206 lines
8.0 KiB
LLVM
206 lines
8.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve.fp -verify-machineinstrs -o - %s | FileCheck %s
|
|
|
|
declare <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32)
|
|
declare <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32)
|
|
|
|
declare <8 x half> @llvm.arm.mve.vcmlaq.v8f16(i32, <8 x half>, <8 x half>, <8 x half>)
|
|
declare <4 x float> @llvm.arm.mve.vcmlaq.v4f32(i32, <4 x float>, <4 x float>, <4 x float>)
|
|
|
|
declare <8 x half> @llvm.arm.mve.vcmlaq.predicated.v8f16.v8i1(i32, <8 x half>, <8 x half>, <8 x half>, <8 x i1>)
|
|
declare <4 x float> @llvm.arm.mve.vcmlaq.predicated.v4f32.v4i1(i32, <4 x float>, <4 x float>, <4 x float>, <4 x i1>)
|
|
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f16 q0, q1, q2, #0
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <8 x half> @llvm.arm.mve.vcmlaq.v8f16(i32 0, <8 x half> %a, <8 x half> %b, <8 x half> %c)
|
|
ret <8 x half> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f32 q0, q1, q2, #0
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <4 x float> @llvm.arm.mve.vcmlaq.v4f32(i32 0, <4 x float> %a, <4 x float> %b, <4 x float> %c)
|
|
ret <4 x float> %0
|
|
}
|
|
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_rot90_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_rot90_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f16 q0, q1, q2, #90
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <8 x half> @llvm.arm.mve.vcmlaq.v8f16(i32 1, <8 x half> %a, <8 x half> %b, <8 x half> %c)
|
|
ret <8 x half> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_rot90_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_rot90_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f32 q0, q1, q2, #90
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <4 x float> @llvm.arm.mve.vcmlaq.v4f32(i32 1, <4 x float> %a, <4 x float> %b, <4 x float> %c)
|
|
ret <4 x float> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_rot180_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_rot180_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f16 q0, q1, q2, #180
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <8 x half> @llvm.arm.mve.vcmlaq.v8f16(i32 2, <8 x half> %a, <8 x half> %b, <8 x half> %c)
|
|
ret <8 x half> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_rot180_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_rot180_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f32 q0, q1, q2, #180
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <4 x float> @llvm.arm.mve.vcmlaq.v4f32(i32 2, <4 x float> %a, <4 x float> %b, <4 x float> %c)
|
|
ret <4 x float> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_rot270_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_rot270_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f16 q0, q1, q2, #270
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <8 x half> @llvm.arm.mve.vcmlaq.v8f16(i32 3, <8 x half> %a, <8 x half> %b, <8 x half> %c)
|
|
ret <8 x half> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_rot270_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
|
|
; CHECK-LABEL: test_vcmlaq_rot270_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vcmla.f32 q0, q1, q2, #270
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = call <4 x float> @llvm.arm.mve.vcmlaq.v4f32(i32 3, <4 x float> %a, <4 x float> %b, <4 x float> %c)
|
|
ret <4 x float> %0
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_m_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_m_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f16 q0, q1, q2, #0
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
|
|
%2 = call <8 x half> @llvm.arm.mve.vcmlaq.predicated.v8f16.v8i1(i32 0, <8 x half> %a, <8 x half> %b, <8 x half> %c, <8 x i1> %1)
|
|
ret <8 x half> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_m_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_m_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f32 q0, q1, q2, #0
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
|
|
%2 = call <4 x float> @llvm.arm.mve.vcmlaq.predicated.v4f32.v4i1(i32 0, <4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x i1> %1)
|
|
ret <4 x float> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_rot90_m_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_rot90_m_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f16 q0, q1, q2, #90
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
|
|
%2 = call <8 x half> @llvm.arm.mve.vcmlaq.predicated.v8f16.v8i1(i32 1, <8 x half> %a, <8 x half> %b, <8 x half> %c, <8 x i1> %1)
|
|
ret <8 x half> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_rot90_m_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_rot90_m_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f32 q0, q1, q2, #90
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
|
|
%2 = call <4 x float> @llvm.arm.mve.vcmlaq.predicated.v4f32.v4i1(i32 1, <4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x i1> %1)
|
|
ret <4 x float> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_rot180_m_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_rot180_m_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f16 q0, q1, q2, #180
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
|
|
%2 = call <8 x half> @llvm.arm.mve.vcmlaq.predicated.v8f16.v8i1(i32 2, <8 x half> %a, <8 x half> %b, <8 x half> %c, <8 x i1> %1)
|
|
ret <8 x half> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_rot180_m_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_rot180_m_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f32 q0, q1, q2, #180
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
|
|
%2 = call <4 x float> @llvm.arm.mve.vcmlaq.predicated.v4f32.v4i1(i32 2, <4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x i1> %1)
|
|
ret <4 x float> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x half> @test_vcmlaq_rot270_m_f16(<8 x half> %a, <8 x half> %b, <8 x half> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_rot270_m_f16:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f16 q0, q1, q2, #270
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <8 x i1> @llvm.arm.mve.pred.i2v.v8i1(i32 %0)
|
|
%2 = call <8 x half> @llvm.arm.mve.vcmlaq.predicated.v8f16.v8i1(i32 3, <8 x half> %a, <8 x half> %b, <8 x half> %c, <8 x i1> %1)
|
|
ret <8 x half> %2
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x float> @test_vcmlaq_rot270_m_f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, i16 zeroext %p) {
|
|
; CHECK-LABEL: test_vcmlaq_rot270_m_f32:
|
|
; CHECK: @ %bb.0: @ %entry
|
|
; CHECK-NEXT: vmsr p0, r0
|
|
; CHECK-NEXT: vpst
|
|
; CHECK-NEXT: vcmlat.f32 q0, q1, q2, #270
|
|
; CHECK-NEXT: bx lr
|
|
entry:
|
|
%0 = zext i16 %p to i32
|
|
%1 = call <4 x i1> @llvm.arm.mve.pred.i2v.v4i1(i32 %0)
|
|
%2 = call <4 x float> @llvm.arm.mve.vcmlaq.predicated.v4f32.v4i1(i32 3, <4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x i1> %1)
|
|
ret <4 x float> %2
|
|
}
|