; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes=lower-matrix-intrinsics,instcombine -fuse-matrix-use-loops=false -fuse-matrix-tile-size=1 -matrix-allow-contract -force-fuse-matrix -verify-dom-info %s -S | FileCheck %s

; REQUIRES: aarch64-registered-target

target datalayout = "e-m:o-i64:64-f80:128-n8:4:32:64-S128"
target triple = "aarch64-apple-ios"

define void @multiply_can_hoist_cast(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply_can_hoist_cast(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[STORE_BEGIN:%.*]] = ptrtoint ptr [[C:%.*]] to i64
; CHECK-NEXT:    [[STORE_END:%.*]] = add nuw nsw i64 [[STORE_BEGIN]], 32
; CHECK-NEXT:    [[LOAD_BEGIN:%.*]] = ptrtoint ptr [[B:%.*]] to i64
; CHECK-NEXT:    [[TMP0:%.*]] = icmp ugt i64 [[STORE_END]], [[LOAD_BEGIN]]
; CHECK-NEXT:    br i1 [[TMP0]], label [[ALIAS_CONT:%.*]], label [[NO_ALIAS:%.*]]
; CHECK:       alias_cont:
; CHECK-NEXT:    [[LOAD_END:%.*]] = add nuw nsw i64 [[LOAD_BEGIN]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = icmp ugt i64 [[LOAD_END]], [[STORE_BEGIN]]
; CHECK-NEXT:    br i1 [[TMP1]], label [[COPY:%.*]], label [[NO_ALIAS]]
; CHECK:       copy:
; CHECK-NEXT:    [[TMP2:%.*]] = alloca [4 x double], align 8
; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(32) [[TMP2]], ptr noundef nonnull align 8 dereferenceable(32) [[B]], i64 32, i1 false)
; CHECK-NEXT:    br label [[NO_ALIAS]]
; CHECK:       no_alias:
; CHECK-NEXT:    [[TMP3:%.*]] = phi ptr [ [[B]], [[ENTRY:%.*]] ], [ [[B]], [[ALIAS_CONT]] ], [ [[TMP2]], [[COPY]] ]
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <1 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[TMP4:%.*]] = fmul contract <1 x double> [[COL_LOAD]], [[COL_LOAD1]]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[TMP5]], align 8
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[TMP6]], align 8
; CHECK-NEXT:    [[TMP7:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD2]], <1 x double> [[COL_LOAD3]], <1 x double> [[TMP4]])
; CHECK-NEXT:    store <1 x double> [[TMP7]], ptr [[C]], align 8
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr double, ptr [[A]], i64 1
; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[TMP8]], align 8
; CHECK-NEXT:    [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[TMP9:%.*]] = fmul contract <1 x double> [[COL_LOAD8]], [[COL_LOAD9]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 3
; CHECK-NEXT:    [[COL_LOAD13:%.*]] = load <1 x double>, ptr [[TMP10]], align 8
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
; CHECK-NEXT:    [[COL_LOAD14:%.*]] = load <1 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT:    [[TMP12:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD13]], <1 x double> [[COL_LOAD14]], <1 x double> [[TMP9]])
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr double, ptr [[C]], i64 1
; CHECK-NEXT:    store <1 x double> [[TMP12]], ptr [[TMP13]], align 8
; CHECK-NEXT:    [[COL_LOAD19:%.*]] = load <1 x double>, ptr [[A]], align 8
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
; CHECK-NEXT:    [[COL_LOAD20:%.*]] = load <1 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT:    [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD19]], [[COL_LOAD20]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[TMP16]], align 8
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
; CHECK-NEXT:    [[COL_LOAD25:%.*]] = load <1 x double>, ptr [[TMP17]], align 8
; CHECK-NEXT:    [[TMP18:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[COL_LOAD25]], <1 x double> [[TMP15]])
; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr double, ptr [[C]], i64 2
; CHECK-NEXT:    store <1 x double> [[TMP18]], ptr [[TMP19]], align 8
; CHECK-NEXT:    [[TMP20:%.*]] = getelementptr double, ptr [[A]], i64 1
; CHECK-NEXT:    [[COL_LOAD30:%.*]] = load <1 x double>, ptr [[TMP20]], align 8
; CHECK-NEXT:    [[TMP21:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
; CHECK-NEXT:    [[COL_LOAD31:%.*]] = load <1 x double>, ptr [[TMP21]], align 8
; CHECK-NEXT:    [[TMP22:%.*]] = fmul contract <1 x double> [[COL_LOAD30]], [[COL_LOAD31]]
; CHECK-NEXT:    [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 3
; CHECK-NEXT:    [[COL_LOAD35:%.*]] = load <1 x double>, ptr [[TMP23]], align 8
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
; CHECK-NEXT:    [[COL_LOAD36:%.*]] = load <1 x double>, ptr [[TMP24]], align 8
; CHECK-NEXT:    [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD35]], <1 x double> [[COL_LOAD36]], <1 x double> [[TMP22]])
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr double, ptr [[C]], i64 3
; CHECK-NEXT:    store <1 x double> [[TMP25]], ptr [[TMP26]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  store <4 x double> %c, ptr %C, align 8
  ret void
}

