brintos / llvm-project-archived public Read only

1// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py2// RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +zve64d \3// RUN: -target-feature +f -target-feature +d -disable-O0-optnone \4// RUN: -mvscale-min=4 -mvscale-max=4 -emit-llvm -o - %s | \5// RUN: opt -S -passes=sroa | FileCheck %s6 7// REQUIRES: riscv-registered-target8 9#include <stdint.h>10 11typedef __rvv_int8m1_t vint8m1_t;12typedef __rvv_uint8m1_t vuint8m1_t;13typedef __rvv_int16m1_t vint16m1_t;14typedef __rvv_uint16m1_t vuint16m1_t;15typedef __rvv_int32m1_t vint32m1_t;16typedef __rvv_uint32m1_t vuint32m1_t;17typedef __rvv_int64m1_t vint64m1_t;18typedef __rvv_uint64m1_t vuint64m1_t;19typedef __rvv_float32m1_t vfloat32m1_t;20typedef __rvv_float64m1_t vfloat64m1_t;21 22typedef vint8m1_t fixed_int8m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));23typedef vint16m1_t fixed_int16m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));24typedef vint32m1_t fixed_int32m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));25typedef vint64m1_t fixed_int64m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));26 27typedef vuint8m1_t fixed_uint8m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));28typedef vuint16m1_t fixed_uint16m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));29typedef vuint32m1_t fixed_uint32m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));30typedef vuint64m1_t fixed_uint64m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));31 32typedef vfloat32m1_t fixed_float32m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));33typedef vfloat64m1_t fixed_float64m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));34 35// ADDITION36 37// CHECK-LABEL: @add_i8(38// CHECK-NEXT:  entry:39// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)40// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)41// CHECK-NEXT:    [[ADD:%.*]] = add <32 x i8> [[A]], [[B]]42// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[ADD]], i64 0)43// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]44//45fixed_int8m1_t add_i8(fixed_int8m1_t a, fixed_int8m1_t b) {46  return a + b;47}48 49// CHECK-LABEL: @add_i16(50// CHECK-NEXT:  entry:51// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)52// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)53// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i16> [[A]], [[B]]54// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[ADD]], i64 0)55// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]56//57fixed_int16m1_t add_i16(fixed_int16m1_t a, fixed_int16m1_t b) {58  return a + b;59}60 61// CHECK-LABEL: @add_i32(62// CHECK-NEXT:  entry:63// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)64// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)65// CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[A]], [[B]]66// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[ADD]], i64 0)67// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]68//69fixed_int32m1_t add_i32(fixed_int32m1_t a, fixed_int32m1_t b) {70  return a + b;71}72 73// CHECK-LABEL: @add_i64(74// CHECK-NEXT:  entry:75// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)76// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)77// CHECK-NEXT:    [[ADD:%.*]] = add <4 x i64> [[A]], [[B]]78// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[ADD]], i64 0)79// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]80//81fixed_int64m1_t add_i64(fixed_int64m1_t a, fixed_int64m1_t b) {82  return a + b;83}84 85// CHECK-LABEL: @add_u8(86// CHECK-NEXT:  entry:87// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)88// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)89// CHECK-NEXT:    [[ADD:%.*]] = add <32 x i8> [[A]], [[B]]90// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[ADD]], i64 0)91// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]92//93fixed_uint8m1_t add_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {94  return a + b;95}96 97// CHECK-LABEL: @add_u16(98// CHECK-NEXT:  entry:99// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)100// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)101// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i16> [[A]], [[B]]102// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[ADD]], i64 0)103// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]104//105fixed_uint16m1_t add_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {106  return a + b;107}108 109// CHECK-LABEL: @add_u32(110// CHECK-NEXT:  entry:111// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)112// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)113// CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[A]], [[B]]114// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[ADD]], i64 0)115// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]116//117fixed_uint32m1_t add_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {118  return a + b;119}120 121// CHECK-LABEL: @add_u64(122// CHECK-NEXT:  entry:123// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)124// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)125// CHECK-NEXT:    [[ADD:%.*]] = add <4 x i64> [[A]], [[B]]126// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[ADD]], i64 0)127// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]128//129fixed_uint64m1_t add_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {130  return a + b;131}132 133// CHECK-LABEL: @add_f32(134// CHECK-NEXT:  entry:135// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)136// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)137// CHECK-NEXT:    [[ADD:%.*]] = fadd <8 x float> [[A]], [[B]]138// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[ADD]], i64 0)139// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]140//141fixed_float32m1_t add_f32(fixed_float32m1_t a, fixed_float32m1_t b) {142  return a + b;143}144 145// CHECK-LABEL: @add_f64(146// CHECK-NEXT:  entry:147// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)148// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)149// CHECK-NEXT:    [[ADD:%.*]] = fadd <4 x double> [[A]], [[B]]150// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[ADD]], i64 0)151// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]152//153fixed_float64m1_t add_f64(fixed_float64m1_t a, fixed_float64m1_t b) {154  return a + b;155}156 157// CHECK-LABEL: @add_inplace_i8(158// CHECK-NEXT:  entry:159// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)160// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)161// CHECK-NEXT:    [[ADD:%.*]] = add <32 x i8> [[A]], [[B]]162// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[ADD]], i64 0)163// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]164//165fixed_int8m1_t add_inplace_i8(fixed_int8m1_t a, fixed_int8m1_t b) {166  return a += b;167}168 169// CHECK-LABEL: @add_inplace_i16(170// CHECK-NEXT:  entry:171// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)172// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)173// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i16> [[A]], [[B]]174// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[ADD]], i64 0)175// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]176//177fixed_int16m1_t add_inplace_i16(fixed_int16m1_t a, fixed_int16m1_t b) {178  return a += b;179}180 181// CHECK-LABEL: @add_inplace_i32(182// CHECK-NEXT:  entry:183// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)184// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)185// CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[A]], [[B]]186// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[ADD]], i64 0)187// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]188//189fixed_int32m1_t add_inplace_i32(fixed_int32m1_t a, fixed_int32m1_t b) {190  return a += b;191}192 193// CHECK-LABEL: @add_inplace_i64(194// CHECK-NEXT:  entry:195// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)196// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)197// CHECK-NEXT:    [[ADD:%.*]] = add <4 x i64> [[A]], [[B]]198// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[ADD]], i64 0)199// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]200//201fixed_int64m1_t add_inplace_i64(fixed_int64m1_t a, fixed_int64m1_t b) {202  return a += b;203}204 205// CHECK-LABEL: @add_inplace_u8(206// CHECK-NEXT:  entry:207// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)208// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)209// CHECK-NEXT:    [[ADD:%.*]] = add <32 x i8> [[A]], [[B]]210// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[ADD]], i64 0)211// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]212//213fixed_uint8m1_t add_inplace_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {214  return a += b;215}216 217// CHECK-LABEL: @add_inplace_u16(218// CHECK-NEXT:  entry:219// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)220// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)221// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i16> [[A]], [[B]]222// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[ADD]], i64 0)223// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]224//225fixed_uint16m1_t add_inplace_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {226  return a += b;227}228 229// CHECK-LABEL: @add_inplace_u32(230// CHECK-NEXT:  entry:231// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)232// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)233// CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[A]], [[B]]234// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[ADD]], i64 0)235// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]236//237fixed_uint32m1_t add_inplace_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {238  return a += b;239}240 241// CHECK-LABEL: @add_inplace_u64(242// CHECK-NEXT:  entry:243// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)244// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)245// CHECK-NEXT:    [[ADD:%.