; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -matrix-allow-contract=false -passes='lower-matrix-intrinsics' -S < %s | FileCheck %s


; Make sure we propagate in multiple iterations. First, we back-propagate the
; shape information from the transpose to %A, in the next iteration we
; forward-propagate it to %Mul, and then back to %B.
define <16 x double> @backpropagation_iterations(ptr %A.Ptr, ptr %B.Ptr) {
; CHECK-LABEL: @backpropagation_iterations(
; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <4 x double>, ptr [[A_PTR:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[A_PTR]], i64 4
; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <4 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT:    [[VEC_GEP2:%.*]] = getelementptr double, ptr [[A_PTR]], i64 8
; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <4 x double>, ptr [[VEC_GEP2]], align 8
; CHECK-NEXT:    [[VEC_GEP4:%.*]] = getelementptr double, ptr [[A_PTR]], i64 12
; CHECK-NEXT:    [[COL_LOAD5:%.*]] = load <4 x double>, ptr [[VEC_GEP4]], align 8
; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 0
; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <4 x double> poison, double [[TMP1]], i64 0
; CHECK-NEXT:    [[TMP3:%.*]] = extractelement <4 x double> [[COL_LOAD1]], i64 0
; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <4 x double> [[TMP2]], double [[TMP3]], i64 1
; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <4 x double> [[COL_LOAD3]], i64 0
; CHECK-NEXT:    [[TMP6:%.*]] = insertelement <4 x double> [[TMP4]], double [[TMP5]], i64 2
; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 0
; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <4 x double> [[TMP6]], double [[TMP7]], i64 3
; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 1
; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <4 x double> poison, double [[TMP9]], i64 0
; CHECK-NEXT:    [[TMP11:%.*]] = extractelement <4 x double> [[COL_LOAD1]], i64 1
; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <4 x double> [[TMP10]], double [[TMP11]], i64 1
; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <4 x double> [[COL_LOAD3]], i64 1
; CHECK-NEXT:    [[TMP14:%.*]] = insertelement <4 x double> [[TMP12]], double [[TMP13]], i64 2
; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 1
; CHECK-NEXT:    [[TMP16:%.*]] = insertelement <4 x double> [[TMP14]], double [[TMP15]], i64 3
; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 2
; CHECK-NEXT:    [[TMP18:%.*]] = insertelement <4 x double> poison, double [[TMP17]], i64 0
; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <4 x double> [[COL_LOAD1]], i64 2
; CHECK-NEXT:    [[TMP20:%.*]] = insertelement <4 x double> [[TMP18]], double [[TMP19]], i64 1
; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <4 x double> [[COL_LOAD3]], i64 2
; CHECK-NEXT:    [[TMP22:%.*]] = insertelement <4 x double> [[TMP20]], double [[TMP21]], i64 2
; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 2
; CHECK-NEXT:    [[TMP24:%.*]] = insertelement <4 x double> [[TMP22]], double [[TMP23]], i64 3
; CHECK-NEXT:    [[TMP25:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 3
; CHECK-NEXT:    [[TMP26:%.*]] = insertelement <4 x double> poison, double [[TMP25]], i64 0
; CHECK-NEXT:    [[TMP27:%.*]] = extractelement <4 x double> [[COL_LOAD1]], i64 3
; CHECK-NEXT:    [[TMP28:%.*]] = insertelement <4 x double> [[TMP26]], double [[TMP27]], i64 1
; CHECK-NEXT:    [[TMP29:%.*]] = extractelement <4 x double> [[COL_LOAD3]], i64 3
; CHECK-NEXT:    [[TMP30:%.*]] = insertelement <4 x double> [[TMP28]], double [[TMP29]], i64 2
; CHECK-NEXT:    [[TMP31:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 3
; CHECK-NEXT:    [[TMP32:%.*]] = insertelement <4 x double> [[TMP30]], double [[TMP31]], i64 3
; CHECK-NEXT:    [[COL_LOAD6:%.*]] = load <4 x double>, ptr [[B_PTR:%.*]], align 8
; CHECK-NEXT:    [[VEC_GEP7:%.*]] = getelementptr double, ptr [[B_PTR]], i64 4
; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <4 x double>, ptr [[VEC_GEP7]], align 8
; CHECK-NEXT:    [[VEC_GEP9:%.*]] = getelementptr double, ptr [[B_PTR]], i64 8
; CHECK-NEXT:    [[COL_LOAD10:%.*]] = load <4 x double>, ptr [[VEC_GEP9]], align 8
; CHECK-NEXT:    [[VEC_GEP11:%.*]] = getelementptr double, ptr [[B_PTR]], i64 12
; CHECK-NEXT:    [[COL_LOAD12:%.*]] = load <4 x double>, ptr [[VEC_GEP11]], align 8
; CHECK-NEXT:    [[TMP33:%.*]] = fmul <4 x double> [[COL_LOAD]], [[COL_LOAD6]]
; CHECK-NEXT:    [[TMP34:%.*]] = fmul <4 x double> [[COL_LOAD1]], [[COL_LOAD8]]
; CHECK-NEXT:    [[TMP35:%.*]] = fmul <4 x double> [[COL_LOAD3]], [[COL_LOAD10]]
; CHECK-NEXT:    [[TMP36:%.*]] = fmul <4 x double> [[COL_LOAD5]], [[COL_LOAD12]]
; CHECK-NEXT:    [[TMP37:%.*]] = shufflevector <4 x double> [[TMP33]], <4 x double> [[TMP34]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; CHECK-NEXT:    [[TMP38:%.*]] = shufflevector <4 x double> [[TMP35]], <4 x double> [[TMP36]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; CHECK-NEXT:    [[TMP39:%.*]] = shufflevector <8 x double> [[TMP37]], <8 x double> [[TMP38]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
; CHECK-NEXT:    ret <16 x double> [[TMP39]]
;
  %A = load <16 x double>, ptr %A.Ptr, align 8
  %A.trans = tail call <16 x double> @llvm.matrix.transpose.v16f64(<16 x double> %A, i32 4, i32 4)
  %B = load <16 x double>, ptr %B.Ptr, align 8
  %Mul = fmul <16 x double> %A, %B
  ret <16 x double> %Mul
}

declare <16 x double> @llvm.matrix.multiply.v16f64.v16f64.v16f64(<16 x double>, <16 x double>, i32 immarg, i32 immarg, i32 immarg)
declare <16 x double> @llvm.matrix.transpose.v16f64(<16 x double>, i32 immarg, i32 immarg)