| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py |
| ; RUN: opt -passes='lower-matrix-intrinsics' -S < %s | FileCheck %s |
| |
| ; Make sure we use a flattened vector when calling @foo and the use its flat vector result properly. |
| define <9 x double> @unsupported_call(i1 %cond, <9 x double> %A, <9 x double> %B) { |
| ; CHECK-LABEL: @unsupported_call( |
| ; CHECK-NEXT: [[SPLIT:%.*]] = shufflevector <9 x double> [[A:%.*]], <9 x double> poison, <3 x i32> <i32 0, i32 1, i32 2> |
| ; CHECK-NEXT: [[SPLIT1:%.*]] = shufflevector <9 x double> [[A]], <9 x double> poison, <3 x i32> <i32 3, i32 4, i32 5> |
| ; CHECK-NEXT: [[SPLIT2:%.*]] = shufflevector <9 x double> [[A]], <9 x double> poison, <3 x i32> <i32 6, i32 7, i32 8> |
| ; CHECK-NEXT: [[TMP1:%.*]] = extractelement <3 x double> [[SPLIT]], i64 0 |
| ; CHECK-NEXT: [[TMP2:%.*]] = insertelement <3 x double> poison, double [[TMP1]], i64 0 |
| ; CHECK-NEXT: [[TMP3:%.*]] = extractelement <3 x double> [[SPLIT1]], i64 0 |
| ; CHECK-NEXT: [[TMP4:%.*]] = insertelement <3 x double> [[TMP2]], double [[TMP3]], i64 1 |
| ; CHECK-NEXT: [[TMP5:%.*]] = extractelement <3 x double> [[SPLIT2]], i64 0 |
| ; CHECK-NEXT: [[TMP6:%.*]] = insertelement <3 x double> [[TMP4]], double [[TMP5]], i64 2 |
| ; CHECK-NEXT: [[TMP7:%.*]] = extractelement <3 x double> [[SPLIT]], i64 1 |
| ; CHECK-NEXT: [[TMP8:%.*]] = insertelement <3 x double> poison, double [[TMP7]], i64 0 |
| ; CHECK-NEXT: [[TMP9:%.*]] = extractelement <3 x double> [[SPLIT1]], i64 1 |
| ; CHECK-NEXT: [[TMP10:%.*]] = insertelement <3 x double> [[TMP8]], double [[TMP9]], i64 1 |
| ; CHECK-NEXT: [[TMP11:%.*]] = extractelement <3 x double> [[SPLIT2]], i64 1 |
| ; CHECK-NEXT: [[TMP12:%.*]] = insertelement <3 x double> [[TMP10]], double [[TMP11]], i64 2 |
| ; CHECK-NEXT: [[TMP13:%.*]] = extractelement <3 x double> [[SPLIT]], i64 2 |
| ; CHECK-NEXT: [[TMP14:%.*]] = insertelement <3 x double> poison, double [[TMP13]], i64 0 |
| ; CHECK-NEXT: [[TMP15:%.*]] = extractelement <3 x double> [[SPLIT1]], i64 2 |
| ; CHECK-NEXT: [[TMP16:%.*]] = insertelement <3 x double> [[TMP14]], double [[TMP15]], i64 1 |
| ; CHECK-NEXT: [[TMP17:%.*]] = extractelement <3 x double> [[SPLIT2]], i64 2 |
| ; CHECK-NEXT: [[TMP18:%.*]] = insertelement <3 x double> [[TMP16]], double [[TMP17]], i64 2 |
| ; CHECK-NEXT: [[TMP19:%.*]] = shufflevector <3 x double> [[TMP6]], <3 x double> [[TMP12]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5> |
| ; CHECK-NEXT: [[TMP20:%.*]] = shufflevector <3 x double> [[TMP18]], <3 x double> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 poison, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP21:%.*]] = shufflevector <6 x double> [[TMP19]], <6 x double> [[TMP20]], <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8> |
| ; CHECK-NEXT: [[A_FOO:%.*]] = call <9 x double> @foo(<9 x double> [[TMP21]]) |
| ; CHECK-NEXT: [[SPLIT3:%.*]] = shufflevector <9 x double> [[B:%.*]], <9 x double> poison, <3 x i32> <i32 0, i32 1, i32 2> |
| ; CHECK-NEXT: [[SPLIT4:%.*]] = shufflevector <9 x double> [[B]], <9 x double> poison, <3 x i32> <i32 3, i32 4, i32 5> |
| ; CHECK-NEXT: [[SPLIT5:%.*]] = shufflevector <9 x double> [[B]], <9 x double> poison, <3 x i32> <i32 6, i32 7, i32 8> |
| ; CHECK-NEXT: [[SPLIT6:%.*]] = shufflevector <9 x double> [[A_FOO]], <9 x double> poison, <3 x i32> <i32 0, i32 1, i32 2> |
| ; CHECK-NEXT: [[SPLIT7:%.*]] = shufflevector <9 x double> [[A_FOO]], <9 x double> poison, <3 x i32> <i32 3, i32 4, i32 5> |
| ; CHECK-NEXT: [[SPLIT8:%.*]] = shufflevector <9 x double> [[A_FOO]], <9 x double> poison, <3 x i32> <i32 6, i32 7, i32 8> |
| ; CHECK-NEXT: [[BLOCK:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP22:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> poison, double [[TMP22]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP23:%.*]] = fmul <1 x double> [[BLOCK]], [[SPLAT_SPLAT]] |
