1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=i686-- -mattr=+sse | FileCheck %s --check-prefix=SSE
3 ; RUN: llc < %s -mtriple=i686-- -mattr=+avx | FileCheck %s --check-prefix=SSE
4 ; RUN: llc < %s -mtriple=i686-- -mattr=+sse,+prfchw | FileCheck %s -check-prefix=PRFCHWSSE
5 ; RUN: llc < %s -mtriple=i686-- -mattr=+prfchw | FileCheck %s -check-prefix=PRFCHWSSE
6 ; RUN: llc < %s -mtriple=i686-- -mcpu=slm | FileCheck %s -check-prefix=PRFCHWSSE
7 ; RUN: llc < %s -mtriple=i686-- -mcpu=btver2 | FileCheck %s -check-prefix=PRFCHWSSE
8 ; RUN: llc < %s -mtriple=i686-- -mcpu=btver2 -mattr=-prfchw | FileCheck %s -check-prefix=SSE
9 ; RUN: llc < %s -mtriple=i686-- -mattr=+sse,+prefetchwt1 | FileCheck %s -check-prefix=PREFETCHWT1
10 ; RUN: llc < %s -mtriple=i686-- -mattr=-sse,+prefetchwt1 | FileCheck %s -check-prefix=PREFETCHWT1
11 ; RUN: llc < %s -mtriple=i686-- -mattr=-sse,+3dnow,+prefetchwt1 | FileCheck %s -check-prefix=PREFETCHWT1
12 ; RUN: llc < %s -mtriple=i686-- -mattr=+3dnow | FileCheck %s -check-prefix=3DNOW
13 ; RUN: llc < %s -mtriple=i686-- -mattr=+3dnow,+prfchw | FileCheck %s -check-prefix=PRFCHW3DNOW
16 ; 3dnow by itself get you just the single prefetch instruction with no hints
17 ; sse provides prefetch0/1/2/nta
18 ; supporting prefetchw, but not 3dnow implicitly provides prefetcht0/1/2/nta regardless of sse setting as we need something to fall back to for the non-write hint.
19 ; supporting prefetchwt1 implies prefetcht0/1/2/nta and prefetchw regardless of other settings. this allows levels for non-write and gives us an instruction for write+T0
20 ; 3dnow prefetch instruction will only get used if you have no other prefetch instructions enabled
24 define void @t(i8* %ptr) nounwind {
26 ; SSE: # %bb.0: # %entry
27 ; SSE-NEXT: movl {{[0-9]+}}(%esp), %eax
28 ; SSE-NEXT: prefetcht2 (%eax)
29 ; SSE-NEXT: prefetcht1 (%eax)
30 ; SSE-NEXT: prefetcht0 (%eax)
31 ; SSE-NEXT: prefetchnta (%eax)
32 ; SSE-NEXT: prefetcht2 (%eax)
33 ; SSE-NEXT: prefetcht1 (%eax)
34 ; SSE-NEXT: prefetcht0 (%eax)
35 ; SSE-NEXT: prefetchnta (%eax)
39 ; PRFCHWSSE: # %bb.0: # %entry
40 ; PRFCHWSSE-NEXT: movl {{[0-9]+}}(%esp), %eax
41 ; PRFCHWSSE-NEXT: prefetcht2 (%eax)
42 ; PRFCHWSSE-NEXT: prefetcht1 (%eax)
43 ; PRFCHWSSE-NEXT: prefetcht0 (%eax)
44 ; PRFCHWSSE-NEXT: prefetchnta (%eax)
45 ; PRFCHWSSE-NEXT: prefetchw (%eax)
46 ; PRFCHWSSE-NEXT: prefetchw (%eax)
47 ; PRFCHWSSE-NEXT: prefetchw (%eax)
48 ; PRFCHWSSE-NEXT: prefetchw (%eax)
49 ; PRFCHWSSE-NEXT: retl
51 ; PREFETCHWT1-LABEL: t:
52 ; PREFETCHWT1: # %bb.0: # %entry
53 ; PREFETCHWT1-NEXT: movl {{[0-9]+}}(%esp), %eax
54 ; PREFETCHWT1-NEXT: prefetcht2 (%eax)
55 ; PREFETCHWT1-NEXT: prefetcht1 (%eax)
56 ; PREFETCHWT1-NEXT: prefetcht0 (%eax)
57 ; PREFETCHWT1-NEXT: prefetchnta (%eax)
58 ; PREFETCHWT1-NEXT: prefetchwt1 (%eax)
59 ; PREFETCHWT1-NEXT: prefetchwt1 (%eax)
60 ; PREFETCHWT1-NEXT: prefetchw (%eax)
61 ; PREFETCHWT1-NEXT: prefetchwt1 (%eax)
62 ; PREFETCHWT1-NEXT: retl
65 ; 3DNOW: # %bb.0: # %entry
66 ; 3DNOW-NEXT: movl {{[0-9]+}}(%esp), %eax
67 ; 3DNOW-NEXT: prefetch (%eax)
68 ; 3DNOW-NEXT: prefetch (%eax)
69 ; 3DNOW-NEXT: prefetch (%eax)
70 ; 3DNOW-NEXT: prefetch (%eax)
71 ; 3DNOW-NEXT: prefetch (%eax)
72 ; 3DNOW-NEXT: prefetch (%eax)
73 ; 3DNOW-NEXT: prefetch (%eax)
74 ; 3DNOW-NEXT: prefetch (%eax)
77 ; PRFCHW3DNOW-LABEL: t:
78 ; PRFCHW3DNOW: # %bb.0: # %entry
79 ; PRFCHW3DNOW-NEXT: movl {{[0-9]+}}(%esp), %eax
80 ; PRFCHW3DNOW-NEXT: prefetch (%eax)
81 ; PRFCHW3DNOW-NEXT: prefetch (%eax)
82 ; PRFCHW3DNOW-NEXT: prefetch (%eax)
83 ; PRFCHW3DNOW-NEXT: prefetch (%eax)
84 ; PRFCHW3DNOW-NEXT: prefetchw (%eax)
85 ; PRFCHW3DNOW-NEXT: prefetchw (%eax)
86 ; PRFCHW3DNOW-NEXT: prefetchw (%eax)
87 ; PRFCHW3DNOW-NEXT: prefetchw (%eax)
88 ; PRFCHW3DNOW-NEXT: retl
90 tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 1, i32 1 )
91 tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 2, i32 1 )
92 tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 3, i32 1 )
93 tail call void @llvm.prefetch( i8* %ptr, i32 0, i32 0, i32 1 )
94 tail call void @llvm.prefetch( i8* %ptr, i32 1, i32 1, i32 1 )
95 tail call void @llvm.prefetch( i8* %ptr, i32 1, i32 2, i32 1 )
96 tail call void @llvm.prefetch( i8* %ptr, i32 1, i32 3, i32 1 )
97 tail call void @llvm.prefetch( i8* %ptr, i32 1, i32 0, i32 1 )
101 declare void @llvm.prefetch(i8*, i32, i32, i32) nounwind