define void @multiply_can_hoist_multiple_insts(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply_can_hoist_multiple_insts(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr [4 x double], ptr [[C:%.*]], i64 2
; CHECK-NEXT:    [[STORE_BEGIN:%.*]] = ptrtoint ptr [[GEP]] to i64
; CHECK-NEXT:    [[STORE_END:%.*]] = add nuw nsw i64 [[STORE_BEGIN]], 32
; CHECK-NEXT:    [[LOAD_BEGIN:%.*]] = ptrtoint ptr [[B:%.*]] to i64
; CHECK-NEXT:    [[TMP0:%.*]] = icmp ugt i64 [[STORE_END]], [[LOAD_BEGIN]]
; CHECK-NEXT:    br i1 [[TMP0]], label [[ALIAS_CONT:%.*]], label [[NO_ALIAS:%.*]]
; CHECK:       alias_cont:
; CHECK-NEXT:    [[LOAD_END:%.*]] = add nuw nsw i64 [[LOAD_BEGIN]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = icmp ugt i64 [[LOAD_END]], [[STORE_BEGIN]]
; CHECK-NEXT:    br i1 [[TMP1]], label [[COPY:%.*]], label [[NO_ALIAS]]
; CHECK:       copy:
; CHECK-NEXT:    [[TMP2:%.*]] = alloca [4 x double], align 8
; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(32) [[TMP2]], ptr noundef nonnull align 8 dereferenceable(32) [[B]], i64 32, i1 false)
; CHECK-NEXT:    br label [[NO_ALIAS]]
; CHECK:       no_alias:
; CHECK-NEXT:    [[TMP3:%.*]] = phi ptr [ [[B]], [[ENTRY:%.*]] ], [ [[B]], [[ALIAS_CONT]] ], [ [[TMP2]], [[COPY]] ]
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <1 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[TMP4:%.*]] = fmul contract <1 x double> [[COL_LOAD]], [[COL_LOAD1]]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[TMP5]], align 8
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[TMP6]], align 8
; CHECK-NEXT:    [[TMP7:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD2]], <1 x double> [[COL_LOAD3]], <1 x double> [[TMP4]])
; CHECK-NEXT:    store <1 x double> [[TMP7]], ptr [[GEP]], align 8
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr double, ptr [[A]], i64 1
; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[TMP8]], align 8
; CHECK-NEXT:    [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[TMP9:%.*]] = fmul contract <1 x double> [[COL_LOAD8]], [[COL_LOAD9]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 3
; CHECK-NEXT:    [[COL_LOAD13:%.*]] = load <1 x double>, ptr [[TMP10]], align 8
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
; CHECK-NEXT:    [[COL_LOAD14:%.*]] = load <1 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT:    [[TMP12:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD13]], <1 x double> [[COL_LOAD14]], <1 x double> [[TMP9]])
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr [4 x double], ptr [[C]], i64 2, i64 1
; CHECK-NEXT:    store <1 x double> [[TMP12]], ptr [[TMP13]], align 8
; CHECK-NEXT:    [[COL_LOAD19:%.*]] = load <1 x double>, ptr [[A]], align 8
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
; CHECK-NEXT:    [[COL_LOAD20:%.*]] = load <1 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT:    [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD19]], [[COL_LOAD20]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[TMP16]], align 8
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
; CHECK-NEXT:    [[COL_LOAD25:%.*]] = load <1 x double>, ptr [[TMP17]], align 8
; CHECK-NEXT:    [[TMP18:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[COL_LOAD25]], <1 x double> [[TMP15]])
; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr [4 x double], ptr [[C]], i64 2, i64 2
; CHECK-NEXT:    store <1 x double> [[TMP18]], ptr [[TMP19]], align 8
; CHECK-NEXT:    [[TMP20:%.*]] = getelementptr double, ptr [[A]], i64 1
; CHECK-NEXT:    [[COL_LOAD30:%.*]] = load <1 x double>, ptr [[TMP20]], align 8
; CHECK-NEXT:    [[TMP21:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
; CHECK-NEXT:    [[COL_LOAD31:%.*]] = load <1 x double>, ptr [[TMP21]], align 8
; CHECK-NEXT:    [[TMP22:%.*]] = fmul contract <1 x double> [[COL_LOAD30]], [[COL_LOAD31]]
; CHECK-NEXT:    [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 3
; CHECK-NEXT:    [[COL_LOAD35:%.*]] = load <1 x double>, ptr [[TMP23]], align 8
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
; CHECK-NEXT:    [[COL_LOAD36:%.*]] = load <1 x double>, ptr [[TMP24]], align 8
; CHECK-NEXT:    [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD35]], <1 x double> [[COL_LOAD36]], <1 x double> [[TMP22]])
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr [4 x double], ptr [[C]], i64 2, i64 3
; CHECK-NEXT:    store <1 x double> [[TMP25]], ptr [[TMP26]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  %gep = getelementptr [4 x double], ptr %C, i32 2
  store <4 x double> %c, ptr %gep, align 8
  ret void
}

