1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py 2; RUN: opt -matrix-allow-contract=false -passes='lower-matrix-intrinsics' -S < %s | FileCheck %s 3 4 5; Make sure we propagate in multiple iterations. First, we back-propagate the 6; shape information from the transpose to %A, in the next iteration we 7; forward-propagate it to %Mul, and then back to %B. 8define <16 x double> @backpropagation_iterations(<16 x double>* %A.Ptr, <16 x double>* %B.Ptr) { 9; CHECK-LABEL: @backpropagation_iterations( 10; CHECK-NEXT: [[TMP1:%.*]] = bitcast <16 x double>* [[A_PTR:%.*]] to double* 11; CHECK-NEXT: [[VEC_CAST:%.*]] = bitcast double* [[TMP1]] to <4 x double>* 12; CHECK-NEXT: [[COL_LOAD:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST]], align 8 13; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, double* [[TMP1]], i64 4 14; CHECK-NEXT: [[VEC_CAST1:%.*]] = bitcast double* [[VEC_GEP]] to <4 x double>* 15; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST1]], align 8 16; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr double, double* [[TMP1]], i64 8 17; CHECK-NEXT: [[VEC_CAST4:%.*]] = bitcast double* [[VEC_GEP3]] to <4 x double>* 18; CHECK-NEXT: [[COL_LOAD5:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST4]], align 8 19; CHECK-NEXT: [[VEC_GEP6:%.*]] = getelementptr double, double* [[TMP1]], i64 12 20; CHECK-NEXT: [[VEC_CAST7:%.*]] = bitcast double* [[VEC_GEP6]] to <4 x double>* 21; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST7]], align 8 22; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 0 23; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x double> poison, double [[TMP2]], i64 0 24; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x double> [[COL_LOAD2]], i64 0 25; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x double> [[TMP3]], double [[TMP4]], i64 1 26; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 0 27; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x double> [[TMP5]], double [[TMP6]], i64 2 28; CHECK-NEXT: [[TMP8:%.*]] = extractelement <4 x double> [[COL_LOAD8]], i64 0 29; CHECK-NEXT: [[TMP9:%.*]] = insertelement <4 x double> [[TMP7]], double [[TMP8]], i64 3 30; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 1 31; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x double> poison, double [[TMP10]], i64 0 32; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x double> [[COL_LOAD2]], i64 1 33; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x double> [[TMP11]], double [[TMP12]], i64 1 34; CHECK-NEXT: [[TMP14:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 1 35; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x double> [[TMP13]], double [[TMP14]], i64 2 36; CHECK-NEXT: [[TMP16:%.*]] = extractelement <4 x double> [[COL_LOAD8]], i64 1 37; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x double> [[TMP15]], double [[TMP16]], i64 3 38; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 2 39; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x double> poison, double [[TMP18]], i64 0 40; CHECK-NEXT: [[TMP20:%.*]] = extractelement <4 x double> [[COL_LOAD2]], i64 2 41; CHECK-NEXT: [[TMP21:%.*]] = insertelement <4 x double> [[TMP19]], double [[TMP20]], i64 1 42; CHECK-NEXT: [[TMP22:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 2 43; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x double> [[TMP21]], double [[TMP22]], i64 2 44; CHECK-NEXT: [[TMP24:%.*]] = extractelement <4 x double> [[COL_LOAD8]], i64 2 45; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x double> [[TMP23]], double [[TMP24]], i64 3 46; CHECK-NEXT: [[TMP26:%.*]] = extractelement <4 x double> [[COL_LOAD]], i64 3 47; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x double> poison, double [[TMP26]], i64 0 48; CHECK-NEXT: [[TMP28:%.*]] = extractelement <4 x double> [[COL_LOAD2]], i64 3 49; CHECK-NEXT: [[TMP29:%.*]] = insertelement <4 x double> [[TMP27]], double [[TMP28]], i64 1 50; CHECK-NEXT: [[TMP30:%.*]] = extractelement <4 x double> [[COL_LOAD5]], i64 3 51; CHECK-NEXT: [[TMP31:%.*]] = insertelement <4 x double> [[TMP29]], double [[TMP30]], i64 2 52; CHECK-NEXT: [[TMP32:%.*]] = extractelement <4 x double> [[COL_LOAD8]], i64 3 53; CHECK-NEXT: [[TMP33:%.*]] = insertelement <4 x double> [[TMP31]], double [[TMP32]], i64 3 54; CHECK-NEXT: [[TMP34:%.*]] = bitcast <16 x double>* [[B_PTR:%.*]] to double* 55; CHECK-NEXT: [[VEC_CAST9:%.*]] = bitcast double* [[TMP34]] to <4 x double>* 56; CHECK-NEXT: [[COL_LOAD10:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST9]], align 8 57; CHECK-NEXT: [[VEC_GEP11:%.*]] = getelementptr double, double* [[TMP34]], i64 4 58; CHECK-NEXT: [[VEC_CAST12:%.*]] = bitcast double* [[VEC_GEP11]] to <4 x double>* 59; CHECK-NEXT: [[COL_LOAD13:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST12]], align 8 60; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr double, double* [[TMP34]], i64 8 61; CHECK-NEXT: [[VEC_CAST15:%.*]] = bitcast double* [[VEC_GEP14]] to <4 x double>* 62; CHECK-NEXT: [[COL_LOAD16:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST15]], align 8 63; CHECK-NEXT: [[VEC_GEP17:%.*]] = getelementptr double, double* [[TMP34]], i64 12 64; CHECK-NEXT: [[VEC_CAST18:%.*]] = bitcast double* [[VEC_GEP17]] to <4 x double>* 65; CHECK-NEXT: [[COL_LOAD19:%.*]] = load <4 x double>, <4 x double>* [[VEC_CAST18]], align 8 66; CHECK-NEXT: [[TMP35:%.*]] = fmul <4 x double> [[COL_LOAD]], [[COL_LOAD10]] 67; CHECK-NEXT: [[TMP36:%.*]] = fmul <4 x double> [[COL_LOAD2]], [[COL_LOAD13]] 68; CHECK-NEXT: [[TMP37:%.*]] = fmul <4 x double> [[COL_LOAD5]], [[COL_LOAD16]] 69; CHECK-NEXT: [[TMP38:%.*]] = fmul <4 x double> [[COL_LOAD8]], [[COL_LOAD19]] 70; CHECK-NEXT: [[TMP39:%.*]] = 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> 71; CHECK-NEXT: [[TMP40:%.*]] = shufflevector <4 x double> [[TMP37]], <4 x double> [[TMP38]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> 72; CHECK-NEXT: [[TMP41:%.*]] = shufflevector <8 x double> [[TMP39]], <8 x double> [[TMP40]], <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> 73; CHECK-NEXT: ret <16 x double> [[TMP41]] 74; 75 %A = load <16 x double>, <16 x double>* %A.Ptr, align 8 76 %A.trans = tail call <16 x double> @llvm.matrix.transpose.v16f64(<16 x double> %A, i32 4, i32 4) 77 %B = load <16 x double>, <16 x double>* %B.Ptr, align 8 78 %Mul = fmul <16 x double> %A, %B 79 ret <16 x double> %Mul 80} 81 82declare <16 x double> @llvm.matrix.multiply.v16f64.v16f64.v16f64(<16 x double>, <16 x double>, i32 immarg, i32 immarg, i32 immarg) 83declare <16 x double> @llvm.matrix.transpose.v16f64(<16 x double>, i32 immarg, i32 immarg) 84