test/CodeGen/AArch64/scalable-expanding-load.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve2p2 < %s | FileCheck %s
; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve -mattr=+sme2p2 < %s | FileCheck %s
; RUN: llc -mtriple=aarch64--linux-gnu -mattr=+sve -mattr=+sme2p2 -force-streaming < %s | FileCheck %s

define <vscale x 2 x i64> @test_expandload(ptr %base, <vscale x 2 x i64> %passthru, <vscale x 2 x i1> %pred) {
; CHECK-LABEL: test_expandload:
; CHECK:       // %bb.0:
; CHECK-NEXT:    cntp x8, p0, p0.d
; CHECK-NEXT:    whilelo p1.d, xzr, x8
; CHECK-NEXT:    ld1d { z1.d }, p1/z, [x0]
; CHECK-NEXT:    expand z1.d, p0, z1.d
; CHECK-NEXT:    mov z0.d, p0/m, z1.d
; CHECK-NEXT:    ret
  %res = call <vscale x 2 x i64> @llvm.masked.expandload(ptr align 1 %base, <vscale x 2 x i1> %pred, <vscale x 2 x i64> %passthru)
  ret <vscale x 2 x i64> %res
}

define <vscale x 8 x float> @test_split_expandload(ptr %base, <vscale x 8 x float> %passthru, <vscale x 8 x i1> %pred) {
; CHECK-LABEL: test_split_expandload:
; CHECK:       // %bb.0:
; CHECK-NEXT:    punpklo p1.h, p0.b
; CHECK-NEXT:    punpkhi p0.h, p0.b
; CHECK-NEXT:    cntp x8, p1, p1.s
; CHECK-NEXT:    cntp x9, p0, p0.s
; CHECK-NEXT:    whilelo p2.s, xzr, x8
; CHECK-NEXT:    whilelo p3.s, xzr, x9
; CHECK-NEXT:    ld1w { z2.s }, p2/z, [x0]
; CHECK-NEXT:    ld1w { z3.s }, p3/z, [x0, x8, lsl #2]
; CHECK-NEXT:    expand z2.s, p1, z2.s
; CHECK-NEXT:    expand z3.s, p0, z3.s
; CHECK-NEXT:    mov z0.s, p1/m, z2.s
; CHECK-NEXT:    mov z1.s, p0/m, z3.s
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x float> @llvm.masked.expandload(ptr align 1 %base, <vscale x 8 x i1> %pred, <vscale x 8 x float> %passthru)
  ret <vscale x 8 x float> %res
}

define <vscale x 1 x i64> @test_widen_expandload(ptr %base, <vscale x 1 x i64> %passthru, <vscale x 1 x i1> %pred) {
; CHECK-LABEL: test_widen_expandload:
; CHECK:       // %bb.0:
; CHECK-NEXT:    pfalse p1.b
; CHECK-NEXT:    uzp1 p0.d, p0.d, p1.d
; CHECK-NEXT:    cntp x8, p0, p0.d
; CHECK-NEXT:    whilelo p1.d, xzr, x8
; CHECK-NEXT:    ld1d { z1.d }, p1/z, [x0]
; CHECK-NEXT:    expand z1.d, p0, z1.d
; CHECK-NEXT:    mov z0.d, p0/m, z1.d
; CHECK-NEXT:    ret
  %res = call <vscale x 1 x i64> @llvm.masked.expandload(ptr align 1 %base, <vscale x 1 x i1> %pred, <vscale x 1 x i64> %passthru)
  ret <vscale x 1 x i64> %res
}

define <vscale x 4 x i8> @test_promote_expandload(ptr %base, <vscale x 4 x i1> %pred, <vscale x 4 x i8> %passthru) {
; CHECK-LABEL: test_promote_expandload:
; CHECK:       // %bb.0:
; CHECK-NEXT:    cntp x8, p0, p0.s
; CHECK-NEXT:    whilelo p1.s, xzr, x8
; CHECK-NEXT:    ld1b { z1.s }, p1/z, [x0]
; CHECK-NEXT:    expand z1.s, p0, z1.s
; CHECK-NEXT:    mov z0.s, p0/m, z1.s
; CHECK-NEXT:    ret
  %res = call <vscale x 4 x i8> @llvm.masked.expandload(ptr align 1 %base, <vscale x 4 x i1> %pred, <vscale x 4 x i8> %passthru)
  ret <vscale x 4 x i8> %res
}

define <vscale x 8 x bfloat> @test_expandload_zero_passthru(ptr %base, <vscale x 8 x i1> %pred) {
; CHECK-LABEL: test_expandload_zero_passthru:
; CHECK:       // %bb.0:
; CHECK-NEXT:    cntp x8, p0, p0.h
; CHECK-NEXT:    whilelo p1.h, xzr, x8
; CHECK-NEXT:    ld1h { z0.h }, p1/z, [x0]
; CHECK-NEXT:    expand z0.h, p0, z0.h
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x bfloat> @llvm.masked.expandload(ptr align 1 %base, <vscale x 8 x i1> %pred, <vscale x 8 x bfloat> zeroinitializer)
  ret <vscale x 8 x bfloat> %res
}

define <vscale x 16 x i8> @test_expandload_poison_passthru(ptr %base, <vscale x 16 x i1> %pred) {
; CHECK-LABEL: test_expandload_poison_passthru:
; CHECK:       // %bb.0:
; CHECK-NEXT:    cntp x8, p0, p0.b
; CHECK-NEXT:    whilelo p1.b, xzr, x8
; CHECK-NEXT:    ld1b { z0.b }, p1/z, [x0]
; CHECK-NEXT:    expand z0.b, p0, z0.b
; CHECK-NEXT:    ret
  %res = call <vscale x 16 x i8> @llvm.masked.expandload(ptr align 1 %base, <vscale x 16 x i1> %pred, <vscale x 16 x i8> poison)
  ret <vscale x 16 x i8> %res
}