHECK-NEXT:    fmulx v0.4h, v0.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %vecinit = insertelement <4 x half> undef, half %c, i32 0
+  %vecinit3 = shufflevector <4 x half> %vecinit, <4 x half> undef, <4 x i32> zeroinitializer
+  %vmulx2.i = tail call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> %vecinit3) #4
+  ret <4 x half> %vmulx2.i
+}
+
+define dso_local <8 x half> @t_vmulxq_n_f16(<8 x half> %a, half %c) {
+; CHECK-LABEL: t_vmulxq_n_f16:
+; CHECK:         dup v1.8h, v1.h[0]
+; CHECK-NEXT:    fmulx v0.8h, v0.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %vecinit = insertelement <8 x half> undef, half %c, i32 0
+  %vecinit7 = shufflevector <8 x half> %vecinit, <8 x half> undef, <8 x i32> zeroinitializer
+  %vmulx2.i = tail call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> %vecinit7) #4
+  ret <8 x half> %vmulx2.i
+}
Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_1op.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_1op.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_1op.ll
@@ -0,0 +1,318 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare i64 @llvm.aarch64.neon.fcvtpu.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtpu.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtps.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtps.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtnu.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtnu.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtns.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtns.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtmu.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtmu.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtms.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtms.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtau.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtau.i32.f16(half)
+declare i64 @llvm.aarch64.neon.fcvtas.i64.f16(half)
+declare i32 @llvm.aarch64.neon.fcvtas.i32.f16(half)
+declare half @llvm.aarch64.neon.frsqrte.f16(half)
+declare half @llvm.aarch64.neon.frecpx.f16(half)
+declare half @llvm.aarch64.neon.frecpe.f16(half)
+
+define dso_local i16 @t2(half %a) {
+; CHECK-LABEL: t2:
+; CHECK:         fcmp h0, #0.0
+; CHECK-NEXT:    csetm w0, eq
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp oeq half %a, 0xH0000
+  %vceqz = sext i1 %0 to i16
+  ret i16 %vceqz
+}
+
+define dso_local i16 @t3(half %a) {
+; CHECK-LABEL: t3:
+; CHECK:         fcmp h0, #0.0
+; CHECK-NEXT:    csetm w0, ge
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp oge half %a, 0xH0000
+  %vcgez = sext i1 %0 to i16
+  ret i16 %vcgez
+}
+
+define dso_local i16 @t4(half %a) {
+; CHECK-LABEL: t4:
+; CHECK:         fcmp h0, #0.0
+; CHECK-NEXT:    csetm w0, gt
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp ogt half %a, 0xH0000
+  %vcgtz = sext i1 %0 to i16
+  ret i16 %vcgtz
+}
+
+define dso_local i16 @t5(half %a) {
+; CHECK-LABEL: t5:
+; CHECK:         fcmp h0, #0.0
+; CHECK-NEXT:    csetm w0, ls
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp ole half %a, 0xH0000
+  %vclez = sext i1 %0 to i16
+  ret i16 %vclez
+}
+
+define dso_local i16 @t6(half %a) {
+; CHECK-LABEL: t6:
+; CHECK:         fcmp h0, #0.0
+; CHECK-NEXT:    csetm w0, mi
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp olt half %a, 0xH0000
+  %vcltz = sext i1 %0 to i16
+  ret i16 %vcltz
+}
+
+define dso_local half @t8(i32 %a) {
+; CHECK-LABEL: t8:
+; CHECK:         scvtf h0, w0
+; CHECK-NEXT:    ret
+entry:
+  %0 = sitofp i32 %a to half
+  ret half %0
+}
+
+define dso_local half @t9(i64 %a) {
+; CHECK-LABEL: t9:
+; CHECK:         scvtf h0, x0
+; CHECK-NEXT:    ret
+entry:
+  %0 = sitofp i64 %a to half