| ; CHECK-NEXT: [[BLOCK9:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP24:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT10:%.*]] = insertelement <1 x double> poison, double [[TMP24]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT11:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT10]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP25:%.*]] = fmul <1 x double> [[BLOCK9]], [[SPLAT_SPLAT11]] |
| ; CHECK-NEXT: [[TMP26:%.*]] = fadd <1 x double> [[TMP23]], [[TMP25]] |
| ; CHECK-NEXT: [[BLOCK12:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP27:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT13:%.*]] = insertelement <1 x double> poison, double [[TMP27]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT14:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT13]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP28:%.*]] = fmul <1 x double> [[BLOCK12]], [[SPLAT_SPLAT14]] |
| ; CHECK-NEXT: [[TMP29:%.*]] = fadd <1 x double> [[TMP26]], [[TMP28]] |
| ; CHECK-NEXT: [[TMP30:%.*]] = shufflevector <1 x double> [[TMP29]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP31:%.*]] = shufflevector <3 x double> poison, <3 x double> [[TMP30]], <3 x i32> <i32 3, i32 1, i32 2> |
| ; CHECK-NEXT: [[BLOCK15:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP32:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT16:%.*]] = insertelement <1 x double> poison, double [[TMP32]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT17:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT16]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP33:%.*]] = fmul <1 x double> [[BLOCK15]], [[SPLAT_SPLAT17]] |
| ; CHECK-NEXT: [[BLOCK18:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP34:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT19:%.*]] = insertelement <1 x double> poison, double [[TMP34]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT20:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT19]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP35:%.*]] = fmul <1 x double> [[BLOCK18]], [[SPLAT_SPLAT20]] |
| ; CHECK-NEXT: [[TMP36:%.*]] = fadd <1 x double> [[TMP33]], [[TMP35]] |
| ; CHECK-NEXT: [[BLOCK21:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP37:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT22:%.*]] = insertelement <1 x double> poison, double [[TMP37]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT23:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT22]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP38:%.*]] = fmul <1 x double> [[BLOCK21]], [[SPLAT_SPLAT23]] |
| ; CHECK-NEXT: [[TMP39:%.*]] = fadd <1 x double> [[TMP36]], [[TMP38]] |
| ; CHECK-NEXT: [[TMP40:%.*]] = shufflevector <1 x double> [[TMP39]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP41:%.*]] = shufflevector <3 x double> [[TMP31]], <3 x double> [[TMP40]], <3 x i32> <i32 0, i32 3, i32 2> |
| ; CHECK-NEXT: [[BLOCK24:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP42:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT25:%.*]] = insertelement <1 x double> poison, double [[TMP42]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT26:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT25]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP43:%.*]] = fmul <1 x double> [[BLOCK24]], [[SPLAT_SPLAT26]] |
| ; CHECK-NEXT: [[BLOCK27:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP44:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT28:%.*]] = insertelement <1 x double> poison, double [[TMP44]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT29:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT28]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP45:%.*]] = fmul <1 x double> [[BLOCK27]], [[SPLAT_SPLAT29]] |
| ; CHECK-NEXT: [[TMP46:%.*]] = fadd <1 x double> [[TMP43]], [[TMP45]] |
| ; CHECK-NEXT: [[BLOCK30:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP47:%.*]] = extractelement <3 x double> [[SPLIT6]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT31:%.*]] = insertelement <1 x double> poison, double [[TMP47]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT32:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT31]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP48:%.*]] = fmul <1 x double> [[BLOCK30]], [[SPLAT_SPLAT32]] |
