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
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
|
; RUN: llc < %s -march=nvptx -mcpu=sm_20 -fp-contract=fast | FileCheck %s
; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 -fp-contract=fast | FileCheck %s
; RUN: %if ptxas %{ llc < %s -march=nvptx -mcpu=sm_20 -fp-contract=fast | %ptxas-verify %}
; RUN: %if ptxas %{ llc < %s -march=nvptx64 -mcpu=sm_20 -fp-contract=fast | %ptxas-verify %}
;; These tests should run for all targets
;;===-- Basic instruction selection tests ---------------------------------===;;
;;; f64
define double @fadd_f64(double %a, double %b) {
; CHECK: add.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}
; CHECK: ret
%ret = fadd double %a, %b
ret double %ret
}
define double @fsub_f64(double %a, double %b) {
; CHECK: sub.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}
; CHECK: ret
%ret = fsub double %a, %b
ret double %ret
}
define double @fmul_f64(double %a, double %b) {
; CHECK: mul.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}
; CHECK: ret
%ret = fmul double %a, %b
ret double %ret
}
define double @fdiv_f64(double %a, double %b) {
; CHECK: div.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}
; CHECK: ret
%ret = fdiv double %a, %b
ret double %ret
}
;; PTX does not have a floating-point rem instruction
;;; f32
define float @fadd_f32(float %a, float %b) {
; CHECK: add.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}
; CHECK: ret
%ret = fadd float %a, %b
ret float %ret
}
define float @fsub_f32(float %a, float %b) {
; CHECK: sub.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}
; CHECK: ret
%ret = fsub float %a, %b
ret float %ret
}
define float @fmul_f32(float %a, float %b) {
; CHECK: mul.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}
; CHECK: ret
%ret = fmul float %a, %b
ret float %ret
}
define float @fdiv_f32(float %a, float %b) {
; CHECK: div.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}
; CHECK: ret
%ret = fdiv float %a, %b
ret float %ret
}
;; PTX does not have a floating-point rem instruction
|
