llvm/test/CodeGen/X86/combine-rndscale.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=sandybridge | FileCheck %s --check-prefixes=AVX,AVX1OR2,AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=AVX,AVX1OR2,AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=AVX,AVX512

define <4 x double> @concat_roundpd_v4f64_v2f64(<2 x double> %a0, <2 x double> %a1) {
; AVX-LABEL: concat_roundpd_v4f64_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX-NEXT:    vroundpd $4, %ymm0, %ymm0
; AVX-NEXT:    retq
  %v0 = call <2 x double> @llvm.x86.sse41.round.pd(<2 x double> %a0, i32 4)
  %v1 = call <2 x double> @llvm.x86.sse41.round.pd(<2 x double> %a1, i32 4)
  %res  = shufflevector <2 x double> %v0, <2 x double> %v1, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  ret <4 x double> %res
}

define <8 x float> @concat_roundps_v8f32_v4f32(<4 x float> %a0, <4 x float> %a1) {
; AVX-LABEL: concat_roundps_v8f32_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX-NEXT:    vroundps $4, %ymm0, %ymm0
; AVX-NEXT:    retq
  %v0 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a0, i32 4)
  %v1 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a1, i32 4)
  %res  = shufflevector <4 x float> %v0, <4 x float> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  ret <8 x float> %res
}

define <8 x double> @concat_roundpd_v8f64_v2f64(<2 x double> %a0, <2 x double> %a1, <2 x double> %a2, <2 x double> %a3) {
; AVX1-LABEL: concat_roundpd_v8f64_v2f64:
; AVX1:       # %bb.0:
; AVX1-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX1-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vroundpd $4, %ymm0, %ymm0
; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm1
; AVX1-NEXT:    vroundpd $4, %ymm1, %ymm1
; AVX1-NEXT:    retq
;
; AVX2-LABEL: concat_roundpd_v8f64_v2f64:
; AVX2:       # %bb.0:
; AVX2-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT:    vroundpd $4, %ymm0, %ymm0
; AVX2-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX2-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm1
; AVX2-NEXT:    vroundpd $4, %ymm1, %ymm1
; AVX2-NEXT:    retq
;
; AVX512-LABEL: concat_roundpd_v8f64_v2f64:
; AVX512:       # %bb.0:
; AVX512-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX512-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX512-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT:    vinsertf64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512-NEXT:    vrndscalepd $4, %zmm0, %zmm0
; AVX512-NEXT:    retq
  %v0 = call <2 x double> @llvm.x86.sse41.round.pd(<2 x double> %a0, i32 4)
  %v1 = call <2 x double> @llvm.x86.sse41.round.pd(<2 x double> %a1, i32 4)
  %v2 = call <2 x double> @llvm.x86.sse41.round.pd(<2 x double> %a2, i32 4)
  %v3 = call <2 x double> @llvm.x86.sse41.round.pd(<2 x double> %a3, i32 4)
  %r01 = shufflevector <2 x double> %v0, <2 x double> %v1, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %r23 = shufflevector <2 x double> %v2, <2 x double> %v3, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %res  = shufflevector <4 x double> %r01, <4 x double> %r23, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  ret <8 x double> %res
}

define <16 x float> @concat_roundps_v16f32_v4f32(<4 x float> %a0, <4 x float> %a1, <4 x float> %a2, <4 x float> %a3) {
; AVX1-LABEL: concat_roundps_v16f32_v4f32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX1-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vroundps $4, %ymm0, %ymm0
; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm1
; AVX1-NEXT:    vroundps $4, %ymm1, %ymm1
; AVX1-NEXT:    retq
;
; AVX2-LABEL: concat_roundps_v16f32_v4f32:
; AVX2:       # %bb.0:
; AVX2-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT:    vroundps $4, %ymm0, %ymm0
; AVX2-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX2-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm1
; AVX2-NEXT:    vroundps $4, %ymm1, %ymm1
; AVX2-NEXT:    retq
;
; AVX512-LABEL: concat_roundps_v16f32_v4f32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX512-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX512-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT:    vinsertf64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512-NEXT:    vrndscaleps $4, %zmm0, %zmm0
; AVX512-NEXT:    retq
  %v0 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a0, i32 4)
  %v1 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a1, i32 4)
  %v2 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a2, i32 4)
  %v3 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a3, i32 4)
  %r01 = shufflevector <4 x float> %v0, <4 x float> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %r23 = shufflevector <4 x float> %v2, <4 x float> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %res  = shufflevector <8 x float> %r01, <8 x float> %r23, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
  ret <16 x float> %res
}

define <8 x double> @concat_roundpd_v8f64_v4f64(<4 x double> %a0, <4 x double> %a1) {
; AVX1OR2-LABEL: concat_roundpd_v8f64_v4f64:
; AVX1OR2:       # %bb.0:
; AVX1OR2-NEXT:    vroundpd $4, %ymm0, %ymm0
; AVX1OR2-NEXT:    vroundpd $4, %ymm1, %ymm1
; AVX1OR2-NEXT:    retq
;
; AVX512-LABEL: concat_roundpd_v8f64_v4f64:
; AVX512:       # %bb.0:
; AVX512-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; AVX512-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT:    vrndscalepd $4, %zmm0, %zmm0
; AVX512-NEXT:    retq
  %v0 = call <4 x double> @llvm.x86.avx.round.pd.256(<4 x double> %a0, i32 4)
  %v1 = call <4 x double> @llvm.x86.avx.round.pd.256(<4 x double> %a1, i32 4)
  %res  = shufflevector <4 x double> %v0, <4 x double> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  ret <8 x double> %res
}

