forked from libxsmm/libxsmm
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path.test-dnn.yml
136 lines (127 loc) · 7.04 KB
/
.test-dnn.yml
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
script:
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST ResNet-50 (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_resnet50.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_resnet50.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_resnet50.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST AlexNet (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_alexnet.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_alexnet.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_alexnet.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST Overfeat (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_overfeat.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_overfeat.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_overfeat.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST GoogleNet-v1 (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_googlenetv1.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_googlenetv1.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_googlenetv1.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST GoogleNet-v3 (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_googlenetv3.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_googlenetv3.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_googlenetv3.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST dcGAN (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_dcgan.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_dcgan.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_dcgan.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
for PAD in 0 1; do
echo; echo "--- TEST VGGa (format=${FORMAT} pad=${PAD} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_vgga.sh ${MB} ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_vgga.sh ${MB} ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_vgga.sh ${MB} ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export OMP_NUM_THREADS=$(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}" | cut -d_ -f1; else echo "0"; fi);
export CHECK=1 ITERS=1;
for FORMAT in $(if [ "" != "${FORMATS}" ]; then echo "${FORMATS}"; else echo "L"; fi); do
for PAD in 0 1; do
echo; echo "--- TEST DeepBench (format=${FORMAT} pad=${PAD})";
./run_deepbench.sh ${ITERS} -1 f32 F ${FORMAT} ${PAD} &&
./run_deepbench.sh ${ITERS} -1 f32 B ${FORMAT} ${PAD} &&
./run_deepbench.sh ${ITERS} -1 f32 U ${FORMAT} ${PAD};
done
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(echo; echo "--- TEST Quicktest";
for MB_NT in $(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}"; else echo "32_0"; fi); do
MB=$(echo ${MB_NT} | cut -d_ -f1);
export OMP_NUM_THREADS=$(echo ${MB_NT} | cut -d_ -f2);
./layer_example_f32 1 299 299 ${MB} 3 32 3 3 0 0 2 U T 1 &&
./layer_example_f32 1 13 13 ${MB} 192 384 3 3 1 1 1 B L 1;
done)
- make -e ${MAKEJ} && cd samples/deeplearning/cnnlayer && make -e ${MAKEJ} &&
(export OMP_NUM_THREADS=$(if [ "" != "${MB_THREADS}" ]; then echo "${MB_THREADS}" | cut -d_ -f1; else echo "0"; fi);
export CHECK=1 CHECK_SCALE=1 ITERS=1 MB=${OMP_NUM_THREADS};
for KIND in F B U; do
echo; echo "--- TEST ResNet-50 (precision=bf16 kind=${KIND} mb=${MB} nt=${OMP_NUM_THREADS})";
./run_resnet50.sh ${MB} ${ITERS} -1 bf16 F L 1 &&
./run_resnet50.sh ${MB} ${ITERS} -1 bf16 B L 1 &&
./run_resnet50.sh ${MB} ${ITERS} -1 bf16 U L 1;
done)