92 lines
3.3 KiB
LLVM
92 lines
3.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
|
|
; RUN: llc < %s -verify-machineinstrs -mtriple=arm64-none-linux-gnu -mattr=+neon | FileCheck %s
|
|
|
|
define <16 x i8> @div16xi8(<16 x i8> %x) {
|
|
; CHECK-LABEL: div16xi8:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: movi v1.16b, #41
|
|
; CHECK-NEXT: smull2 v2.8h, v0.16b, v1.16b
|
|
; CHECK-NEXT: smull v0.8h, v0.8b, v1.8b
|
|
; CHECK-NEXT: uzp2 v0.16b, v0.16b, v2.16b
|
|
; CHECK-NEXT: sshr v0.16b, v0.16b, #2
|
|
; CHECK-NEXT: usra v0.16b, v0.16b, #7
|
|
; CHECK-NEXT: ret
|
|
%div = sdiv <16 x i8> %x, <i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25, i8 25>
|
|
ret <16 x i8> %div
|
|
}
|
|
|
|
define <8 x i16> @div8xi16(<8 x i16> %x) {
|
|
; CHECK-LABEL: div8xi16:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #40815 // =0x9f6f
|
|
; CHECK-NEXT: dup v1.8h, w8
|
|
; CHECK-NEXT: smull2 v2.4s, v0.8h, v1.8h
|
|
; CHECK-NEXT: smull v1.4s, v0.4h, v1.4h
|
|
; CHECK-NEXT: uzp2 v1.8h, v1.8h, v2.8h
|
|
; CHECK-NEXT: add v0.8h, v1.8h, v0.8h
|
|
; CHECK-NEXT: sshr v0.8h, v0.8h, #12
|
|
; CHECK-NEXT: usra v0.8h, v0.8h, #15
|
|
; CHECK-NEXT: ret
|
|
%div = sdiv <8 x i16> %x, <i16 6577, i16 6577, i16 6577, i16 6577, i16 6577, i16 6577, i16 6577, i16 6577>
|
|
ret <8 x i16> %div
|
|
}
|
|
|
|
define <4 x i32> @div32xi4(<4 x i32> %x) {
|
|
; CHECK-LABEL: div32xi4:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #7527 // =0x1d67
|
|
; CHECK-NEXT: movk w8, #28805, lsl #16
|
|
; CHECK-NEXT: dup v1.4s, w8
|
|
; CHECK-NEXT: smull2 v2.2d, v0.4s, v1.4s
|
|
; CHECK-NEXT: smull v0.2d, v0.2s, v1.2s
|
|
; CHECK-NEXT: uzp2 v1.4s, v0.4s, v2.4s
|
|
; CHECK-NEXT: sshr v0.4s, v1.4s, #22
|
|
; CHECK-NEXT: usra v0.4s, v1.4s, #31
|
|
; CHECK-NEXT: ret
|
|
%div = sdiv <4 x i32> %x, <i32 9542677, i32 9542677, i32 9542677, i32 9542677>
|
|
ret <4 x i32> %div
|
|
}
|
|
|
|
define <16 x i8> @udiv16xi8(<16 x i8> %x) {
|
|
; CHECK-LABEL: udiv16xi8:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: movi v1.16b, #121
|
|
; CHECK-NEXT: umull2 v2.8h, v0.16b, v1.16b
|
|
; CHECK-NEXT: umull v0.8h, v0.8b, v1.8b
|
|
; CHECK-NEXT: uzp2 v0.16b, v0.16b, v2.16b
|
|
; CHECK-NEXT: ushr v0.16b, v0.16b, #5
|
|
; CHECK-NEXT: ret
|
|
%div = udiv <16 x i8> %x, <i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68, i8 68>
|
|
ret <16 x i8> %div
|
|
}
|
|
|
|
define <8 x i16> @udiv8xi16(<8 x i16> %x) {
|
|
; CHECK-LABEL: udiv8xi16:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #16593 // =0x40d1
|
|
; CHECK-NEXT: dup v1.8h, w8
|
|
; CHECK-NEXT: umull2 v2.4s, v0.8h, v1.8h
|
|
; CHECK-NEXT: umull v1.4s, v0.4h, v1.4h
|
|
; CHECK-NEXT: uzp2 v1.8h, v1.8h, v2.8h
|
|
; CHECK-NEXT: sub v0.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: usra v1.8h, v0.8h, #1
|
|
; CHECK-NEXT: ushr v0.8h, v1.8h, #12
|
|
; CHECK-NEXT: ret
|
|
%div = udiv <8 x i16> %x, <i16 6537, i16 6537, i16 6537, i16 6537, i16 6537, i16 6537, i16 6537, i16 6537>
|
|
ret <8 x i16> %div
|
|
}
|
|
|
|
define <4 x i32> @udiv32xi4(<4 x i32> %x) {
|
|
; CHECK-LABEL: udiv32xi4:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mov w8, #16747 // =0x416b
|
|
; CHECK-NEXT: movk w8, #31439, lsl #16
|
|
; CHECK-NEXT: dup v1.4s, w8
|
|
; CHECK-NEXT: umull2 v2.2d, v0.4s, v1.4s
|
|
; CHECK-NEXT: umull v0.2d, v0.2s, v1.2s
|
|
; CHECK-NEXT: uzp2 v0.4s, v0.4s, v2.4s
|
|
; CHECK-NEXT: ushr v0.4s, v0.4s, #22
|
|
; CHECK-NEXT: ret
|
|
%div = udiv <4 x i32> %x, <i32 8743143, i32 8743143, i32 8743143, i32 8743143>
|
|
ret <4 x i32> %div
|
|
}
|