*]] = add <4 x i64> [[A]], [[B]]246// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[ADD]], i64 0)247// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]248//249fixed_uint64m1_t add_inplace_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {250  return a += b;251}252 253// CHECK-LABEL: @add_inplace_f32(254// CHECK-NEXT:  entry:255// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)256// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)257// CHECK-NEXT:    [[ADD:%.*]] = fadd <8 x float> [[A]], [[B]]258// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[ADD]], i64 0)259// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]260//261fixed_float32m1_t add_inplace_f32(fixed_float32m1_t a, fixed_float32m1_t b) {262  return a += b;263}264 265// CHECK-LABEL: @add_inplace_f64(266// CHECK-NEXT:  entry:267// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)268// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)269// CHECK-NEXT:    [[ADD:%.*]] = fadd <4 x double> [[A]], [[B]]270// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[ADD]], i64 0)271// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]272//273fixed_float64m1_t add_inplace_f64(fixed_float64m1_t a, fixed_float64m1_t b) {274  return a += b;275}276 277// CHECK-LABEL: @add_scalar_i8(278// CHECK-NEXT:  entry:279// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)280// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 0281// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer282// CHECK-NEXT:    [[ADD:%.*]] = add <32 x i8> [[A]], [[SPLAT_SPLAT]]283// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[ADD]], i64 0)284// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]285//286fixed_int8m1_t add_scalar_i8(fixed_int8m1_t a, int8_t b) {287  return a + b;288}289 290// CHECK-LABEL: @add_scalar_i16(291// CHECK-NEXT:  entry:292// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)293// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 0294// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer295// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i16> [[A]], [[SPLAT_SPLAT]]296// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[ADD]], i64 0)297// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]298//299fixed_int16m1_t add_scalar_i16(fixed_int16m1_t a, int16_t b) {300  return a + b;301}302 303// CHECK-LABEL: @add_scalar_i32(304// CHECK-NEXT:  entry:305// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)306// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 0307// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer308// CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[A]], [[SPLAT_SPLAT]]309// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[ADD]], i64 0)310// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]311//312fixed_int32m1_t add_scalar_i32(fixed_int32m1_t a, int32_t b) {313  return a + b;314}315 316// CHECK-LABEL: @add_scalar_i64(317// CHECK-NEXT:  entry:318// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)319// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 0320// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer321// CHECK-NEXT:    [[ADD:%.*]] = add <4 x i64> [[A]], [[SPLAT_SPLAT]]322// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[ADD]], i64 0)323// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]324//325fixed_int64m1_t add_scalar_i64(fixed_int64m1_t a, int64_t b) {326  return a + b;327}328 329// CHECK-LABEL: @add_scalar_u8(330// CHECK-NEXT:  entry:331// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)332// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 0333// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer334// CHECK-NEXT:    [[ADD:%.*]] = add <32 x i8> [[A]], [[SPLAT_SPLAT]]335// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[ADD]], i64 0)336// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]337//338fixed_uint8m1_t add_scalar_u8(fixed_uint8m1_t a, uint8_t b) {339  return a + b;340}341 342// CHECK-LABEL: @add_scalar_u16(343// CHECK-NEXT:  entry:344// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)345// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 0346// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer347// CHECK-NEXT:    [[ADD:%.*]] = add <16 x i16> [[A]], [[SPLAT_SPLAT]]348// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[ADD]], i64 0)349// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]350//351fixed_uint16m1_t add_scalar_u16(fixed_uint16m1_t a, uint16_t b) {352  return a + b;353}354 355// CHECK-LABEL: @add_scalar_u32(356// CHECK-NEXT:  entry:357// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)358// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 0359// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer360// CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[A]], [[SPLAT_SPLAT]]361// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[ADD]], i64 0)362// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]363//364fixed_uint32m1_t add_scalar_u32(fixed_uint32m1_t a, uint32_t b) {365  return a + b;366}367 368// CHECK-LABEL: @add_scalar_u64(369// CHECK-NEXT:  entry:370// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)371// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 0372// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer373// CHECK-NEXT:    [[ADD:%.*]] = add <4 x i64> [[A]], [[SPLAT_SPLAT]]374// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[ADD]], i64 0)375// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]376//377fixed_uint64m1_t add_scalar_u64(fixed_uint64m1_t a, uint64_t b) {378  return a + b;379}380 381// CHECK-LABEL: @add_scalar_f32(382// CHECK-NEXT:  entry:383// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)384// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x float> poison, float [[B:%.*]], i64 0385// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x float> [[SPLAT_SPLATINSERT]], <8 x float> poison, <8 x i32> zeroinitializer386// CHECK-NEXT:    [[ADD:%.*]] = fadd <8 x float> [[A]], [[SPLAT_SPLAT]]387// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[ADD]], i64 0)388// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]389//390fixed_float32m1_t add_scalar_f32(fixed_float32m1_t a, float b) {391  return a + b;392}393 394// CHECK-LABEL: @add_scalar_f64(395// CHECK-NEXT:  entry:396// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)397// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x double> poison, double [[B:%.*]], i64 0398// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x double> [[SPLAT_SPLATINSERT]], <4 x double> poison, <4 x i32> zeroinitializer399// CHECK-NEXT:    [[ADD:%.*]] = fadd <4 x double> [[A]], [[SPLAT_SPLAT]]400// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[ADD]], i64 0)401// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]402//403fixed_float64m1_t add_scalar_f64(fixed_float64m1_t a, double b) {404  return a + b;405}406 407// SUBTRACTION408 409// CHECK-LABEL: @sub_i8(410// CHECK-NEXT:  entry:411// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)412// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)413// CHECK-NEXT:    [[SUB:%.*]] = sub <32 x i8> [[A]], [[B]]414// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[SUB]], i64 0)415// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]416//417fixed_int8m1_t sub_i8(fixed_int8m1_t a, fixed_int8m1_t b) {418  return a - b;419}420 421// CHECK-LABEL: @sub_i16(422// CHECK-NEXT:  entry:423// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)424// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)425// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i16> [[A]], [[B]]426// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[SUB]], i64 0)427// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]428//429fixed_int16m1_t sub_i16(fixed_int16m1_t a, fixed_int16m1_t b) {430  return a - b;431}432 433// CHECK-LABEL: @sub_i32(434// CHECK-NEXT:  entry:435// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)436// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)437// CHECK-NEXT:    [[SUB:%.*]] = sub <8 x i32> [[A]], [[B]]438// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[SUB]], i64 0)439// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]440//441fixed_int32m1_t sub_i32(fixed_int32m1_t a, fixed_int32m1_t b) {442  return a - b;443}444 445// CHECK-LABEL: @sub_i64(446// CHECK-NEXT:  entry:447// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)448// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)449// CHECK-NEXT:    [[SUB:%.*]] = sub <4 x i64> [[A]], [[B]]450// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[SUB]], i64 0)451// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]452//453fixed_int64m1_t sub_i64(fixed_int64m1_t a, fixed_int64m1_t b) {454  return a - b;455}456 457// CHECK-LABEL: @sub_u8(458// CHECK-NEXT:  entry:459// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)460// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)461// CHECK-NEXT:    [[SUB:%.