92 lines
3.7 KiB
LLVM
92 lines
3.7 KiB
LLVM
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
||
|
|
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme -verify-machineinstrs < %s | FileCheck %s
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_b(<vscale x 16 x i1> %p1, <vscale x 16 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_b:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.b[w12, 0]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv16i1(<vscale x 16 x i1> %p1, <vscale x 16 x i1> %p2, i32 %idx)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_b_imm(<vscale x 16 x i1> %p1, <vscale x 16 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_b_imm:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.b[w12, 15]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%add = add i32 %idx, 15
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv16i1(<vscale x 16 x i1> %p1, <vscale x 16 x i1> %p2, i32 %add)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_h(<vscale x 16 x i1> %p1, <vscale x 8 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_h:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.h[w12, 0]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv8i1(<vscale x 16 x i1> %p1, <vscale x 8 x i1> %p2, i32 %idx)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_h_imm(<vscale x 16 x i1> %p1, <vscale x 8 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_h_imm:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.h[w12, 7]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%add = add i32 %idx, 7
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv8i1(<vscale x 16 x i1> %p1, <vscale x 8 x i1> %p2, i32 %add)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_s(<vscale x 16 x i1> %p1, <vscale x 4 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_s:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.s[w12, 0]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv4i1(<vscale x 16 x i1> %p1, <vscale x 4 x i1> %p2, i32 %idx)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_s_imm(<vscale x 16 x i1> %p1, <vscale x 4 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_s_imm:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.s[w12, 3]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%add = add i32 %idx, 3
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv4i1(<vscale x 16 x i1> %p1, <vscale x 4 x i1> %p2, i32 %add)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_d(<vscale x 16 x i1> %p1, <vscale x 2 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_d:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.d[w12, 0]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv2i1(<vscale x 16 x i1> %p1, <vscale x 2 x i1> %p2, i32 %idx)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
define <vscale x 16 x i1> @psel_d_imm(<vscale x 16 x i1> %p1, <vscale x 2 x i1> %p2, i32 %idx) {
|
||
|
|
; CHECK-LABEL: psel_d_imm:
|
||
|
|
; CHECK: // %bb.0:
|
||
|
|
; CHECK-NEXT: mov w12, w0
|
||
|
|
; CHECK-NEXT: psel p0, p0, p1.d[w12, 1]
|
||
|
|
; CHECK-NEXT: ret
|
||
|
|
%add = add i32 %idx, 1
|
||
|
|
%res = call <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv2i1(<vscale x 16 x i1> %p1, <vscale x 2 x i1> %p2, i32 %add)
|
||
|
|
ret <vscale x 16 x i1> %res
|
||
|
|
}
|
||
|
|
|
||
|
|
declare <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv16i1(<vscale x 16 x i1>, <vscale x 16 x i1>, i32)
|
||
|
|
declare <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv8i1(<vscale x 16 x i1>, <vscale x 8 x i1>, i32)
|
||
|
|
declare <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv4i1(<vscale x 16 x i1>, <vscale x 4 x i1>, i32)
|
||
|
|
declare <vscale x 16 x i1> @llvm.aarch64.sve.psel.nxv2i1(<vscale x 16 x i1>, <vscale x 2 x i1>, i32)
|