; Make sure the correct instruction order is preserved when hoisting.
define void @multiply_can_hoist_multiple_insts2(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply_can_hoist_multiple_insts2(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[GEP_1:%.*]] = getelementptr <4 x double>, ptr [[C:%.*]], i64 42
; CHECK-NEXT:    [[STORE_BEGIN:%.*]] = ptrtoint ptr [[GEP_1]] to i64
; CHECK-NEXT:    [[STORE_END:%.*]] = add nuw nsw i64 [[STORE_BEGIN]], 32
; CHECK-NEXT:    [[LOAD_BEGIN:%.*]] = ptrtoint ptr [[B:%.*]] to i64
; CHECK-NEXT:    [[TMP0:%.*]] = icmp ugt i64 [[STORE_END]], [[LOAD_BEGIN]]
; CHECK-NEXT:    br i1 [[TMP0]], label [[ALIAS_CONT:%.*]], label [[NO_ALIAS:%.*]]
; CHECK:       alias_cont:
; CHECK-NEXT:    [[LOAD_END:%.*]] = add nuw nsw i64 [[LOAD_BEGIN]], 32
; CHECK-NEXT:    [[TMP1:%.*]] = icmp ugt i64 [[LOAD_END]], [[STORE_BEGIN]]
; CHECK-NEXT:    br i1 [[TMP1]], label [[COPY:%.*]], label [[NO_ALIAS]]
; CHECK:       copy:
; CHECK-NEXT:    [[TMP2:%.*]] = alloca [4 x double], align 8
; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(32) [[TMP2]], ptr noundef nonnull align 8 dereferenceable(32) [[B]], i64 32, i1 false)
; CHECK-NEXT:    br label [[NO_ALIAS]]
; CHECK:       no_alias:
; CHECK-NEXT:    [[TMP3:%.*]] = phi ptr [ [[B]], [[ENTRY:%.*]] ], [ [[B]], [[ALIAS_CONT]] ], [ [[TMP2]], [[COPY]] ]
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <1 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[TMP4:%.*]] = fmul contract <1 x double> [[COL_LOAD]], [[COL_LOAD1]]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[TMP5]], align 8
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[TMP6]], align 8
; CHECK-NEXT:    [[TMP7:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD2]], <1 x double> [[COL_LOAD3]], <1 x double> [[TMP4]])
; CHECK-NEXT:    store <1 x double> [[TMP7]], ptr [[GEP_1]], align 8
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr double, ptr [[A]], i64 1
; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[TMP8]], align 8
; CHECK-NEXT:    [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[TMP9:%.*]] = fmul contract <1 x double> [[COL_LOAD8]], [[COL_LOAD9]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 3
; CHECK-NEXT:    [[COL_LOAD13:%.*]] = load <1 x double>, ptr [[TMP10]], align 8
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