| ; CHECK-NEXT: [[TMP49:%.*]] = fadd <1 x double> [[TMP46]], [[TMP48]] |
| ; CHECK-NEXT: [[TMP50:%.*]] = shufflevector <1 x double> [[TMP49]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP51:%.*]] = shufflevector <3 x double> [[TMP41]], <3 x double> [[TMP50]], <3 x i32> <i32 0, i32 1, i32 3> |
| ; CHECK-NEXT: [[BLOCK33:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP52:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT34:%.*]] = insertelement <1 x double> poison, double [[TMP52]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT35:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT34]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP53:%.*]] = fmul <1 x double> [[BLOCK33]], [[SPLAT_SPLAT35]] |
| ; CHECK-NEXT: [[BLOCK36:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP54:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT37:%.*]] = insertelement <1 x double> poison, double [[TMP54]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT38:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT37]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP55:%.*]] = fmul <1 x double> [[BLOCK36]], [[SPLAT_SPLAT38]] |
| ; CHECK-NEXT: [[TMP56:%.*]] = fadd <1 x double> [[TMP53]], [[TMP55]] |
| ; CHECK-NEXT: [[BLOCK39:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP57:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT40:%.*]] = insertelement <1 x double> poison, double [[TMP57]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT41:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT40]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP58:%.*]] = fmul <1 x double> [[BLOCK39]], [[SPLAT_SPLAT41]] |
| ; CHECK-NEXT: [[TMP59:%.*]] = fadd <1 x double> [[TMP56]], [[TMP58]] |
| ; CHECK-NEXT: [[TMP60:%.*]] = shufflevector <1 x double> [[TMP59]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP61:%.*]] = shufflevector <3 x double> poison, <3 x double> [[TMP60]], <3 x i32> <i32 3, i32 1, i32 2> |
| ; CHECK-NEXT: [[BLOCK42:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP62:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT43:%.*]] = insertelement <1 x double> poison, double [[TMP62]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT44:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT43]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP63:%.*]] = fmul <1 x double> [[BLOCK42]], [[SPLAT_SPLAT44]] |
| ; CHECK-NEXT: [[BLOCK45:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP64:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT46:%.*]] = insertelement <1 x double> poison, double [[TMP64]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT47:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT46]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP65:%.*]] = fmul <1 x double> [[BLOCK45]], [[SPLAT_SPLAT47]] |
| ; CHECK-NEXT: [[TMP66:%.*]] = fadd <1 x double> [[TMP63]], [[TMP65]] |
| ; CHECK-NEXT: [[BLOCK48:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP67:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT49:%.*]] = insertelement <1 x double> poison, double [[TMP67]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT50:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT49]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP68:%.*]] = fmul <1 x double> [[BLOCK48]], [[SPLAT_SPLAT50]] |
| ; CHECK-NEXT: [[TMP69:%.*]] = fadd <1 x double> [[TMP66]], [[TMP68]] |
| ; CHECK-NEXT: [[TMP70:%.*]] = shufflevector <1 x double> [[TMP69]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP71:%.*]] = shufflevector <3 x double> [[TMP61]], <3 x double> [[TMP70]], <3 x i32> <i32 0, i32 3, i32 2> |