*]] = sub <32 x i8> [[A]], [[B]]462// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[SUB]], i64 0)463// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]464//465fixed_uint8m1_t sub_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {466  return a - b;467}468 469// CHECK-LABEL: @sub_u16(470// CHECK-NEXT:  entry:471// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)472// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)473// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i16> [[A]], [[B]]474// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[SUB]], i64 0)475// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]476//477fixed_uint16m1_t sub_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {478  return a - b;479}480 481// CHECK-LABEL: @sub_u32(482// CHECK-NEXT:  entry:483// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)484// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)485// CHECK-NEXT:    [[SUB:%.*]] = sub <8 x i32> [[A]], [[B]]486// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[SUB]], i64 0)487// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]488//489fixed_uint32m1_t sub_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {490  return a - b;491}492 493// CHECK-LABEL: @sub_u64(494// CHECK-NEXT:  entry:495// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)496// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)497// CHECK-NEXT:    [[SUB:%.*]] = sub <4 x i64> [[A]], [[B]]498// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[SUB]], i64 0)499// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]500//501fixed_uint64m1_t sub_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {502  return a - b;503}504 505// CHECK-LABEL: @sub_f32(506// CHECK-NEXT:  entry:507// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)508// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)509// CHECK-NEXT:    [[SUB:%.*]] = fsub <8 x float> [[A]], [[B]]510// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[SUB]], i64 0)511// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]512//513fixed_float32m1_t sub_f32(fixed_float32m1_t a, fixed_float32m1_t b) {514  return a - b;515}516 517// CHECK-LABEL: @sub_f64(518// CHECK-NEXT:  entry:519// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)520// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)521// CHECK-NEXT:    [[SUB:%.*]] = fsub <4 x double> [[A]], [[B]]522// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[SUB]], i64 0)523// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]524//525fixed_float64m1_t sub_f64(fixed_float64m1_t a, fixed_float64m1_t b) {526  return a - b;527}528 529// CHECK-LABEL: @sub_inplace_i8(530// CHECK-NEXT:  entry:531// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)532// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)533// CHECK-NEXT:    [[SUB:%.*]] = sub <32 x i8> [[A]], [[B]]534// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[SUB]], i64 0)535// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]536//537fixed_int8m1_t sub_inplace_i8(fixed_int8m1_t a, fixed_int8m1_t b) {538  return a - b;539}540 541// CHECK-LABEL: @sub_inplace_i16(542// CHECK-NEXT:  entry:543// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)544// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)545// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i16> [[A]], [[B]]546// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[SUB]], i64 0)547// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]548//549fixed_int16m1_t sub_inplace_i16(fixed_int16m1_t a, fixed_int16m1_t b) {550  return a - b;551}552 553// CHECK-LABEL: @sub_inplace_i32(554// CHECK-NEXT:  entry:555// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)556// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)557// CHECK-NEXT:    [[SUB:%.*]] = sub <8 x i32> [[A]], [[B]]558// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[SUB]], i64 0)559// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]560//561fixed_int32m1_t sub_inplace_i32(fixed_int32m1_t a, fixed_int32m1_t b) {562  return a - b;563}564 565// CHECK-LABEL: @sub_inplace_i64(566// CHECK-NEXT:  entry:567// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)568// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)569// CHECK-NEXT:    [[SUB:%.*]] = sub <4 x i64> [[A]], [[B]]570// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[SUB]], i64 0)571// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]572//573fixed_int64m1_t sub_inplace_i64(fixed_int64m1_t a, fixed_int64m1_t b) {574  return a - b;575}576 577// CHECK-LABEL: @sub_inplace_u8(578// CHECK-NEXT:  entry:579// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)580// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)581// CHECK-NEXT:    [[SUB:%.*]] = sub <32 x i8> [[A]], [[B]]582// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[SUB]], i64 0)583// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]584//585fixed_uint8m1_t sub_inplace_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {586  return a - b;587}588 589// CHECK-LABEL: @sub_inplace_u16(590// CHECK-NEXT:  entry:591// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)592// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)593// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i16> [[A]], [[B]]594// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[SUB]], i64 0)595// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]596//597fixed_uint16m1_t sub_inplace_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {598  return a - b;599}600 601// CHECK-LABEL: @sub_inplace_u32(602// CHECK-NEXT:  entry:603// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)604// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)605// CHECK-NEXT:    [[SUB:%.*]] = sub <8 x i32> [[A]], [[B]]606// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[SUB]], i64 0)607// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]608//609fixed_uint32m1_t sub_inplace_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {610  return a - b;611}612 613// CHECK-LABEL: @sub_inplace_u64(614// CHECK-NEXT:  entry:615// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)616// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)617// CHECK-NEXT:    [[SUB:%.*]] = sub <4 x i64> [[A]], [[B]]618// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[SUB]], i64 0)619// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]620//621fixed_uint64m1_t sub_inplace_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {622  return a - b;623}624 625// CHECK-LABEL: @sub_inplace_f32(626// CHECK-NEXT:  entry:627// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)628// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)629// CHECK-NEXT:    [[SUB:%.*]] = fsub <8 x float> [[A]], [[B]]630// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[SUB]], i64 0)631// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]632//633fixed_float32m1_t sub_inplace_f32(fixed_float32m1_t a, fixed_float32m1_t b) {634  return a - b;635}636 637// CHECK-LABEL: @sub_inplace_f64(638// CHECK-NEXT:  entry:639// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)640// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)641// CHECK-NEXT:    [[SUB:%.*]] = fsub <4 x double> [[A]], [[B]]642// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[SUB]], i64 0)643// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]644//645fixed_float64m1_t sub_inplace_f64(fixed_float64m1_t a, fixed_float64m1_t b) {646  return a - b;647}648 649// CHECK-LABEL: @sub_scalar_i8(650// CHECK-NEXT:  entry:651// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)652// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 0653// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer654// CHECK-NEXT:    [[SUB:%.*]] = sub <32 x i8> [[A]], [[SPLAT_SPLAT]]655// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[SUB]], i64 0)656// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]657//658fixed_int8m1_t sub_scalar_i8(fixed_int8m1_t a, int8_t b) {659  return a - b;660}661 662// CHECK-LABEL: @sub_scalar_i16(663// CHECK-NEXT:  entry:664// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)665// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 0666// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer667// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i16> [[A]], [[SPLAT_SPLAT]]668// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[SUB]], i64 0)669// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]670//671fixed_int16m1_t sub_scalar_i16(fixed_int16m1_t a, int16_t b) {672  return a - b;673}674 675// CHECK-LABEL: @sub_scalar_i32(676// CHECK-NEXT:  entry:677// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)678// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 0679// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer680// CHECK-NEXT:    [[SUB:%.*]] = sub <8 x i32> [[A]], [[SPLAT_SPLAT]]681// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[SUB]], i64 0)682// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]683//684fixed_int32m1_t sub_scalar_i32(fixed_int32m1_t a, int32_t b) {685  return a - b;686}687 688// CHECK-LABEL: @sub_scalar_i64(689// CHECK-NEXT:  entry:690// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)691// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 0692// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer693// CHECK-NEXT:    [[SUB:%.*]] = sub <4 x i64> [[A]], [[SPLAT_SPLAT]]694// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[SUB]], i64 0)695// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]696//697fixed_int64m1_t sub_scalar_i64(fixed_int64m1_t a, int64_t b) {698  return a - b;699}700 701// CHECK-LABEL: @sub_scalar_u8(702// CHECK-NEXT:  entry:703// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)704// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 0705// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer706// CHECK-NEXT:    [[SUB:%.