define <16 x float> @concat_roundps_v16f32_v8f32(<8 x float> %a0, <8 x float> %a1) {
; AVX1OR2-LABEL: concat_roundps_v16f32_v8f32:
; AVX1OR2:       # %bb.0:
; AVX1OR2-NEXT:    vroundps $4, %ymm0, %ymm0
; AVX1OR2-NEXT:    vroundps $4, %ymm1, %ymm1
; AVX1OR2-NEXT:    retq
;
; AVX512-LABEL: concat_roundps_v16f32_v8f32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; AVX512-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT:    vrndscaleps $4, %zmm0, %zmm0
; AVX512-NEXT:    retq
  %v0 = call <8 x float> @llvm.x86.avx.round.ps.256(<8 x float> %a0, i32 4)
  %v1 = call <8 x float> @llvm.x86.avx.round.ps.256(<8 x float> %a1, i32 4)
  %res  = shufflevector <8 x float> %v0, <8 x float> %v1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
  ret <16 x float> %res
}

; negative test - rounding mode mismatch
define <8 x float> @concat_roundps_v8f32_v4f32_mismatch(<4 x float> %a0, <4 x float> %a1) {
; AVX-LABEL: concat_roundps_v8f32_v4f32_mismatch:
; AVX:       # %bb.0:
; AVX-NEXT:    vroundps $0, %xmm0, %xmm0
; AVX-NEXT:    vroundps $4, %xmm1, %xmm1
; AVX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX-NEXT:    retq
  %v0 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a0, i32 0)
  %v1 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %a1, i32 4)
  %res  = shufflevector <4 x float> %v0, <4 x float> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  ret <8 x float> %res
}

; FIXME: late expansion of roundps prevents concatenation of comparisons
define i8 @concat_roundps_cmp_v8f32_v4f32(<4 x float> %x0, <4 x float> %x1, <4 x float> %y0, <4 x float> %y1) {
; AVX1OR2-LABEL: concat_roundps_cmp_v8f32_v4f32:
; AVX1OR2:       # %bb.0:
; AVX1OR2-NEXT:    vroundps $4, %xmm0, %xmm0
; AVX1OR2-NEXT:    vroundps $4, %xmm1, %xmm1
; AVX1OR2-NEXT:    vcmpltps %xmm2, %xmm0, %xmm0
; AVX1OR2-NEXT:    vcmpltps %xmm3, %xmm1, %xmm1
; AVX1OR2-NEXT:    vpackssdw %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX1OR2-NEXT:    vpmovmskb %xmm0, %eax
; AVX1OR2-NEXT:    # kill: def $al killed $al killed $eax
; AVX1OR2-NEXT:    retq
;
; AVX512-LABEL: concat_roundps_cmp_v8f32_v4f32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX512-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX512-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT:    vroundps $4, %ymm0, %ymm0
; AVX512-NEXT:    vcmpltps %ymm2, %ymm0, %k0
; AVX512-NEXT:    kmovd %k0, %eax
; AVX512-NEXT:    # kill: def $al killed $al killed $eax
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %rnd0 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %x0, i32 4)
  %rnd1 = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %x1, i32 4)
  %cmp0 = fcmp olt <4 x float> %rnd0, %y0
  %cmp1 = fcmp olt <4 x float> %rnd1, %y1
  %concat = shufflevector <4 x i1> %cmp0, <4 x i1> %cmp1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
  %res = bitcast <8 x i1> %concat to i8
  ret i8 %res
}

define i16 @concat_roundps_cmp_v8f32(<8 x float> %x0, <8 x float> %x1, <8 x float> %y0, <8 x float> %y1) {
; AVX1OR2-LABEL: concat_roundps_cmp_v8f32:
; AVX1OR2:       # %bb.0:
; AVX1OR2-NEXT:    vroundps $4, %ymm0, %ymm0
; AVX1OR2-NEXT:    vroundps $4, %ymm1, %ymm1
; AVX1OR2-NEXT:    vcmpltps %ymm2, %ymm0, %ymm0
; AVX1OR2-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1OR2-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1OR2-NEXT:    vcmpltps %ymm3, %ymm1, %ymm1
; AVX1OR2-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1OR2-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1OR2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT:    vpmovmskb %xmm0, %eax
; AVX1OR2-NEXT:    # kill: def $ax killed $ax killed $eax
; AVX1OR2-NEXT:    vzeroupper
; AVX1OR2-NEXT:    retq
;
; AVX512-LABEL: concat_roundps_cmp_v8f32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    # kill: def $ymm2 killed $ymm2 def $zmm2
; AVX512-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; AVX512-NEXT:    vinsertf64x4 $1, %ymm3, %zmm2, %zmm2
; AVX512-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT:    vrndscaleps $4, %zmm0, %zmm0
; AVX512-NEXT:    vcmpltps %zmm2, %zmm0, %k0
; AVX512-NEXT:    kmovd %k0, %eax
; AVX512-NEXT:    # kill: def $ax killed $ax killed $eax
; AVX512-NEXT:    vzeroupper
; AVX512-NEXT:    retq
  %rnd0 = call <8 x float> @llvm.x86.avx.round.ps.256(<8 x float> %x0, i32 4)
  %rnd1 = call <8 x float> @llvm.x86.avx.round.ps.256(<8 x float> %x1, i32 4)
  %cmp0 = fcmp olt <8 x float> %rnd0, %y0
  %cmp1 = fcmp olt <8 x float> %rnd1, %y1
  %concat = shufflevector <8 x i1> %cmp0, <8 x i1> %cmp1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
  %res = bitcast <16 x i1> %concat to i16
  ret i16 %res
}