; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -march=hexagon < %s | FileCheck %s

define <4 x i8> @f0(<4 x i8> %a0, <4 x i8> %a1) #0 {
; CHECK-LABEL: f0:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r2 = r1
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = combine(#0,#0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5:4 = vmpybu(r0,r1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     p1 = vcmpb.gt(r1:0,#-1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     p0 = vcmpb.gt(r3:2,#-1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r3:2 = vmux(p1,r7:6,r3:2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vmux(p0,r7:6,r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r4 = vtrunohb(r5:4)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = vmpybu(r0,r0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vaddub(r3:2,r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5 = vtrunohb(r7:6)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vsubub(r5:4,r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = sext <4 x i8> %a0 to <4 x i16>
  %v1 = sext <4 x i8> %a1 to <4 x i16>
  %v2 = mul <4 x i16> %v0, %v1
  %v3 = lshr <4 x i16> %v2, <i16 8, i16 8, i16 8, i16 8>
  %v4 = trunc <4 x i16> %v3 to <4 x i8>
  ret <4 x i8> %v4
}

define <4 x i8> @f1(<4 x i8> %a0, <4 x i8> %a1) #0 {
; CHECK-LABEL: f1:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vmpybu(r0,r1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = vtrunohb(r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = zext <4 x i8> %a0 to <4 x i16>
  %v1 = zext <4 x i8> %a1 to <4 x i16>
  %v2 = mul <4 x i16> %v0, %v1
  %v3 = lshr <4 x i16> %v2, <i16 8, i16 8, i16 8, i16 8>
  %v4 = trunc <4 x i16> %v3 to <4 x i8>
  ret <4 x i8> %v4
}

define <8 x i8> @f2(<8 x i8> %a0, <8 x i8> %a1) #0 {
; CHECK-LABEL: f2:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = combine(#0,#0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     p0 = vcmpb.gt(r3:2,#-1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5:4 = vmpybu(r0,r2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r9:8 = vmux(p0,r7:6,r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     p0 = vcmpb.gt(r1:0,#-1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vmpybu(r1,r3)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = vmux(p0,r7:6,r3:2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r4 = vtrunohb(r5:4)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r3:2 = vaddub(r7:6,r9:8)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5 = vtrunohb(r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vsubub(r5:4,r3:2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = sext <8 x i8> %a0 to <8 x i16>
  %v1 = sext <8 x i8> %a1 to <8 x i16>
  %v2 = mul <8 x i16> %v0, %v1
  %v3 = lshr <8 x i16> %v2, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
  %v4 = trunc <8 x i16> %v3 to <8 x i8>
  ret <8 x i8> %v4
}

define <8 x i8> @f3(<8 x i8> %a0, <8 x i8> %a1) #0 {
; CHECK-LABEL: f3:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r5:4 = vmpybu(r0,r2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = vmpybu(r1,r3)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = vtrunohb(r5:4)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1 = vtrunohb(r7:6)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = zext <8 x i8> %a0 to <8 x i16>
  %v1 = zext <8 x i8> %a1 to <8 x i16>
  %v2 = mul <8 x i16> %v0, %v1
  %v3 = lshr <8 x i16> %v2, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8>
  %v4 = trunc <8 x i16> %v3 to <8 x i8>
  ret <8 x i8> %v4
}

define <2 x i16> @f4(<2 x i16> %a0, <2 x i16> %a1) #0 {
; CHECK-LABEL: f4:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vmpyh(r0,r1):sat
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = combine(r1.h,r0.h)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = sext <2 x i16> %a0 to <2 x i32>
  %v1 = sext <2 x i16> %a1 to <2 x i32>
  %v2 = mul <2 x i32> %v0, %v1
  %v3 = lshr <2 x i32> %v2, <i32 16, i32 16>
  %v4 = trunc <2 x i32> %v3 to <2 x i16>
  ret <2 x i16> %v4
}