*]] = sub <32 x i8> [[A]], [[SPLAT_SPLAT]]707// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[SUB]], i64 0)708// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]709//710fixed_uint8m1_t sub_scalar_u8(fixed_uint8m1_t a, uint8_t b) {711  return a - b;712}713 714// CHECK-LABEL: @sub_scalar_u16(715// CHECK-NEXT:  entry:716// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)717// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 0718// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer719// CHECK-NEXT:    [[SUB:%.*]] = sub <16 x i16> [[A]], [[SPLAT_SPLAT]]720// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[SUB]], i64 0)721// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]722//723fixed_uint16m1_t sub_scalar_u16(fixed_uint16m1_t a, uint16_t b) {724  return a - b;725}726 727// CHECK-LABEL: @sub_scalar_u32(728// CHECK-NEXT:  entry:729// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)730// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 0731// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer732// CHECK-NEXT:    [[SUB:%.*]] = sub <8 x i32> [[A]], [[SPLAT_SPLAT]]733// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[SUB]], i64 0)734// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]735//736fixed_uint32m1_t sub_scalar_u32(fixed_uint32m1_t a, uint32_t b) {737  return a - b;738}739 740// CHECK-LABEL: @sub_scalar_u64(741// CHECK-NEXT:  entry:742// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)743// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 0744// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer745// CHECK-NEXT:    [[SUB:%.*]] = sub <4 x i64> [[A]], [[SPLAT_SPLAT]]746// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[SUB]], i64 0)747// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]748//749fixed_uint64m1_t sub_scalar_u64(fixed_uint64m1_t a, uint64_t b) {750  return a - b;751}752 753// CHECK-LABEL: @sub_scalar_f32(754// CHECK-NEXT:  entry:755// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)756// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x float> poison, float [[B:%.*]], i64 0757// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x float> [[SPLAT_SPLATINSERT]], <8 x float> poison, <8 x i32> zeroinitializer758// CHECK-NEXT:    [[SUB:%.*]] = fsub <8 x float> [[A]], [[SPLAT_SPLAT]]759// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[SUB]], i64 0)760// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]761//762fixed_float32m1_t sub_scalar_f32(fixed_float32m1_t a, float b) {763  return a - b;764}765 766// CHECK-LABEL: @sub_scalar_f64(767// CHECK-NEXT:  entry:768// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)769// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x double> poison, double [[B:%.*]], i64 0770// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x double> [[SPLAT_SPLATINSERT]], <4 x double> poison, <4 x i32> zeroinitializer771// CHECK-NEXT:    [[SUB:%.*]] = fsub <4 x double> [[A]], [[SPLAT_SPLAT]]772// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[SUB]], i64 0)773// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]774//775fixed_float64m1_t sub_scalar_f64(fixed_float64m1_t a, double b) {776  return a - b;777}778 779// MULTIPLICATION780 781// CHECK-LABEL: @mul_i8(782// CHECK-NEXT:  entry:783// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)784// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)785// CHECK-NEXT:    [[MUL:%.*]] = mul <32 x i8> [[A]], [[B]]786// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[MUL]], i64 0)787// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]788//789fixed_int8m1_t mul_i8(fixed_int8m1_t a, fixed_int8m1_t b) {790  return a * b;791}792 793// CHECK-LABEL: @mul_i16(794// CHECK-NEXT:  entry:795// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)796// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)797// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i16> [[A]], [[B]]798// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[MUL]], i64 0)799// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]800//801fixed_int16m1_t mul_i16(fixed_int16m1_t a, fixed_int16m1_t b) {802  return a * b;803}804 805// CHECK-LABEL: @mul_i32(806// CHECK-NEXT:  entry:807// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)808// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)809// CHECK-NEXT:    [[MUL:%.*]] = mul <8 x i32> [[A]], [[B]]810// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[MUL]], i64 0)811// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]812//813fixed_int32m1_t mul_i32(fixed_int32m1_t a, fixed_int32m1_t b) {814  return a * b;815}816 817// CHECK-LABEL: @mul_i64(818// CHECK-NEXT:  entry:819// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)820// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)821// CHECK-NEXT:    [[MUL:%.*]] = mul <4 x i64> [[A]], [[B]]822// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[MUL]], i64 0)823// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]824//825fixed_int64m1_t mul_i64(fixed_int64m1_t a, fixed_int64m1_t b) {826  return a * b;827}828 829// CHECK-LABEL: @mul_u8(830// CHECK-NEXT:  entry:831// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)832// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)833// CHECK-NEXT:    [[MUL:%.*]] = mul <32 x i8> [[A]], [[B]]834// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[MUL]], i64 0)835// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]836//837fixed_uint8m1_t mul_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {838  return a * b;839}840 841// CHECK-LABEL: @mul_u16(842// CHECK-NEXT:  entry:843// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)844// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)845// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i16> [[A]], [[B]]846// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[MUL]], i64 0)847// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]848//849fixed_uint16m1_t mul_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {850  return a * b;851}852 853// CHECK-LABEL: @mul_u32(854// CHECK-NEXT:  entry:855// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)856// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)857// CHECK-NEXT:    [[MUL:%.*]] = mul <8 x i32> [[A]], [[B]]858// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[MUL]], i64 0)859// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]860//861fixed_uint32m1_t mul_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {862  return a * b;863}864 865// CHECK-LABEL: @mul_u64(866// CHECK-NEXT:  entry:867// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)868// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)869// CHECK-NEXT:    [[MUL:%.*]] = mul <4 x i64> [[A]], [[B]]870// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[MUL]], i64 0)871// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]872//873fixed_uint64m1_t mul_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {874  return a * b;875}876 877// CHECK-LABEL: @mul_f32(878// CHECK-NEXT:  entry:879// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)880// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)881// CHECK-NEXT:    [[MUL:%.*]] = fmul <8 x float> [[A]], [[B]]882// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[MUL]], i64 0)883// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]884//885fixed_float32m1_t mul_f32(fixed_float32m1_t a, fixed_float32m1_t b) {886  return a * b;887}888 889// CHECK-LABEL: @mul_f64(890// CHECK-NEXT:  entry:891// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)892// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)893// CHECK-NEXT:    [[MUL:%.*]] = fmul <4 x double> [[A]], [[B]]894// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[MUL]], i64 0)895// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]896//897fixed_float64m1_t mul_f64(fixed_float64m1_t a, fixed_float64m1_t b) {898  return a * b;899}900 901// CHECK-LABEL: @mul_inplace_i8(902// CHECK-NEXT:  entry:903// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)904// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)905// CHECK-NEXT:    [[MUL:%.*]] = mul <32 x i8> [[A]], [[B]]906// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[MUL]], i64 0)907// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]908//909fixed_int8m1_t mul_inplace_i8(fixed_int8m1_t a, fixed_int8m1_t b) {910  return a * b;911}912 913// CHECK-LABEL: @mul_inplace_i16(914// CHECK-NEXT:  entry:915// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)916// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)917// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i16> [[A]], [[B]]918// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[MUL]], i64 0)919// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]920//921fixed_int16m1_t mul_inplace_i16(fixed_int16m1_t a, fixed_int16m1_t b) {922  return a * b;923}924 925// CHECK-LABEL: @mul_inplace_i32(926// CHECK-NEXT:  entry:927// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)928// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)929// CHECK-NEXT:    [[MUL:%.