| ; CHECK-NEXT: [[BLOCK51:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP72:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT52:%.*]] = insertelement <1 x double> poison, double [[TMP72]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT53:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT52]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP73:%.*]] = fmul <1 x double> [[BLOCK51]], [[SPLAT_SPLAT53]] |
| ; CHECK-NEXT: [[BLOCK54:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP74:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT55:%.*]] = insertelement <1 x double> poison, double [[TMP74]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT56:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT55]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP75:%.*]] = fmul <1 x double> [[BLOCK54]], [[SPLAT_SPLAT56]] |
| ; CHECK-NEXT: [[TMP76:%.*]] = fadd <1 x double> [[TMP73]], [[TMP75]] |
| ; CHECK-NEXT: [[BLOCK57:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP77:%.*]] = extractelement <3 x double> [[SPLIT7]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT58:%.*]] = insertelement <1 x double> poison, double [[TMP77]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT59:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT58]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP78:%.*]] = fmul <1 x double> [[BLOCK57]], [[SPLAT_SPLAT59]] |
| ; CHECK-NEXT: [[TMP79:%.*]] = fadd <1 x double> [[TMP76]], [[TMP78]] |
| ; CHECK-NEXT: [[TMP80:%.*]] = shufflevector <1 x double> [[TMP79]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP81:%.*]] = shufflevector <3 x double> [[TMP71]], <3 x double> [[TMP80]], <3 x i32> <i32 0, i32 1, i32 3> |
| ; CHECK-NEXT: [[BLOCK60:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP82:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT61:%.*]] = insertelement <1 x double> poison, double [[TMP82]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT62:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT61]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP83:%.*]] = fmul <1 x double> [[BLOCK60]], [[SPLAT_SPLAT62]] |
| ; CHECK-NEXT: [[BLOCK63:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP84:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT64:%.*]] = insertelement <1 x double> poison, double [[TMP84]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT65:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT64]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP85:%.*]] = fmul <1 x double> [[BLOCK63]], [[SPLAT_SPLAT65]] |
| ; CHECK-NEXT: [[TMP86:%.*]] = fadd <1 x double> [[TMP83]], [[TMP85]] |
| ; CHECK-NEXT: [[BLOCK66:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP87:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT67:%.*]] = insertelement <1 x double> poison, double [[TMP87]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT68:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT67]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP88:%.*]] = fmul <1 x double> [[BLOCK66]], [[SPLAT_SPLAT68]] |
| ; CHECK-NEXT: [[TMP89:%.*]] = fadd <1 x double> [[TMP86]], [[TMP88]] |
| ; CHECK-NEXT: [[TMP90:%.*]] = shufflevector <1 x double> [[TMP89]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP91:%.*]] = shufflevector <3 x double> poison, <3 x double> [[TMP90]], <3 x i32> <i32 3, i32 1, i32 2> |
| ; CHECK-NEXT: [[BLOCK69:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP92:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT70:%.*]] = insertelement <1 x double> poison, double [[TMP92]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT71:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT70]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP93:%.*]] = fmul <1 x double> [[BLOCK69]], [[SPLAT_SPLAT71]] |
| ; CHECK-NEXT: [[BLOCK72:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP94:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT73:%.*]] = insertelement <1 x double> poison, double [[TMP94]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT74:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT73]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP95:%.*]] = fmul <1 x double> [[BLOCK72]], [[SPLAT_SPLAT74]] |