define <2 x i16> @f5(<2 x i16> %a0, <2 x i16> %a1) #0 {
; CHECK-LABEL: f5:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r3:2 = combine(r0,r1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vasrh(r3:2,#15)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5:4 = vmpyh(r3,r2):sat
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = and(r3,r0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1 = and(r2,r1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r4 = combine(r5.h,r4.h)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = vaddh(r0,r1)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = vaddh(r4,r0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = zext <2 x i16> %a0 to <2 x i32>
  %v1 = zext <2 x i16> %a1 to <2 x i32>
  %v2 = mul <2 x i32> %v0, %v1
  %v3 = lshr <2 x i32> %v2, <i32 16, i32 16>
  %v4 = trunc <2 x i32> %v3 to <2 x i16>
  ret <2 x i16> %v4
}

define <4 x i16> @f6(<4 x i16> %a0, <4 x i16> %a1) #0 {
; CHECK-LABEL: f6:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r5:4 = vmpyh(r0,r2):sat
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = vmpyh(r1,r3):sat
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = combine(r5.h,r4.h)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1 = combine(r7.h,r6.h)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = sext <4 x i16> %a0 to <4 x i32>
  %v1 = sext <4 x i16> %a1 to <4 x i32>
  %v2 = mul <4 x i32> %v0, %v1
  %v3 = lshr <4 x i32> %v2, <i32 16, i32 16, i32 16, i32 16>
  %v4 = trunc <4 x i32> %v3 to <4 x i16>
  ret <4 x i16> %v4
}

define <4 x i16> @f7(<4 x i16> %a0, <4 x i16> %a1) #0 {
; CHECK-LABEL: f7:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = vasrh(r1:0,#15)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r9:8 = vasrh(r3:2,#15)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5:4 = vmpyh(r0,r2):sat
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r7:6 = and(r3:2,r7:6)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r3:2 = vmpyh(r1,r3):sat
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = and(r1:0,r9:8)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r4 = combine(r5.h,r4.h)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r5 = combine(r3.h,r2.h)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vaddh(r1:0,r7:6)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1:0 = vaddh(r5:4,r1:0)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = zext <4 x i16> %a0 to <4 x i32>
  %v1 = zext <4 x i16> %a1 to <4 x i32>
  %v2 = mul <4 x i32> %v0, %v1
  %v3 = lshr <4 x i32> %v2, <i32 16, i32 16, i32 16, i32 16>
  %v4 = trunc <4 x i32> %v3 to <4 x i16>
  ret <4 x i16> %v4
}

define <2 x i32> @f8(<2 x i32> %a0, <2 x i32> %a1) #0 {
; CHECK-LABEL: f8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = mpy(r0,r2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1 = mpy(r1,r3)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = sext <2 x i32> %a0 to <2 x i64>
  %v1 = sext <2 x i32> %a1 to <2 x i64>
  %v2 = mul <2 x i64> %v0, %v1
  %v3 = lshr <2 x i64> %v2, <i64 32, i64 32>
  %v4 = trunc <2 x i64> %v3 to <2 x i32>
  ret <2 x i32> %v4
}

define <2 x i32> @f9(<2 x i32> %a0, <2 x i32> %a1) #0 {
; CHECK-LABEL: f9:
; CHECK:       // %bb.0:
; CHECK-NEXT:    {
; CHECK-NEXT:     r0 = mpyu(r0,r2)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     r1 = mpyu(r1,r3)
; CHECK-NEXT:    }
; CHECK-NEXT:    {
; CHECK-NEXT:     jumpr r31
; CHECK-NEXT:    }
  %v0 = zext <2 x i32> %a0 to <2 x i64>
  %v1 = zext <2 x i32> %a1 to <2 x i64>
  %v2 = mul <2 x i64> %v0, %v1
  %v3 = lshr <2 x i64> %v2, <i64 32, i64 32>
  %v4 = trunc <2 x i64> %v3 to <2 x i32>
  ret <2 x i32> %v4
}

attributes #0 = { nounwind memory(none) "target-features"="-packets" }