*]] = mul <8 x i32> [[A]], [[B]]930// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[MUL]], i64 0)931// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]932//933fixed_int32m1_t mul_inplace_i32(fixed_int32m1_t a, fixed_int32m1_t b) {934  return a * b;935}936 937// CHECK-LABEL: @mul_inplace_i64(938// CHECK-NEXT:  entry:939// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)940// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)941// CHECK-NEXT:    [[MUL:%.*]] = mul <4 x i64> [[A]], [[B]]942// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[MUL]], i64 0)943// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]944//945fixed_int64m1_t mul_inplace_i64(fixed_int64m1_t a, fixed_int64m1_t b) {946  return a * b;947}948 949// CHECK-LABEL: @mul_inplace_u8(950// CHECK-NEXT:  entry:951// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)952// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)953// CHECK-NEXT:    [[MUL:%.*]] = mul <32 x i8> [[A]], [[B]]954// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[MUL]], i64 0)955// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]956//957fixed_uint8m1_t mul_inplace_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {958  return a * b;959}960 961// CHECK-LABEL: @mul_inplace_u16(962// CHECK-NEXT:  entry:963// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)964// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)965// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i16> [[A]], [[B]]966// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[MUL]], i64 0)967// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]968//969fixed_uint16m1_t mul_inplace_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {970  return a * b;971}972 973// CHECK-LABEL: @mul_inplace_u32(974// CHECK-NEXT:  entry:975// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)976// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)977// CHECK-NEXT:    [[MUL:%.*]] = mul <8 x i32> [[A]], [[B]]978// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[MUL]], i64 0)979// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]980//981fixed_uint32m1_t mul_inplace_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {982  return a * b;983}984 985// CHECK-LABEL: @mul_inplace_u64(986// CHECK-NEXT:  entry:987// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)988// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)989// CHECK-NEXT:    [[MUL:%.*]] = mul <4 x i64> [[A]], [[B]]990// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[MUL]], i64 0)991// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]992//993fixed_uint64m1_t mul_inplace_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {994  return a * b;995}996 997// CHECK-LABEL: @mul_inplace_f32(998// CHECK-NEXT:  entry:999// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)1000// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)1001// CHECK-NEXT:    [[MUL:%.*]] = fmul <8 x float> [[A]], [[B]]1002// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[MUL]], i64 0)1003// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]1004//1005fixed_float32m1_t mul_inplace_f32(fixed_float32m1_t a, fixed_float32m1_t b) {1006  return a * b;1007}1008 1009// CHECK-LABEL: @mul_inplace_f64(1010// CHECK-NEXT:  entry:1011// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)1012// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)1013// CHECK-NEXT:    [[MUL:%.*]] = fmul <4 x double> [[A]], [[B]]1014// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[MUL]], i64 0)1015// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]1016//1017fixed_float64m1_t mul_inplace_f64(fixed_float64m1_t a, fixed_float64m1_t b) {1018  return a * b;1019}1020 1021// CHECK-LABEL: @mul_scalar_i8(1022// CHECK-NEXT:  entry:1023// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1024// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 01025// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer1026// CHECK-NEXT:    [[MUL:%.*]] = mul <32 x i8> [[A]], [[SPLAT_SPLAT]]1027// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[MUL]], i64 0)1028// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1029//1030fixed_int8m1_t mul_scalar_i8(fixed_int8m1_t a, int8_t b) {1031  return a * b;1032}1033 1034// CHECK-LABEL: @mul_scalar_i16(1035// CHECK-NEXT:  entry:1036// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1037// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 01038// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer1039// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i16> [[A]], [[SPLAT_SPLAT]]1040// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[MUL]], i64 0)1041// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1042//1043fixed_int16m1_t mul_scalar_i16(fixed_int16m1_t a, int16_t b) {1044  return a * b;1045}1046 1047// CHECK-LABEL: @mul_scalar_i32(1048// CHECK-NEXT:  entry:1049// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1050// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 01051// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer1052// CHECK-NEXT:    [[MUL:%.*]] = mul <8 x i32> [[A]], [[SPLAT_SPLAT]]1053// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[MUL]], i64 0)1054// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1055//1056fixed_int32m1_t mul_scalar_i32(fixed_int32m1_t a, int32_t b) {1057  return a * b;1058}1059 1060// CHECK-LABEL: @mul_scalar_i64(1061// CHECK-NEXT:  entry:1062// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1063// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 01064// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer1065// CHECK-NEXT:    [[MUL:%.*]] = mul <4 x i64> [[A]], [[SPLAT_SPLAT]]1066// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[MUL]], i64 0)1067// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1068//1069fixed_int64m1_t mul_scalar_i64(fixed_int64m1_t a, int64_t b) {1070  return a * b;1071}1072 1073// CHECK-LABEL: @mul_scalar_u8(1074// CHECK-NEXT:  entry:1075// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1076// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 01077// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer1078// CHECK-NEXT:    [[MUL:%.*]] = mul <32 x i8> [[A]], [[SPLAT_SPLAT]]1079// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[MUL]], i64 0)1080// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1081//1082fixed_uint8m1_t mul_scalar_u8(fixed_uint8m1_t a, uint8_t b) {1083  return a * b;1084}1085 1086// CHECK-LABEL: @mul_scalar_u16(1087// CHECK-NEXT:  entry:1088// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1089// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 01090// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer1091// CHECK-NEXT:    [[MUL:%.*]] = mul <16 x i16> [[A]], [[SPLAT_SPLAT]]1092// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[MUL]], i64 0)1093// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1094//1095fixed_uint16m1_t mul_scalar_u16(fixed_uint16m1_t a, uint16_t b) {1096  return a * b;1097}1098 1099// CHECK-LABEL: @mul_scalar_u32(1100// CHECK-NEXT:  entry:1101// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1102// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 01103// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer1104// CHECK-NEXT:    [[MUL:%.*]] = mul <8 x i32> [[A]], [[SPLAT_SPLAT]]1105// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[MUL]], i64 0)1106// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1107//1108fixed_uint32m1_t mul_scalar_u32(fixed_uint32m1_t a, uint32_t b) {1109  return a * b;1110}1111 1112// CHECK-LABEL: @mul_scalar_u64(1113// CHECK-NEXT:  entry:1114// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1115// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 01116// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer1117// CHECK-NEXT:    [[MUL:%.*]] = mul <4 x i64> [[A]], [[SPLAT_SPLAT]]1118// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[MUL]], i64 0)1119// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1120//1121fixed_uint64m1_t mul_scalar_u64(fixed_uint64m1_t a, uint64_t b) {1122  return a * b;1123}1124 1125// CHECK-LABEL: @mul_scalar_f32(1126// CHECK-NEXT:  entry:1127// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)1128// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x float> poison, float [[B:%.*]], i64 01129// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x float> [[SPLAT_SPLATINSERT]], <8 x float> poison, <8 x i32> zeroinitializer1130// CHECK-NEXT:    [[MUL:%.*]] = fmul <8 x float> [[A]], [[SPLAT_SPLAT]]1131// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[MUL]], i64 0)1132// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]1133//1134fixed_float32m1_t mul_scalar_f32(fixed_float32m1_t a, float b) {1135  return a * b;1136}1137 1138// CHECK-LABEL: @mul_scalar_f64(1139// CHECK-NEXT:  entry:1140// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)1141// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x double> poison, double [[B:%.*]], i64 01142// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x double> [[SPLAT_SPLATINSERT]], <4 x double> poison, <4 x i32> zeroinitializer1143// CHECK-NEXT:    [[MUL:%.*]] = fmul <4 x double> [[A]], [[SPLAT_SPLAT]]1144// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[MUL]], i64 0)1145// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]1146//1147fixed_float64m1_t mul_scalar_f64(fixed_float64m1_t a, double b) {1148  return a * b;1149}1150 1151// DIVISION1152 1153// CHECK-LABEL: @div_i8(1154// CHECK-NEXT:  entry:1155// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1156// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1157// CHECK-NEXT:    [[DIV:%.