; CHECK-NEXT:    [[COL_LOAD14:%.*]] = load <1 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT:    [[TMP12:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD13]], <1 x double> [[COL_LOAD14]], <1 x double> [[TMP9]])
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[C]], i64 1352
; CHECK-NEXT:    store <1 x double> [[TMP12]], ptr [[TMP13]], align 8
; CHECK-NEXT:    [[COL_LOAD19:%.*]] = load <1 x double>, ptr [[A]], align 8
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
; CHECK-NEXT:    [[COL_LOAD20:%.*]] = load <1 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT:    [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD19]], [[COL_LOAD20]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[TMP16]], align 8
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
; CHECK-NEXT:    [[COL_LOAD25:%.*]] = load <1 x double>, ptr [[TMP17]], align 8
; CHECK-NEXT:    [[TMP18:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[COL_LOAD25]], <1 x double> [[TMP15]])
; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr i8, ptr [[C]], i64 1360
; CHECK-NEXT:    store <1 x double> [[TMP18]], ptr [[TMP19]], align 8
; CHECK-NEXT:    [[TMP20:%.*]] = getelementptr double, ptr [[A]], i64 1
; CHECK-NEXT:    [[COL_LOAD30:%.*]] = load <1 x double>, ptr [[TMP20]], align 8
; CHECK-NEXT:    [[TMP21:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
; CHECK-NEXT:    [[COL_LOAD31:%.*]] = load <1 x double>, ptr [[TMP21]], align 8
; CHECK-NEXT:    [[TMP22:%.*]] = fmul contract <1 x double> [[COL_LOAD30]], [[COL_LOAD31]]
; CHECK-NEXT:    [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 3
; CHECK-NEXT:    [[COL_LOAD35:%.*]] = load <1 x double>, ptr [[TMP23]], align 8
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
; CHECK-NEXT:    [[COL_LOAD36:%.*]] = load <1 x double>, ptr [[TMP24]], align 8
; CHECK-NEXT:    [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD35]], <1 x double> [[COL_LOAD36]], <1 x double> [[TMP22]])
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr i8, ptr [[C]], i64 1368
; CHECK-NEXT:    store <1 x double> [[TMP25]], ptr [[TMP26]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  %off.0 = add i32 10, 10
  %off.1 = add i32 %off.0, 2
  %off.2 = add i32 %off.0, %off.1
  %gep.1 = getelementptr <4 x double>, ptr %C, i32 %off.2
  store <4 x double> %c, ptr %gep.1, align 8
  ret void
}

