Repository navigation
Expand file tree
/
Copy pathbench_test.go
More file actions
169 lines (153 loc) · 3.88 KB
/
Copy pathbench_test.go
File metadata and controls
169 lines (153 loc) · 3.88 KB
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
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
package ndarray
import (
"fmt"
"testing"
)
// Benchmarks for the operations compared against numpy in docs/perf.md. Sizes
// span small (threshold-relevant) to large (where multicore + SIMD win). Each
// benchmark builds its operands once and times the operation in the loop.
var benchElemSizes = []int{1 << 10, 1 << 14, 1 << 18, 1 << 22}
func benchVec(n int) *Array {
d := make([]float64, n)
for i := range d {
d[i] = float64(i%97) * 0.5
}
a, _ := FromData(d, n)
return a
}
func BenchmarkAdd(b *testing.B) {
for _, n := range benchElemSizes {
x, y := benchVec(n), benchVec(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
b.SetBytes(int64(n * 8))
for i := 0; i < b.N; i++ {
_, _ = x.Add(y)
}
})
}
}
func BenchmarkMul(b *testing.B) {
for _, n := range benchElemSizes {
x, y := benchVec(n), benchVec(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
b.SetBytes(int64(n * 8))
for i := 0; i < b.N; i++ {
_, _ = x.Mul(y)
}
})
}
}
func BenchmarkSqrt(b *testing.B) {
for _, n := range benchElemSizes {
x := benchVec(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_ = x.Sqrt()
}
})
}
}
func BenchmarkSum(b *testing.B) {
for _, n := range benchElemSizes {
x := benchVec(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_ = x.Sum()
}
})
}
}
func BenchmarkMax(b *testing.B) {
for _, n := range benchElemSizes {
x := benchVec(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_, _ = x.Max()
}
})
}
}
// The *Into benchmarks are the no-allocation parity path (np.add(x,y,out=z)):
// they reuse a caller-provided output buffer, so they measure the SIMD kernel
// without the per-op make that NumPy's temp-array cache avoids. This is where Go
// reaches and beats NumPy on small contiguous ops.
func BenchmarkAddInto(b *testing.B) {
for _, n := range benchElemSizes {
x, y := benchVec(n), benchVec(n)
out, _ := New(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
b.SetBytes(int64(n * 8))
for i := 0; i < b.N; i++ {
_ = x.AddInto(out, y)
}
})
}
}
func BenchmarkMulInto(b *testing.B) {
for _, n := range benchElemSizes {
x, y := benchVec(n), benchVec(n)
out, _ := New(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
b.SetBytes(int64(n * 8))
for i := 0; i < b.N; i++ {
_ = x.MulInto(out, y)
}
})
}
}
func BenchmarkSqrtInto(b *testing.B) {
for _, n := range benchElemSizes {
x := benchVec(n)
out, _ := New(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_ = x.SqrtInto(out)
}
})
}
}
func benchMat(n int) *Array {
d := make([]float64, n*n)
for i := range d {
d[i] = float64(i%101) * 0.25
}
a, _ := FromData(d, n, n)
return a
}
// BenchmarkMatMulVariants covers what BenchmarkMatMul does not: float32
// operands, a transposed operand (read in place by the GEMM) and a strided one
// (a[:, ::2], which the GEMM entry copies first).
func BenchmarkMatMulVariants(b *testing.B) {
for _, n := range []int{64, 256, 1024} {
x, y := benchMat(n), benchMat(n)
x32, y32 := x.AsType(Float32), y.AsType(Float32)
wide, _ := FromData(make([]float64, 2*n*n), n, 2*n)
half, _ := wide.Slice(All(), Step(2))
xt := x.Transpose()
b.Run(fmt.Sprintf("f32/n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_, _ = x32.MatMul(y32)
}
})
b.Run(fmt.Sprintf("transA/n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_, _ = xt.MatMul(y)
}
})
b.Run(fmt.Sprintf("strided/n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_, _ = half.MatMul(y)
}
})
}
}
func BenchmarkMatMul(b *testing.B) {
for _, n := range []int{64, 128, 256, 512, 1024} {
x, y := benchMat(n), benchMat(n)
b.Run(fmt.Sprintf("n=%d", n), func(b *testing.B) {
for i := 0; i < b.N; i++ {
_, _ = x.MatMul(y)
}
})
}
}