*]] = sdiv <32 x i8> [[A]], [[B]]1158// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[DIV]], i64 0)1159// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1160//1161fixed_int8m1_t div_i8(fixed_int8m1_t a, fixed_int8m1_t b) {1162  return a / b;1163}1164 1165// CHECK-LABEL: @div_i16(1166// CHECK-NEXT:  entry:1167// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1168// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1169// CHECK-NEXT:    [[DIV:%.*]] = sdiv <16 x i16> [[A]], [[B]]1170// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[DIV]], i64 0)1171// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1172//1173fixed_int16m1_t div_i16(fixed_int16m1_t a, fixed_int16m1_t b) {1174  return a / b;1175}1176 1177// CHECK-LABEL: @div_i32(1178// CHECK-NEXT:  entry:1179// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1180// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1181// CHECK-NEXT:    [[DIV:%.*]] = sdiv <8 x i32> [[A]], [[B]]1182// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[DIV]], i64 0)1183// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1184//1185fixed_int32m1_t div_i32(fixed_int32m1_t a, fixed_int32m1_t b) {1186  return a / b;1187}1188 1189// CHECK-LABEL: @div_i64(1190// CHECK-NEXT:  entry:1191// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1192// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1193// CHECK-NEXT:    [[DIV:%.*]] = sdiv <4 x i64> [[A]], [[B]]1194// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[DIV]], i64 0)1195// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1196//1197fixed_int64m1_t div_i64(fixed_int64m1_t a, fixed_int64m1_t b) {1198  return a / b;1199}1200 1201// CHECK-LABEL: @div_u8(1202// CHECK-NEXT:  entry:1203// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1204// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1205// CHECK-NEXT:    [[DIV:%.*]] = udiv <32 x i8> [[A]], [[B]]1206// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[DIV]], i64 0)1207// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1208//1209fixed_uint8m1_t div_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {1210  return a / b;1211}1212 1213// CHECK-LABEL: @div_u16(1214// CHECK-NEXT:  entry:1215// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1216// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1217// CHECK-NEXT:    [[DIV:%.*]] = udiv <16 x i16> [[A]], [[B]]1218// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[DIV]], i64 0)1219// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1220//1221fixed_uint16m1_t div_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {1222  return a / b;1223}1224 1225// CHECK-LABEL: @div_u32(1226// CHECK-NEXT:  entry:1227// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1228// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1229// CHECK-NEXT:    [[DIV:%.*]] = udiv <8 x i32> [[A]], [[B]]1230// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[DIV]], i64 0)1231// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1232//1233fixed_uint32m1_t div_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {1234  return a / b;1235}1236 1237// CHECK-LABEL: @div_u64(1238// CHECK-NEXT:  entry:1239// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1240// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1241// CHECK-NEXT:    [[DIV:%.*]] = udiv <4 x i64> [[A]], [[B]]1242// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[DIV]], i64 0)1243// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1244//1245fixed_uint64m1_t div_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {1246  return a / b;1247}1248 1249// CHECK-LABEL: @div_f32(1250// CHECK-NEXT:  entry:1251// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)1252// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)1253// CHECK-NEXT:    [[DIV:%.*]] = fdiv <8 x float> [[A]], [[B]]1254// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[DIV]], i64 0)1255// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]1256//1257fixed_float32m1_t div_f32(fixed_float32m1_t a, fixed_float32m1_t b) {1258  return a / b;1259}1260 1261// CHECK-LABEL: @div_f64(1262// CHECK-NEXT:  entry:1263// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)1264// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)1265// CHECK-NEXT:    [[DIV:%.*]] = fdiv <4 x double> [[A]], [[B]]1266// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[DIV]], i64 0)1267// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]1268//1269fixed_float64m1_t div_f64(fixed_float64m1_t a, fixed_float64m1_t b) {1270  return a / b;1271}1272 1273// CHECK-LABEL: @div_inplace_i8(1274// CHECK-NEXT:  entry:1275// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1276// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1277// CHECK-NEXT:    [[DIV:%.*]] = sdiv <32 x i8> [[A]], [[B]]1278// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[DIV]], i64 0)1279// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1280//1281fixed_int8m1_t div_inplace_i8(fixed_int8m1_t a, fixed_int8m1_t b) {1282  return a / b;1283}1284 1285// CHECK-LABEL: @div_inplace_i16(1286// CHECK-NEXT:  entry:1287// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1288// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1289// CHECK-NEXT:    [[DIV:%.*]] = sdiv <16 x i16> [[A]], [[B]]1290// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[DIV]], i64 0)1291// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1292//1293fixed_int16m1_t div_inplace_i16(fixed_int16m1_t a, fixed_int16m1_t b) {1294  return a / b;1295}1296 1297// CHECK-LABEL: @div_inplace_i32(1298// CHECK-NEXT:  entry:1299// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1300// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1301// CHECK-NEXT:    [[DIV:%.*]] = sdiv <8 x i32> [[A]], [[B]]1302// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[DIV]], i64 0)1303// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1304//1305fixed_int32m1_t div_inplace_i32(fixed_int32m1_t a, fixed_int32m1_t b) {1306  return a / b;1307}1308 1309// CHECK-LABEL: @div_inplace_i64(1310// CHECK-NEXT:  entry:1311// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1312// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1313// CHECK-NEXT:    [[DIV:%.*]] = sdiv <4 x i64> [[A]], [[B]]1314// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[DIV]], i64 0)1315// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1316//1317fixed_int64m1_t div_inplace_i64(fixed_int64m1_t a, fixed_int64m1_t b) {1318  return a / b;1319}1320 1321// CHECK-LABEL: @div_inplace_u8(1322// CHECK-NEXT:  entry:1323// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1324// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1325// CHECK-NEXT:    [[DIV:%.*]] = udiv <32 x i8> [[A]], [[B]]1326// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[DIV]], i64 0)1327// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1328//1329fixed_uint8m1_t div_inplace_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {1330  return a / b;1331}1332 1333// CHECK-LABEL: @div_inplace_u16(1334// CHECK-NEXT:  entry:1335// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1336// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1337// CHECK-NEXT:    [[DIV:%.*]] = udiv <16 x i16> [[A]], [[B]]1338// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[DIV]], i64 0)1339// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1340//1341fixed_uint16m1_t div_inplace_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {1342  return a / b;1343}1344 1345// CHECK-LABEL: @div_inplace_u32(1346// CHECK-NEXT:  entry:1347// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1348// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1349// CHECK-NEXT:    [[DIV:%.*]] = udiv <8 x i32> [[A]], [[B]]1350// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[DIV]], i64 0)1351// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1352//1353fixed_uint32m1_t div_inplace_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {1354  return a / b;1355}1356 1357// CHECK-LABEL: @div_inplace_u64(1358// CHECK-NEXT:  entry:1359// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1360// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1361// CHECK-NEXT:    [[DIV:%.*]] = udiv <4 x i64> [[A]], [[B]]1362// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[DIV]], i64 0)1363// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1364//1365fixed_uint64m1_t div_inplace_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {1366  return a / b;1367}1368 1369// CHECK-LABEL: @div_inplace_f32(1370// CHECK-NEXT:  entry:1371// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)1372// CHECK-NEXT:    [[B:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[B_COERCE:%.*]], i64 0)1373// CHECK-NEXT:    [[DIV:%.*]] = fdiv <8 x float> [[A]], [[B]]1374// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[DIV]], i64 0)1375// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]1376//1377fixed_float32m1_t div_inplace_f32(fixed_float32m1_t a, fixed_float32m1_t b) {1378  return a / b;1379}1380 1381// CHECK-LABEL: @div_inplace_f64(1382// CHECK-NEXT:  entry:1383// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)1384// CHECK-NEXT:    [[B:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[B_COERCE:%.*]], i64 0)1385// CHECK-NEXT:    [[DIV:%.*]] = fdiv <4 x double> [[A]], [[B]]1386// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[DIV]], i64 0)1387// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]1388//1389fixed_float64m1_t div_inplace_f64(fixed_float64m1_t a, fixed_float64m1_t b) {1390  return a / b;1391}1392 1393// CHECK-LABEL: @div_scalar_i8(1394// CHECK-NEXT:  entry:1395// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1396// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 01397// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer1398// CHECK-NEXT:    [[DIV:%.