+  ret half %0
+}
+
+define dso_local half @t12(i64 %a) {
+; CHECK-LABEL: t12:
+; CHECK:         ucvtf h0, x0
+; CHECK-NEXT:    ret
+entry:
+  %0 = uitofp i64 %a to half
+  ret half %0
+}
+
+define dso_local i16 @t13(half %a) {
+; CHECK-LABEL: t13:
+; CHECK:         fcvtzs w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = fptosi half %a to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t15(half %a) {
+; CHECK-LABEL: t15:
+; CHECK:         fcvtzs x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = fptosi half %a to i64
+  ret i64 %0
+}
+
+define dso_local i16 @t16(half %a) {
+; CHECK-LABEL: t16:
+; CHECK:         fcvtzs w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = fptoui half %a to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t18(half %a) {
+; CHECK-LABEL: t18:
+; CHECK:         fcvtzu x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = fptoui half %a to i64
+  ret i64 %0
+}
+
+define dso_local i16 @t19(half %a) {
+; CHECK-LABEL: t19:
+; CHECK:         fcvtas w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtas.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t21(half %a) {
+; CHECK-LABEL: t21:
+; CHECK:         fcvtas x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtah_s64_f16 = tail call i64 @llvm.aarch64.neon.fcvtas.i64.f16(half %a)
+  ret i64 %vcvtah_s64_f16
+}
+
+define dso_local i16 @t22(half %a) {
+; CHECK-LABEL: t22:
+; CHECK:         fcvtau w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtau.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t24(half %a) {
+; CHECK-LABEL: t24:
+; CHECK:         fcvtau x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtah_u64_f16 = tail call i64 @llvm.aarch64.neon.fcvtau.i64.f16(half %a)
+  ret i64 %vcvtah_u64_f16
+}
+
+define dso_local i16 @t25(half %a) {
+; CHECK-LABEL: t25:
+; CHECK:         fcvtms w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtms.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t27(half %a) {
+; CHECK-LABEL: t27:
+; CHECK:         fcvtms x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtmh_s64_f16 = tail call i64 @llvm.aarch64.neon.fcvtms.i64.f16(half %a)
+  ret i64 %vcvtmh_s64_f16
+}
+
+define dso_local i16 @t28(half %a) {
+; CHECK-LABEL: t28:
+; CHECK:         fcvtmu w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtmu.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t30(half %a) {
+; CHECK-LABEL: t30:
+; CHECK:         fcvtmu x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtmh_u64_f16 = tail call i64 @llvm.aarch64.neon.fcvtmu.i64.f16(half %a)
+  ret i64 %vcvtmh_u64_f16
+}
+
+define dso_local i16 @t31(half %a) {
+; CHECK-LABEL: t31:
+; CHECK:         fcvtns w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtns.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t33(half %a) {
+; CHECK-LABEL: t33:
+; CHECK:         fcvtns x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtnh_s64_f16 = tail call i64 @llvm.aarch64.neon.fcvtns.i64.f16(half %a)
+  ret i64 %vcvtnh_s64_f16
+}
+
+define dso_local i16 @t34(half %a) {
+; CHECK-LABEL: t34:
+; CHECK:         fcvtnu w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtnu.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t36(half %a) {
+; CHECK-LABEL: t36:
+; CHECK:         fcvtnu x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtnh_u64_f16 = tail call i64 @llvm.aarch64.neon.fcvtnu.i64.f16(half %a)
+  ret i64 %vcvtnh_u64_f16
+}
+
+define dso_local i16 @t37(half %a) {
+; CHECK-LABEL: t37:
+; CHECK:         fcvtps w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtps.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t39(half %a) {
+; CHECK-LABEL: t39:
