File: arm-neon-shifts.c

package info (click to toggle)
llvm-toolchain-20 1%3A20.1.6-1~exp1
  • links: PTS, VCS
  • area: main
  • in suites: experimental
  • size: 2,111,304 kB
  • sloc: cpp: 7,438,677; ansic: 1,393,822; asm: 1,012,926; python: 241,650; f90: 86,635; objc: 75,479; lisp: 42,144; pascal: 17,286; sh: 10,027; ml: 5,082; perl: 4,730; awk: 3,523; makefile: 3,349; javascript: 2,251; xml: 892; fortran: 672
file content (47 lines) | stat: -rw-r--r-- 1,429 bytes parent folder | download | duplicates (6)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
// REQUIRES: arm-registered-target
// RUN: %clang_cc1 -triple thumbv7-apple-darwin \
// RUN:   -disable-O0-optnone \
// RUN:   -target-cpu cortex-a8 \
// RUN:   -ffreestanding \
// RUN:   -emit-llvm -w -o - %s | opt -S -passes=mem2reg | FileCheck %s

#include <arm_neon.h>

uint8x8_t test_shift_vshr(uint8x8_t a) {
  // CHECK-LABEL: test_shift_vshr
  // CHECK: %{{.*}} = lshr <8 x i8> %a, splat (i8 5)
  return vshr_n_u8(a, 5);
}

int8x8_t test_shift_vshr_smax(int8x8_t a) {
  // CHECK-LABEL: test_shift_vshr_smax
  // CHECK: %{{.*}} = ashr <8 x i8> %a, splat (i8 7)
  return vshr_n_s8(a, 8);
}

uint8x8_t test_shift_vshr_umax(uint8x8_t a) {
  // CHECK-LABEL: test_shift_vshr_umax
  // CHECK: ret <8 x i8> zeroinitializer
  return vshr_n_u8(a, 8);
}

uint8x8_t test_shift_vsra(uint8x8_t a, uint8x8_t b) {
  // CHECK-LABEL: test_shift_vsra
  // CHECK: %[[SHR:.*]] = lshr <8 x i8> %b, splat (i8 5)
  // CHECK: %{{.*}} = add <8 x i8> %a, %[[SHR]]
  return vsra_n_u8(a, b, 5);
}

int8x8_t test_shift_vsra_smax(int8x8_t a, int8x8_t b) {
  // CHECK-LABEL: test_shift_vsra_smax
  // CHECK: %[[SHR:.*]] = ashr <8 x i8> %b, splat (i8 7)
  // CHECK: %{{.*}} = add <8 x i8> %a, %[[SHR]]
  return vsra_n_s8(a, b, 8);
}

uint8x8_t test_shift_vsra_umax(uint8x8_t a, uint8x8_t b) {
  // CHECK-LABEL: test_shift_vsra_umax
  // CHECK: [[RES:%.*]] = add <8 x i8> %a, zeroinitializer
  // CHECK: ret <8 x i8> [[RES]]
  return vsra_n_u8(a, b, 8);
}