*]] = sdiv <32 x i8> [[A]], [[SPLAT_SPLAT]]1399// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[DIV]], i64 0)1400// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1401//1402fixed_int8m1_t div_scalar_i8(fixed_int8m1_t a, int8_t b) {1403  return a / b;1404}1405 1406// CHECK-LABEL: @div_scalar_i16(1407// CHECK-NEXT:  entry:1408// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1409// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 01410// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer1411// CHECK-NEXT:    [[DIV:%.*]] = sdiv <16 x i16> [[A]], [[SPLAT_SPLAT]]1412// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[DIV]], i64 0)1413// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1414//1415fixed_int16m1_t div_scalar_i16(fixed_int16m1_t a, int16_t b) {1416  return a / b;1417}1418 1419// CHECK-LABEL: @div_scalar_i32(1420// CHECK-NEXT:  entry:1421// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1422// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 01423// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer1424// CHECK-NEXT:    [[DIV:%.*]] = sdiv <8 x i32> [[A]], [[SPLAT_SPLAT]]1425// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[DIV]], i64 0)1426// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1427//1428fixed_int32m1_t div_scalar_i32(fixed_int32m1_t a, int32_t b) {1429  return a / b;1430}1431 1432// CHECK-LABEL: @div_scalar_i64(1433// CHECK-NEXT:  entry:1434// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1435// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 01436// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer1437// CHECK-NEXT:    [[DIV:%.*]] = sdiv <4 x i64> [[A]], [[SPLAT_SPLAT]]1438// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[DIV]], i64 0)1439// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1440//1441fixed_int64m1_t div_scalar_i64(fixed_int64m1_t a, int64_t b) {1442  return a / b;1443}1444 1445// CHECK-LABEL: @div_scalar_u8(1446// CHECK-NEXT:  entry:1447// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1448// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 01449// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer1450// CHECK-NEXT:    [[DIV:%.*]] = udiv <32 x i8> [[A]], [[SPLAT_SPLAT]]1451// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[DIV]], i64 0)1452// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1453//1454fixed_uint8m1_t div_scalar_u8(fixed_uint8m1_t a, uint8_t b) {1455  return a / b;1456}1457 1458// CHECK-LABEL: @div_scalar_u16(1459// CHECK-NEXT:  entry:1460// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1461// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 01462// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer1463// CHECK-NEXT:    [[DIV:%.*]] = udiv <16 x i16> [[A]], [[SPLAT_SPLAT]]1464// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[DIV]], i64 0)1465// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1466//1467fixed_uint16m1_t div_scalar_u16(fixed_uint16m1_t a, uint16_t b) {1468  return a / b;1469}1470 1471// CHECK-LABEL: @div_scalar_u32(1472// CHECK-NEXT:  entry:1473// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1474// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 01475// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer1476// CHECK-NEXT:    [[DIV:%.*]] = udiv <8 x i32> [[A]], [[SPLAT_SPLAT]]1477// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[DIV]], i64 0)1478// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1479//1480fixed_uint32m1_t div_scalar_u32(fixed_uint32m1_t a, uint32_t b) {1481  return a / b;1482}1483 1484// CHECK-LABEL: @div_scalar_u64(1485// CHECK-NEXT:  entry:1486// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1487// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 01488// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer1489// CHECK-NEXT:    [[DIV:%.*]] = udiv <4 x i64> [[A]], [[SPLAT_SPLAT]]1490// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[DIV]], i64 0)1491// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1492//1493fixed_uint64m1_t div_scalar_u64(fixed_uint64m1_t a, uint64_t b) {1494  return a / b;1495}1496 1497// CHECK-LABEL: @div_scalar_f32(1498// CHECK-NEXT:  entry:1499// CHECK-NEXT:    [[A:%.*]] = call <8 x float> @llvm.vector.extract.v8f32.nxv2f32(<vscale x 2 x float> [[A_COERCE:%.*]], i64 0)1500// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x float> poison, float [[B:%.*]], i64 01501// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x float> [[SPLAT_SPLATINSERT]], <8 x float> poison, <8 x i32> zeroinitializer1502// CHECK-NEXT:    [[DIV:%.*]] = fdiv <8 x float> [[A]], [[SPLAT_SPLAT]]1503// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x float> @llvm.vector.insert.nxv2f32.v8f32(<vscale x 2 x float> poison, <8 x float> [[DIV]], i64 0)1504// CHECK-NEXT:    ret <vscale x 2 x float> [[CAST_SCALABLE]]1505//1506fixed_float32m1_t div_scalar_f32(fixed_float32m1_t a, float b) {1507  return a / b;1508}1509 1510// CHECK-LABEL: @div_scalar_f64(1511// CHECK-NEXT:  entry:1512// CHECK-NEXT:    [[A:%.*]] = call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[A_COERCE:%.*]], i64 0)1513// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x double> poison, double [[B:%.*]], i64 01514// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x double> [[SPLAT_SPLATINSERT]], <4 x double> poison, <4 x i32> zeroinitializer1515// CHECK-NEXT:    [[DIV:%.*]] = fdiv <4 x double> [[A]], [[SPLAT_SPLAT]]1516// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> poison, <4 x double> [[DIV]], i64 0)1517// CHECK-NEXT:    ret <vscale x 1 x double> [[CAST_SCALABLE]]1518//1519fixed_float64m1_t div_scalar_f64(fixed_float64m1_t a, double b) {1520  return a / b;1521}1522 1523// REMAINDER1524 1525// CHECK-LABEL: @rem_i8(1526// CHECK-NEXT:  entry:1527// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1528// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1529// CHECK-NEXT:    [[REM:%.*]] = srem <32 x i8> [[A]], [[B]]1530// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[REM]], i64 0)1531// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1532//1533fixed_int8m1_t rem_i8(fixed_int8m1_t a, fixed_int8m1_t b) {1534  return a % b;1535}1536 1537// CHECK-LABEL: @rem_i16(1538// CHECK-NEXT:  entry:1539// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1540// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1541// CHECK-NEXT:    [[REM:%.*]] = srem <16 x i16> [[A]], [[B]]1542// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[REM]], i64 0)1543// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1544//1545fixed_int16m1_t rem_i16(fixed_int16m1_t a, fixed_int16m1_t b) {1546  return a % b;1547}1548 1549// CHECK-LABEL: @rem_i32(1550// CHECK-NEXT:  entry:1551// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1552// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1553// CHECK-NEXT:    [[REM:%.*]] = srem <8 x i32> [[A]], [[B]]1554// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[REM]], i64 0)1555// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1556//1557fixed_int32m1_t rem_i32(fixed_int32m1_t a, fixed_int32m1_t b) {1558  return a % b;1559}1560 1561// CHECK-LABEL: @rem_i64(1562// CHECK-NEXT:  entry:1563// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1564// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1565// CHECK-NEXT:    [[REM:%.*]] = srem <4 x i64> [[A]], [[B]]1566// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[REM]], i64 0)1567// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1568//1569fixed_int64m1_t rem_i64(fixed_int64m1_t a, fixed_int64m1_t b) {1570  return a % b;1571}1572 1573// CHECK-LABEL: @rem_u8(1574// CHECK-NEXT:  entry:1575// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1576// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1577// CHECK-NEXT:    [[REM:%.*]] = urem <32 x i8> [[A]], [[B]]1578// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[REM]], i64 0)1579// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1580//1581fixed_uint8m1_t rem_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {1582  return a % b;1583}1584 1585// CHECK-LABEL: @rem_u16(1586// CHECK-NEXT:  entry:1587// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1588// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1589// CHECK-NEXT:    [[REM:%.*]] = urem <16 x i16> [[A]], [[B]]1590// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[REM]], i64 0)1591// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1592//1593fixed_uint16m1_t rem_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {1594  return a % b;1595}1596 1597// CHECK-LABEL: @rem_u32(1598// CHECK-NEXT:  entry:1599// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1600// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1601// CHECK-NEXT:    [[REM:%.*]] = urem <8 x i32> [[A]], [[B]]1602// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[REM]], i64 0)1603// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1604//1605fixed_uint32m1_t rem_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {1606  return a % b;1607}1608 1609// CHECK-LABEL: @rem_u64(1610// CHECK-NEXT:  entry:1611// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1612// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1613// CHECK-NEXT:    [[REM:%.