+; CHECK:         fcvtps x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtph_s64_f16 = tail call i64 @llvm.aarch64.neon.fcvtps.i64.f16(half %a)
+  ret i64 %vcvtph_s64_f16
+}
+
+define dso_local i16 @t40(half %a) {
+; CHECK-LABEL: t40:
+; CHECK:         fcvtpu w0, h0
+; CHECK-NEXT:    ret
+entry:
+  %fcvt = tail call i32 @llvm.aarch64.neon.fcvtpu.i32.f16(half %a)
+  %0 = trunc i32 %fcvt to i16
+  ret i16 %0
+}
+
+define dso_local i64 @t42(half %a) {
+; CHECK-LABEL: t42:
+; CHECK:         fcvtpu x0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vcvtph_u64_f16 = tail call i64 @llvm.aarch64.neon.fcvtpu.i64.f16(half %a)
+  ret i64 %vcvtph_u64_f16
+}
+
+define dso_local half @t44(half %a) {
+; CHECK-LABEL: t44:
+; CHECK:         frecpe h0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vrecpeh_f16 = tail call half @llvm.aarch64.neon.frecpe.f16(half %a)
+  ret half %vrecpeh_f16
+}
+
+define dso_local half @t45(half %a) {
+; CHECK-LABEL: t45:
+; CHECK:         frecpx h0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vrecpxh_f16 = tail call half @llvm.aarch64.neon.frecpx.f16(half %a)
+  ret half %vrecpxh_f16
+}
+
+define dso_local half @t53(half %a) {
+; CHECK-LABEL: t53:
+; CHECK:         frsqrte h0, h0
+; CHECK-NEXT:    ret
+entry:
+  %vrsqrteh_f16 = tail call half @llvm.aarch64.neon.frsqrte.f16(half %a)
+  ret half %vrsqrteh_f16
+}
Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_2op.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_2op.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_2op.ll
@@ -0,0 +1,117 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare half @llvm.aarch64.sisd.fabd.f16(half, half)
+declare half @llvm.aarch64.neon.fmax.f16(half, half)
+declare half @llvm.aarch64.neon.fmin.f16(half, half)
+declare half @llvm.aarch64.neon.frsqrts.f16(half, half)
+declare half @llvm.aarch64.neon.frecps.f16(half, half)
+declare half @llvm.aarch64.neon.fmulx.f16(half, half)
+
+define dso_local half @t_vabdh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vabdh_f16:
+; CHECK:         fabd h0, h0, h1
+; CHECK-NEXT:    ret
+entry:
+  %vabdh_f16 = tail call half @llvm.aarch64.sisd.fabd.f16(half %a, half %b)
+  ret half %vabdh_f16
+}
+
+define dso_local i16 @t_vceqh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vceqh_f16:
+; CHECK:         fcmp h0, h1
+; CHECK-NEXT:    csetm w0, eq
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp oeq half %a, %b
+  %vcmpd = sext i1 %0 to i16
+  ret i16 %vcmpd
+}
+
+define dso_local i16 @t_vcgeh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vcgeh_f16:
+; CHECK:         fcmp h0, h1
+; CHECK-NEXT:    csetm w0, ge
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp oge half %a, %b
+  %vcmpd = sext i1 %0 to i16
+  ret i16 %vcmpd
+}
+
+define dso_local i16 @t_vcgth_f16(half %a, half %b) {
+; CHECK-LABEL: t_vcgth_f16:
+; CHECK:         fcmp h0, h1
+; CHECK-NEXT:    csetm w0, gt
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp ogt half %a, %b
+  %vcmpd = sext i1 %0 to i16
+  ret i16 %vcmpd
+}
+
+define dso_local i16 @t_vcleh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vcleh_f16:
+; CHECK:         fcmp h0, h1
+; CHECK-NEXT:    csetm w0, ls
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp ole half %a, %b
+  %vcmpd = sext i1 %0 to i16
+  ret i16 %vcmpd
+}
+
+define dso_local i16 @t_vclth_f16(half %a, half %b) {
+; CHECK-LABEL: t_vclth_f16:
+; CHECK:         fcmp h0, h1
+; CHECK-NEXT:    csetm w0, mi
+; CHECK-NEXT:    ret
+entry:
+  %0 = fcmp olt half %a, %b
+  %vcmpd = sext i1 %0 to i16
+  ret i16 %vcmpd
+}
+
+define dso_local half @t_vmaxh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vmaxh_f16:
+; CHECK:         fmax h0, h0, h1
+; CHECK-NEXT:    ret
+entry:
