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
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
|
; RUN: llvm-as -o - %s | llc -march=cellspu > %t1.s
; RUN: grep fa %t1.s | count 2 &&
; RUN: grep fs %t1.s | count 2 &&
; RUN: grep fm %t1.s | count 6 &&
; RUN: grep fma %t1.s | count 2 &&
; RUN: grep fms %t1.s | count 2 &&
; RUN: grep fnms %t1.s | count 3
;
; This file includes standard floating point arithmetic instructions
; NOTE fdiv is tested separately since it is a compound operation
target datalayout = "E-p:32:32:128-f64:64:128-f32:32:128-i64:32:128-i32:32:128-i16:16:128-i8:8:128-i1:8:128-a0:0:128-v128:128:128-s0:128:128"
target triple = "spu"
define float @fp_add(float %arg1, float %arg2) {
%A = add float %arg1, %arg2 ; <float> [#uses=1]
ret float %A
}
define <4 x float> @fp_add_vec(<4 x float> %arg1, <4 x float> %arg2) {
%A = add <4 x float> %arg1, %arg2 ; <<4 x float>> [#uses=1]
ret <4 x float> %A
}
define float @fp_sub(float %arg1, float %arg2) {
%A = sub float %arg1, %arg2 ; <float> [#uses=1]
ret float %A
}
define <4 x float> @fp_sub_vec(<4 x float> %arg1, <4 x float> %arg2) {
%A = sub <4 x float> %arg1, %arg2 ; <<4 x float>> [#uses=1]
ret <4 x float> %A
}
define float @fp_mul(float %arg1, float %arg2) {
%A = mul float %arg1, %arg2 ; <float> [#uses=1]
ret float %A
}
define <4 x float> @fp_mul_vec(<4 x float> %arg1, <4 x float> %arg2) {
%A = mul <4 x float> %arg1, %arg2 ; <<4 x float>> [#uses=1]
ret <4 x float> %A
}
define float @fp_mul_add(float %arg1, float %arg2, float %arg3) {
%A = mul float %arg1, %arg2 ; <float> [#uses=1]
%B = add float %A, %arg3 ; <float> [#uses=1]
ret float %B
}
define <4 x float> @fp_mul_add_vec(<4 x float> %arg1, <4 x float> %arg2, <4 x float> %arg3) {
%A = mul <4 x float> %arg1, %arg2 ; <<4 x float>> [#uses=1]
%B = add <4 x float> %A, %arg3 ; <<4 x float>> [#uses=1]
ret <4 x float> %B
}
define float @fp_mul_sub(float %arg1, float %arg2, float %arg3) {
%A = mul float %arg1, %arg2 ; <float> [#uses=1]
%B = sub float %A, %arg3 ; <float> [#uses=1]
ret float %B
}
define <4 x float> @fp_mul_sub_vec(<4 x float> %arg1, <4 x float> %arg2, <4 x float> %arg3) {
%A = mul <4 x float> %arg1, %arg2 ; <<4 x float>> [#uses=1]
%B = sub <4 x float> %A, %arg3 ; <<4 x float>> [#uses=1]
ret <4 x float> %B
}
; Test the straightforward way of getting fnms
; c - a * b
define float @fp_neg_mul_sub_1(float %arg1, float %arg2, float %arg3) {
%A = mul float %arg1, %arg2
%B = sub float %arg3, %A
ret float %B
}
; Test another way of getting fnms
; - ( a *b -c ) = c - a * b
define float @fp_neg_mul_sub_2(float %arg1, float %arg2, float %arg3) {
%A = mul float %arg1, %arg2
%B = sub float %A, %arg3
%C = sub float -0.0, %B
ret float %C
}
define <4 x float> @fp_neg_mul_sub_vec(<4 x float> %arg1, <4 x float> %arg2, <4 x float> %arg3) {
%A = mul <4 x float> %arg1, %arg2
%B = sub <4 x float> %A, %arg3
%D = sub <4 x float> < float -0.0, float -0.0, float -0.0, float -0.0 >, %B
ret <4 x float> %D
}
|