*]] = urem <4 x i64> [[A]], [[B]]1614// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[REM]], i64 0)1615// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1616//1617fixed_uint64m1_t rem_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {1618  return a % b;1619}1620 1621// CHECK-LABEL: @rem_inplace_i8(1622// CHECK-NEXT:  entry:1623// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1624// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1625// CHECK-NEXT:    [[REM:%.*]] = srem <32 x i8> [[A]], [[B]]1626// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[REM]], i64 0)1627// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1628//1629fixed_int8m1_t rem_inplace_i8(fixed_int8m1_t a, fixed_int8m1_t b) {1630  return a % b;1631}1632 1633// CHECK-LABEL: @rem_inplace_i16(1634// CHECK-NEXT:  entry:1635// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1636// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1637// CHECK-NEXT:    [[REM:%.*]] = srem <16 x i16> [[A]], [[B]]1638// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[REM]], i64 0)1639// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1640//1641fixed_int16m1_t rem_inplace_i16(fixed_int16m1_t a, fixed_int16m1_t b) {1642  return a % b;1643}1644 1645// CHECK-LABEL: @rem_inplace_i32(1646// CHECK-NEXT:  entry:1647// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1648// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1649// CHECK-NEXT:    [[REM:%.*]] = srem <8 x i32> [[A]], [[B]]1650// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[REM]], i64 0)1651// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1652//1653fixed_int32m1_t rem_inplace_i32(fixed_int32m1_t a, fixed_int32m1_t b) {1654  return a % b;1655}1656 1657// CHECK-LABEL: @rem_inplace_i64(1658// CHECK-NEXT:  entry:1659// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1660// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1661// CHECK-NEXT:    [[REM:%.*]] = srem <4 x i64> [[A]], [[B]]1662// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[REM]], i64 0)1663// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1664//1665fixed_int64m1_t rem_inplace_i64(fixed_int64m1_t a, fixed_int64m1_t b) {1666  return a % b;1667}1668 1669// CHECK-LABEL: @rem_inplace_u8(1670// CHECK-NEXT:  entry:1671// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1672// CHECK-NEXT:    [[B:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[B_COERCE:%.*]], i64 0)1673// CHECK-NEXT:    [[REM:%.*]] = urem <32 x i8> [[A]], [[B]]1674// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[REM]], i64 0)1675// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1676//1677fixed_uint8m1_t rem_inplace_u8(fixed_uint8m1_t a, fixed_uint8m1_t b) {1678  return a % b;1679}1680 1681// CHECK-LABEL: @rem_inplace_u16(1682// CHECK-NEXT:  entry:1683// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1684// CHECK-NEXT:    [[B:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[B_COERCE:%.*]], i64 0)1685// CHECK-NEXT:    [[REM:%.*]] = urem <16 x i16> [[A]], [[B]]1686// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[REM]], i64 0)1687// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1688//1689fixed_uint16m1_t rem_inplace_u16(fixed_uint16m1_t a, fixed_uint16m1_t b) {1690  return a % b;1691}1692 1693// CHECK-LABEL: @rem_inplace_u32(1694// CHECK-NEXT:  entry:1695// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1696// CHECK-NEXT:    [[B:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[B_COERCE:%.*]], i64 0)1697// CHECK-NEXT:    [[REM:%.*]] = urem <8 x i32> [[A]], [[B]]1698// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[REM]], i64 0)1699// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1700//1701fixed_uint32m1_t rem_inplace_u32(fixed_uint32m1_t a, fixed_uint32m1_t b) {1702  return a % b;1703}1704 1705// CHECK-LABEL: @rem_inplace_u64(1706// CHECK-NEXT:  entry:1707// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1708// CHECK-NEXT:    [[B:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[B_COERCE:%.*]], i64 0)1709// CHECK-NEXT:    [[REM:%.*]] = urem <4 x i64> [[A]], [[B]]1710// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[REM]], i64 0)1711// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1712//1713fixed_uint64m1_t rem_inplace_u64(fixed_uint64m1_t a, fixed_uint64m1_t b) {1714  return a % b;1715}1716 1717// CHECK-LABEL: @rem_scalar_i8(1718// CHECK-NEXT:  entry:1719// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1720// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 01721// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer1722// CHECK-NEXT:    [[REM:%.*]] = srem <32 x i8> [[A]], [[SPLAT_SPLAT]]1723// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[REM]], i64 0)1724// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1725//1726fixed_int8m1_t rem_scalar_i8(fixed_int8m1_t a, int8_t b) {1727  return a % b;1728}1729 1730// CHECK-LABEL: @rem_scalar_i16(1731// CHECK-NEXT:  entry:1732// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1733// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 01734// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer1735// CHECK-NEXT:    [[REM:%.*]] = srem <16 x i16> [[A]], [[SPLAT_SPLAT]]1736// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[REM]], i64 0)1737// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1738//1739fixed_int16m1_t rem_scalar_i16(fixed_int16m1_t a, int16_t b) {1740  return a % b;1741}1742 1743// CHECK-LABEL: @rem_scalar_i32(1744// CHECK-NEXT:  entry:1745// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1746// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 01747// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer1748// CHECK-NEXT:    [[REM:%.*]] = srem <8 x i32> [[A]], [[SPLAT_SPLAT]]1749// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[REM]], i64 0)1750// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1751//1752fixed_int32m1_t rem_scalar_i32(fixed_int32m1_t a, int32_t b) {1753  return a % b;1754}1755 1756// CHECK-LABEL: @rem_scalar_i64(1757// CHECK-NEXT:  entry:1758// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1759// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 01760// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer1761// CHECK-NEXT:    [[REM:%.*]] = srem <4 x i64> [[A]], [[SPLAT_SPLAT]]1762// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[REM]], i64 0)1763// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1764//1765fixed_int64m1_t rem_scalar_i64(fixed_int64m1_t a, int64_t b) {1766  return a % b;1767}1768 1769// CHECK-LABEL: @rem_scalar_u8(1770// CHECK-NEXT:  entry:1771// CHECK-NEXT:    [[A:%.*]] = call <32 x i8> @llvm.vector.extract.v32i8.nxv8i8(<vscale x 8 x i8> [[A_COERCE:%.*]], i64 0)1772// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <32 x i8> poison, i8 [[B:%.*]], i64 01773// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <32 x i8> [[SPLAT_SPLATINSERT]], <32 x i8> poison, <32 x i32> zeroinitializer1774// CHECK-NEXT:    [[REM:%.*]] = urem <32 x i8> [[A]], [[SPLAT_SPLAT]]1775// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 8 x i8> @llvm.vector.insert.nxv8i8.v32i8(<vscale x 8 x i8> poison, <32 x i8> [[REM]], i64 0)1776// CHECK-NEXT:    ret <vscale x 8 x i8> [[CAST_SCALABLE]]1777//1778fixed_uint8m1_t rem_scalar_u8(fixed_uint8m1_t a, uint8_t b) {1779  return a % b;1780}1781 1782// CHECK-LABEL: @rem_scalar_u16(1783// CHECK-NEXT:  entry:1784// CHECK-NEXT:    [[A:%.*]] = call <16 x i16> @llvm.vector.extract.v16i16.nxv4i16(<vscale x 4 x i16> [[A_COERCE:%.*]], i64 0)1785// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <16 x i16> poison, i16 [[B:%.*]], i64 01786// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <16 x i16> [[SPLAT_SPLATINSERT]], <16 x i16> poison, <16 x i32> zeroinitializer1787// CHECK-NEXT:    [[REM:%.*]] = urem <16 x i16> [[A]], [[SPLAT_SPLAT]]1788// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i16> @llvm.vector.insert.nxv4i16.v16i16(<vscale x 4 x i16> poison, <16 x i16> [[REM]], i64 0)1789// CHECK-NEXT:    ret <vscale x 4 x i16> [[CAST_SCALABLE]]1790//1791fixed_uint16m1_t rem_scalar_u16(fixed_uint16m1_t a, uint16_t b) {1792  return a % b;1793}1794 1795// CHECK-LABEL: @rem_scalar_u32(1796// CHECK-NEXT:  entry:1797// CHECK-NEXT:    [[A:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[A_COERCE:%.*]], i64 0)1798// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <8 x i32> poison, i32 [[B:%.*]], i64 01799// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <8 x i32> [[SPLAT_SPLATINSERT]], <8 x i32> poison, <8 x i32> zeroinitializer1800// CHECK-NEXT:    [[REM:%.*]] = urem <8 x i32> [[A]], [[SPLAT_SPLAT]]1801// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> poison, <8 x i32> [[REM]], i64 0)1802// CHECK-NEXT:    ret <vscale x 2 x i32> [[CAST_SCALABLE]]1803//1804fixed_uint32m1_t rem_scalar_u32(fixed_uint32m1_t a, uint32_t b) {1805  return a % b;1806}1807 1808// CHECK-LABEL: @rem_scalar_u64(1809// CHECK-NEXT:  entry:1810// CHECK-NEXT:    [[A:%.*]] = call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[A_COERCE:%.*]], i64 0)1811// CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[B:%.*]], i64 01812// CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <4 x i64> [[SPLAT_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer1813// CHECK-NEXT:    [[REM:%.*]] = urem <4 x i64> [[A]], [[SPLAT_SPLAT]]1814// CHECK-NEXT:    [[CAST_SCALABLE:%.*]] = call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> poison, <4 x i64> [[REM]], i64 0)1815// CHECK-NEXT:    ret <vscale x 1 x i64> [[CAST_SCALABLE]]1816//1817fixed_uint64m1_t rem_scalar_u64(fixed_uint64m1_t a, uint64_t b) {1818  return a % b;1819}1820