128 lines
3.9 KiB
LLVM
128 lines
3.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=aarch64-- | FileCheck %s --check-prefixes=CHECK,CHECK-SD
|
|
; RUN: llc < %s -mtriple=aarch64-- -global-isel -global-isel-abort=2 2>&1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI
|
|
|
|
declare i4 @llvm.uadd.sat.i4(i4, i4)
|
|
declare i8 @llvm.uadd.sat.i8(i8, i8)
|
|
declare i16 @llvm.uadd.sat.i16(i16, i16)
|
|
declare i32 @llvm.uadd.sat.i32(i32, i32)
|
|
declare i64 @llvm.uadd.sat.i64(i64, i64)
|
|
|
|
define i32 @func32(i32 %x, i32 %y, i32 %z) nounwind {
|
|
; CHECK-SD-LABEL: func32:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: mul w8, w1, w2
|
|
; CHECK-SD-NEXT: adds w8, w0, w8
|
|
; CHECK-SD-NEXT: csinv w0, w8, wzr, lo
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: func32:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: mul w8, w1, w2
|
|
; CHECK-GI-NEXT: adds w8, w0, w8
|
|
; CHECK-GI-NEXT: cset w9, hs
|
|
; CHECK-GI-NEXT: tst w9, #0x1
|
|
; CHECK-GI-NEXT: csinv w0, w8, wzr, eq
|
|
; CHECK-GI-NEXT: ret
|
|
%a = mul i32 %y, %z
|
|
%tmp = call i32 @llvm.uadd.sat.i32(i32 %x, i32 %a)
|
|
ret i32 %tmp
|
|
}
|
|
|
|
define i64 @func64(i64 %x, i64 %y, i64 %z) nounwind {
|
|
; CHECK-SD-LABEL: func64:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: adds x8, x0, x2
|
|
; CHECK-SD-NEXT: csinv x0, x8, xzr, lo
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: func64:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: adds x8, x0, x2
|
|
; CHECK-GI-NEXT: cset w9, hs
|
|
; CHECK-GI-NEXT: tst w9, #0x1
|
|
; CHECK-GI-NEXT: csinv x0, x8, xzr, eq
|
|
; CHECK-GI-NEXT: ret
|
|
%a = mul i64 %y, %z
|
|
%tmp = call i64 @llvm.uadd.sat.i64(i64 %x, i64 %z)
|
|
ret i64 %tmp
|
|
}
|
|
|
|
define i16 @func16(i16 %x, i16 %y, i16 %z) nounwind {
|
|
; CHECK-SD-LABEL: func16:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: mul w8, w1, w2
|
|
; CHECK-SD-NEXT: and w9, w0, #0xffff
|
|
; CHECK-SD-NEXT: add w8, w9, w8, uxth
|
|
; CHECK-SD-NEXT: mov w9, #65535 // =0xffff
|
|
; CHECK-SD-NEXT: cmp w8, w9
|
|
; CHECK-SD-NEXT: csel w0, w8, w9, lo
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: func16:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: mul w8, w1, w2
|
|
; CHECK-GI-NEXT: and w8, w8, #0xffff
|
|
; CHECK-GI-NEXT: add w8, w8, w0, uxth
|
|
; CHECK-GI-NEXT: cmp w8, w8, uxth
|
|
; CHECK-GI-NEXT: csinv w0, w8, wzr, eq
|
|
; CHECK-GI-NEXT: ret
|
|
%a = mul i16 %y, %z
|
|
%tmp = call i16 @llvm.uadd.sat.i16(i16 %x, i16 %a)
|
|
ret i16 %tmp
|
|
}
|
|
|
|
define i8 @func8(i8 %x, i8 %y, i8 %z) nounwind {
|
|
; CHECK-SD-LABEL: func8:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: mul w8, w1, w2
|
|
; CHECK-SD-NEXT: and w9, w0, #0xff
|
|
; CHECK-SD-NEXT: add w8, w9, w8, uxtb
|
|
; CHECK-SD-NEXT: mov w9, #255 // =0xff
|
|
; CHECK-SD-NEXT: cmp w8, #255
|
|
; CHECK-SD-NEXT: csel w0, w8, w9, lo
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: func8:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: mul w8, w1, w2
|
|
; CHECK-GI-NEXT: and w8, w8, #0xff
|
|
; CHECK-GI-NEXT: add w8, w8, w0, uxtb
|
|
; CHECK-GI-NEXT: cmp w8, w8, uxtb
|
|
; CHECK-GI-NEXT: csinv w0, w8, wzr, eq
|
|
; CHECK-GI-NEXT: ret
|
|
%a = mul i8 %y, %z
|
|
%tmp = call i8 @llvm.uadd.sat.i8(i8 %x, i8 %a)
|
|
ret i8 %tmp
|
|
}
|
|
|
|
define i4 @func4(i4 %x, i4 %y, i4 %z) nounwind {
|
|
; CHECK-SD-LABEL: func4:
|
|
; CHECK-SD: // %bb.0:
|
|
; CHECK-SD-NEXT: mul w8, w1, w2
|
|
; CHECK-SD-NEXT: and w9, w0, #0xf
|
|
; CHECK-SD-NEXT: and w8, w8, #0xf
|
|
; CHECK-SD-NEXT: add w8, w9, w8
|
|
; CHECK-SD-NEXT: mov w9, #15 // =0xf
|
|
; CHECK-SD-NEXT: cmp w8, #15
|
|
; CHECK-SD-NEXT: csel w0, w8, w9, lo
|
|
; CHECK-SD-NEXT: ret
|
|
;
|
|
; CHECK-GI-LABEL: func4:
|
|
; CHECK-GI: // %bb.0:
|
|
; CHECK-GI-NEXT: mul w8, w1, w2
|
|
; CHECK-GI-NEXT: and w9, w0, #0xf
|
|
; CHECK-GI-NEXT: mov w10, #15 // =0xf
|
|
; CHECK-GI-NEXT: and w8, w8, #0xf
|
|
; CHECK-GI-NEXT: add w8, w9, w8
|
|
; CHECK-GI-NEXT: and w9, w8, #0xf
|
|
; CHECK-GI-NEXT: cmp w8, w9
|
|
; CHECK-GI-NEXT: csel w0, w10, w8, ne
|
|
; CHECK-GI-NEXT: ret
|
|
%a = mul i4 %y, %z
|
|
%tmp = call i4 @llvm.uadd.sat.i4(i4 %x, i4 %a)
|
|
ret i4 %tmp
|
|
}
|
|
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
|
|
; CHECK: {{.*}}
|