+  %vmax = tail call half @llvm.aarch64.neon.fmax.f16(half %a, half %b)
+  ret half %vmax
+}
+
+define dso_local half @t_vminh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vminh_f16:
+; CHECK:         fmin h0, h0, h1
+; CHECK-NEXT:    ret
+entry:
+  %vmin = tail call half @llvm.aarch64.neon.fmin.f16(half %a, half %b)
+  ret half %vmin
+}
+
+define dso_local half @t_vmulxh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vmulxh_f16:
+; CHECK:         fmulx h0, h0, h1
+; CHECK-NEXT:    ret
+entry:
+  %vmulxh_f16 = tail call half @llvm.aarch64.neon.fmulx.f16(half %a, half %b)
+  ret half %vmulxh_f16
+}
+
+define dso_local half @t_vrecpsh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vrecpsh_f16:
+; CHECK:         frecps h0, h0, h1
+; CHECK-NEXT:    ret
+entry:
+  %vrecps = tail call half @llvm.aarch64.neon.frecps.f16(half %a, half %b)
+  ret half %vrecps
+}
+
+define dso_local half @t_vrsqrtsh_f16(half %a, half %b) {
+; CHECK-LABEL: t_vrsqrtsh_f16:
+; CHECK:         frsqrts h0, h0, h1
+; CHECK-NEXT:    ret
+entry:
+  %vrsqrtsh_f16 = tail call half @llvm.aarch64.neon.frsqrts.f16(half %a, half %b)
+  ret half %vrsqrtsh_f16
+}
Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_3op.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_3op.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_scalar_3op.ll
@@ -0,0 +1,13 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare half @llvm.fma.f16(half, half, half)
+
+define dso_local half @t_vfmah_f16(half %a, half %b, half %c) {
+; CHECK-LABEL: t_vfmah_f16:
+; CHECK:         fmadd h0, h1, h2, h0
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call half @llvm.fma.f16(half %b, half %c, half %a)
+  ret half %0
+}
+
Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_1op.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_1op.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_1op.ll
@@ -0,0 +1,42 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare <4 x half> @llvm.nearbyint.v4f16(<4 x half>)
+declare <8 x half> @llvm.nearbyint.v8f16(<8 x half>)
+declare <4 x half> @llvm.sqrt.v4f16(<4 x half>)
+declare <8 x half> @llvm.sqrt.v8f16(<8 x half>)
+
+define dso_local <4 x half> @t_vrndi_f16(<4 x half> %a) {
+; CHECK-LABEL: t_vrndi_f16:
+; CHECK:         frinti v0.4h, v0.4h
+; CHECK-NEXT:    ret
+entry:
+  %vrndi1.i = tail call <4 x half> @llvm.nearbyint.v4f16(<4 x half> %a)
+  ret <4 x half> %vrndi1.i
+}
+
+define dso_local <8 x half> @t_vrndiq_f16(<8 x half> %a) {
+; CHECK-LABEL: t_vrndiq_f16:
+; CHECK:         frinti v0.8h, v0.8h
+; CHECK-NEXT:    ret
+entry:
+  %vrndi1.i = tail call <8 x half> @llvm.nearbyint.v8f16(<8 x half> %a)
+  ret <8 x half> %vrndi1.i
+}
+
+define dso_local <4 x half> @t_vsqrt_f16(<4 x half> %a) {
+; CHECK-LABEL: t_vsqrt_f16:
+; CHECK:         fsqrt v0.4h, v0.4h
+; CHECK-NEXT:    ret
+entry:
+  %vsqrt.i = tail call <4 x half> @llvm.sqrt.v4f16(<4 x half> %a)
+  ret <4 x half> %vsqrt.i
+}
+
+define dso_local <8 x half> @t_vsqrtq_f16(<8 x half> %a) {
+; CHECK-LABEL: t_vsqrtq_f16:
+; CHECK:         fsqrt v0.8h, v0.8h
+; CHECK-NEXT:    ret
+entry:
+  %vsqrt.i = tail call <8 x half> @llvm.sqrt.v8f16(<8 x half> %a)
+  ret <8 x half> %vsqrt.i
+}
Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_2op.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_2op.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_2op.ll
@@ -0,0 +1,80 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half>, <4 x half>)
+declare <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half>, <8 x half>)
+declare <4 x half> @llvm.aarch64.neon.fminnmp.v4f16(<4 x half>, <4 x half>)
+declare <8 x half> @llvm.aarch64.neon.fminnmp.v8f16(<8 x half>, <8 x half>)
+declare <4 x half> @llvm.aarch64.neon.fmaxnmp.v4f16(<4 x half>, <4 x half>)