define void @multiply_dont_hoist_phi(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply_dont_hoist_phi(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br label [[NEXT:%.*]]
; CHECK:       next:
; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[A:%.*]], i64 2
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B:%.*]], i64 2
; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A]], align 8
; CHECK-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT10]]
; CHECK-NEXT:    [[TMP1:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP0]])
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B]], align 8
; CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP2:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
; CHECK-NEXT:    [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT7]], <2 x double> [[TMP2]])
; CHECK-NEXT:    [[GEP_1:%.*]] = getelementptr <4 x double>, ptr [[C:%.*]], i64 26
; CHECK-NEXT:    store <2 x double> [[TMP3]], ptr [[GEP_1]], align 8
; CHECK-NEXT:    [[VEC_GEP14:%.*]] = getelementptr i8, ptr [[C]], i64 848
; CHECK-NEXT:    store <2 x double> [[TMP1]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  br label %next

next:
  %p = phi i32 [ 2, %entry ]
  %off.0 = add i32 10, %p
  %off.1 = add i32 %off.0, 2
  %off.2 = add i32 %off.0, %off.1
  %gep.1 = getelementptr <4 x double>, ptr %C, i32 %off.2
  store <4 x double> %c, ptr %gep.1, align 8
  ret void
}

; The address load may alias, so avoid moving it for now.
define void @multiply_dont_hoist_cast_due_to_operand(ptr noalias %A, ptr %B, ptr %C.ptr) {
; CHECK-LABEL: @multiply_dont_hoist_cast_due_to_operand(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B]], i64 2
; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
; CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[TMP1:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT7]], <2 x double> [[TMP0]])
; CHECK-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP2:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT10]]
; CHECK-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
; CHECK-NEXT:    [[C:%.*]] = load ptr, ptr [[C_PTR:%.*]], align 8
; CHECK-NEXT:    store <2 x double> [[TMP1]], ptr [[C]], align 8
; CHECK-NEXT:    [[VEC_GEP14:%.*]] = getelementptr double, ptr [[C]], i64 2
; CHECK-NEXT:    store <2 x double> [[TMP3]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  %C = load ptr, ptr %C.ptr
  store <4 x double> %c, ptr %C, align 8
  ret void
}

; The address load may alias, so avoid moving it for now.
define void @multiply_dont_hoist_load(ptr noalias %A, ptr %B, ptr %C.ptr) {
; CHECK-LABEL: @multiply_dont_hoist_load(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B]], i64 2
; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
; CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[TMP1:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT7]], <2 x double> [[TMP0]])
; CHECK-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP2:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT10]]
; CHECK-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
; CHECK-NEXT:    [[C:%.*]] = load ptr, ptr [[C_PTR:%.*]], align 8
; CHECK-NEXT:    store <2 x double> [[TMP1]], ptr [[C]], align 8
; CHECK-NEXT:    [[VEC_GEP14:%.*]] = getelementptr double, ptr [[C]], i64 2
; CHECK-NEXT:    store <2 x double> [[TMP3]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  %C = load ptr, ptr %C.ptr
  store <4 x double> %c, ptr %C, align 8
  ret void
}

; The call to @get_adress may clobber memory, avoid moving it for now.
define void @multiply_dont_hoist_call(ptr noalias %A, ptr %B) {
; CHECK-LABEL: @multiply_dont_hoist_call(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B]], i64 2
; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
; CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[TMP1:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT7]], <2 x double> [[TMP0]])
; CHECK-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP2:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT10]]
; CHECK-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT:    [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
; CHECK-NEXT:    [[C:%.*]] = call ptr @get_address()
; CHECK-NEXT:    store <2 x double> [[TMP1]], ptr [[C]], align 8
; CHECK-NEXT:    [[VEC_GEP14:%.*]] = getelementptr double, ptr [[C]], i64 2
; CHECK-NEXT:    store <2 x double> [[TMP3]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT:    ret void
;
entry:
  %a = load <4 x double>, ptr %A, align 8
  %b = load <4 x double>, ptr %B, align 8
  %c = call <4 x double> @llvm.matrix.multiply(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
  %C = call ptr @get_address()
  store <4 x double> %c, ptr %C, align 8
  ret void
}

declare ptr @get_address()


declare <4 x double> @llvm.matrix.multiply(<4 x double>, <4 x double>, i32, i32, i32)