| ; CHECK-NEXT: [[TMP96:%.*]] = fadd <1 x double> [[TMP93]], [[TMP95]] |
| ; CHECK-NEXT: [[BLOCK75:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> <i32 1> |
| ; CHECK-NEXT: [[TMP97:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT76:%.*]] = insertelement <1 x double> poison, double [[TMP97]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT77:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT76]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP98:%.*]] = fmul <1 x double> [[BLOCK75]], [[SPLAT_SPLAT77]] |
| ; CHECK-NEXT: [[TMP99:%.*]] = fadd <1 x double> [[TMP96]], [[TMP98]] |
| ; CHECK-NEXT: [[TMP100:%.*]] = shufflevector <1 x double> [[TMP99]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP101:%.*]] = shufflevector <3 x double> [[TMP91]], <3 x double> [[TMP100]], <3 x i32> <i32 0, i32 3, i32 2> |
| ; CHECK-NEXT: [[BLOCK78:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP102:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT79:%.*]] = insertelement <1 x double> poison, double [[TMP102]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT80:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT79]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP103:%.*]] = fmul <1 x double> [[BLOCK78]], [[SPLAT_SPLAT80]] |
| ; CHECK-NEXT: [[BLOCK81:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP104:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 1 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT82:%.*]] = insertelement <1 x double> poison, double [[TMP104]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT83:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT82]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP105:%.*]] = fmul <1 x double> [[BLOCK81]], [[SPLAT_SPLAT83]] |
| ; CHECK-NEXT: [[TMP106:%.*]] = fadd <1 x double> [[TMP103]], [[TMP105]] |
| ; CHECK-NEXT: [[BLOCK84:%.*]] = shufflevector <3 x double> [[SPLIT5]], <3 x double> poison, <1 x i32> <i32 2> |
| ; CHECK-NEXT: [[TMP107:%.*]] = extractelement <3 x double> [[SPLIT8]], i64 2 |
| ; CHECK-NEXT: [[SPLAT_SPLATINSERT85:%.*]] = insertelement <1 x double> poison, double [[TMP107]], i64 0 |
| ; CHECK-NEXT: [[SPLAT_SPLAT86:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT85]], <1 x double> poison, <1 x i32> zeroinitializer |
| ; CHECK-NEXT: [[TMP108:%.*]] = fmul <1 x double> [[BLOCK84]], [[SPLAT_SPLAT86]] |
| ; CHECK-NEXT: [[TMP109:%.*]] = fadd <1 x double> [[TMP106]], [[TMP108]] |
| ; CHECK-NEXT: [[TMP110:%.*]] = shufflevector <1 x double> [[TMP109]], <1 x double> poison, <3 x i32> <i32 0, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP111:%.*]] = shufflevector <3 x double> [[TMP101]], <3 x double> [[TMP110]], <3 x i32> <i32 0, i32 1, i32 3> |
| ; CHECK-NEXT: [[TMP112:%.*]] = shufflevector <3 x double> [[TMP51]], <3 x double> [[TMP81]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5> |
| ; CHECK-NEXT: [[TMP113:%.*]] = shufflevector <3 x double> [[TMP111]], <3 x double> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 poison, i32 poison, i32 poison> |
| ; CHECK-NEXT: [[TMP114:%.*]] = shufflevector <6 x double> [[TMP112]], <6 x double> [[TMP113]], <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8> |
| ; CHECK-NEXT: ret <9 x double> [[TMP114]] |
| ; |
| %A.trans = tail call <9 x double> @llvm.matrix.transpose.v9f64(<9 x double> %A, i32 3, i32 3) |
| %A.foo = call <9 x double> @foo(<9 x double> %A.trans) |
| %res = tail call <9 x double> @llvm.matrix.multiply.v9f64.v9f64.v9f64(<9 x double> %B, <9 x double> %A.foo, i32 3, i32 3, i32 3) |
| ret <9 x double> %res |
| } |
| |
| declare <9 x double> @llvm.matrix.multiply.v9f64.v9f64.v9f64(<9 x double>, <9 x double>, i32 immarg, i32 immarg, i32 immarg) |
| declare <9 x double> @llvm.matrix.transpose.v9f64(<9 x double>, i32 immarg, i32 immarg) |
| declare <9 x double> @foo(<9 x double>) |