+declare <8 x half> @llvm.aarch64.neon.fmaxnmp.v8f16(<8 x half>, <8 x half>)
+
+define dso_local <4 x half> @t_vdiv_f16(<4 x half> %a, <4 x half> %b) {
+; CHECK-LABEL: t_vdiv_f16:
+; CHECK:         fdiv v0.4h, v0.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %div.i = fdiv <4 x half> %a, %b
+  ret <4 x half> %div.i
+}
+
+define dso_local <8 x half> @t_vdivq_f16(<8 x half> %a, <8 x half> %b) {
+; CHECK-LABEL: t_vdivq_f16:
+; CHECK:         fdiv v0.8h, v0.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %div.i = fdiv <8 x half> %a, %b
+  ret <8 x half> %div.i
+}
+
+define dso_local <4 x half> @t_vmulx_f16(<4 x half> %a, <4 x half> %b) {
+; CHECK-LABEL: t_vmulx_f16:
+; CHECK:         fmulx v0.4h, v0.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %vmulx2.i = tail call <4 x half> @llvm.aarch64.neon.fmulx.v4f16(<4 x half> %a, <4 x half> %b)
+  ret <4 x half> %vmulx2.i
+}
+
+define dso_local <8 x half> @t_vmulxq_f16(<8 x half> %a, <8 x half> %b) {
+; CHECK-LABEL: t_vmulxq_f16:
+; CHECK:         fmulx v0.8h, v0.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %vmulx2.i = tail call <8 x half> @llvm.aarch64.neon.fmulx.v8f16(<8 x half> %a, <8 x half> %b)
+  ret <8 x half> %vmulx2.i
+}
+
+define dso_local <4 x half> @t_vpminnm_f16(<4 x half> %a, <4 x half> %b) {
+; CHECK-LABEL: t_vpminnm_f16:
+; CHECK:         fminnmp v0.4h, v0.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %vpminnm2.i = tail call <4 x half> @llvm.aarch64.neon.fminnmp.v4f16(<4 x half> %a, <4 x half> %b)
+  ret <4 x half> %vpminnm2.i
+}
+
+define dso_local <8 x half> @t_vpminnmq_f16(<8 x half> %a, <8 x half> %b) {
+; CHECK-LABEL: t_vpminnmq_f16:
+; CHECK:         fminnmp v0.8h, v0.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %vpminnm2.i = tail call <8 x half> @llvm.aarch64.neon.fminnmp.v8f16(<8 x half> %a, <8 x half> %b)
+  ret <8 x half> %vpminnm2.i
+}
+
+define dso_local <4 x half> @t_vpmaxnm_f16(<4 x half> %a, <4 x half> %b) {
+; CHECK-LABEL: t_vpmaxnm_f16:
+; CHECK:         fmaxnmp v0.4h, v0.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %vpmaxnm2.i = tail call <4 x half> @llvm.aarch64.neon.fmaxnmp.v4f16(<4 x half> %a, <4 x half> %b)
+  ret <4 x half> %vpmaxnm2.i
+}
+
+define dso_local <8 x half> @t_vpmaxnmq_f16(<8 x half> %a, <8 x half> %b) {
+; CHECK-LABEL: t_vpmaxnmq_f16:
+; CHECK:         fmaxnmp v0.8h, v0.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %vpmaxnm2.i = tail call <8 x half> @llvm.aarch64.neon.fmaxnmp.v8f16(<8 x half> %a, <8 x half> %b)
+  ret <8 x half> %vpmaxnm2.i
+}
Index: llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_3op.ll
===================================================================
--- llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_3op.ll
+++ llvm/trunk/test/CodeGen/AArch64/fp16_intrinsic_vector_3op.ll
@@ -0,0 +1,22 @@
+; RUN: llc < %s -mtriple=aarch64-eabi -mattr=+v8.2a,+fullfp16  | FileCheck %s
+
+declare <4 x half> @llvm.fma.v4f16(<4 x half>, <4 x half>, <4 x half>)
+declare <8 x half> @llvm.fma.v8f16(<8 x half>, <8 x half>, <8 x half>)
+
+define dso_local <4 x half> @t_vfma_f16(<4 x half> %a, <4 x half> %b, <4 x half> %c) {
+; CHECK-LABEL: t_vfma_f16:
+; CHECK:         fmla v0.4h, v2.4h, v1.4h
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call <4 x half> @llvm.fma.v4f16(<4 x half> %b, <4 x half> %c, <4 x half> %a)
+  ret <4 x half> %0
+}
+
+define dso_local <8 x half> @t_vfmaq_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
+; CHECK-LABEL: t_vfmaq_f16:
+; CHECK:         fmla v0.8h, v2.8h, v1.8h
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call <8 x half> @llvm.fma.v8f16(<8 x half> %b, <8 x half> %c, <8 x half> %a)
+  ret <8 x half> %0
+}