|
105 | 105 |
|
106 | 106 | ###################################################################### |
107 | 107 | # The output will look like (omitting some columns): |
108 | | - |
109 | | -# --------------------------------- ------------ ------------ ------------ ------------ |
110 | | -# Name Self CPU CPU total CPU time avg # of Calls |
111 | | -# --------------------------------- ------------ ------------ ------------ ------------ |
112 | | -# model_inference 5.509ms 57.503ms 57.503ms 1 |
113 | | -# aten::conv2d 231.000us 31.931ms 1.597ms 20 |
114 | | -# aten::convolution 250.000us 31.700ms 1.585ms 20 |
115 | | -# aten::_convolution 336.000us 31.450ms 1.573ms 20 |
116 | | -# aten::mkldnn_convolution 30.838ms 31.114ms 1.556ms 20 |
117 | | -# aten::batch_norm 211.000us 14.693ms 734.650us 20 |
118 | | -# aten::_batch_norm_impl_index 319.000us 14.482ms 724.100us 20 |
119 | | -# aten::native_batch_norm 9.229ms 14.109ms 705.450us 20 |
120 | | -# aten::mean 332.000us 2.631ms 125.286us 21 |
121 | | -# aten::select 1.668ms 2.292ms 8.988us 255 |
122 | | -# --------------------------------- ------------ ------------ ------------ ------------ |
123 | | -# Self CPU time total: 57.549m |
| 108 | +# |
| 109 | +# .. code-block:: sh |
| 110 | +# |
| 111 | +# --------------------------------- ------------ ------------ ------------ ------------ |
| 112 | +# Name Self CPU CPU total CPU time avg # of Calls |
| 113 | +# --------------------------------- ------------ ------------ ------------ ------------ |
| 114 | +# model_inference 5.509ms 57.503ms 57.503ms 1 |
| 115 | +# aten::conv2d 231.000us 31.931ms 1.597ms 20 |
| 116 | +# aten::convolution 250.000us 31.700ms 1.585ms 20 |
| 117 | +# aten::_convolution 336.000us 31.450ms 1.573ms 20 |
| 118 | +# aten::mkldnn_convolution 30.838ms 31.114ms 1.556ms 20 |
| 119 | +# aten::batch_norm 211.000us 14.693ms 734.650us 20 |
| 120 | +# aten::_batch_norm_impl_index 319.000us 14.482ms 724.100us 20 |
| 121 | +# aten::native_batch_norm 9.229ms 14.109ms 705.450us 20 |
| 122 | +# aten::mean 332.000us 2.631ms 125.286us 21 |
| 123 | +# aten::select 1.668ms 2.292ms 8.988us 255 |
| 124 | +# --------------------------------- ------------ ------------ ------------ ------------ |
| 125 | +# Self CPU time total: 57.549m |
124 | 126 | # |
125 | 127 |
|
126 | 128 | ###################################################################### |
|
218 | 220 | # |
219 | 221 | # .. code-block:: sh |
220 | 222 | # |
221 | | -# ------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ |
222 | | -# Name Self XPU Self XPU % XPU total XPU time avg # of Calls |
223 | | -# ------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ |
224 | | -# model_inference 0.000us 0.00% 2.567ms 2.567ms 1 |
225 | | -# aten::conv2d 0.000us 0.00% 1.871ms 93.560us 20 |
226 | | -# aten::convolution 0.000us 0.00% 1.871ms 93.560us 20 |
227 | | -# aten::_convolution 0.000us 0.00% 1.871ms 93.560us 20 |
228 | | -# aten::convolution_overrideable 1.871ms 72.89% 1.871ms 93.560us 20 |
229 | | -# gen_conv 1.484ms 57.82% 1.484ms 74.216us 20 |
230 | | -# aten::batch_norm 0.000us 0.00% 432.640us 21.632us 20 |
231 | | -# aten::_batch_norm_impl_index 0.000us 0.00% 432.640us 21.632us 20 |
232 | | -# aten::native_batch_norm 432.640us 16.85% 432.640us 21.632us 20 |
233 | | -# conv_reorder 386.880us 15.07% 386.880us 6.448us 60 |
234 | | -# ------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ |
235 | | -# Self CPU time total: 712.486ms |
236 | | -# Self XPU time total: 2.567ms |
| 223 | +# ------------------------------ ------------ ------------ ------------ ------------ ------------ |
| 224 | +# Name Self XPU Self XPU % XPU total XPU time avg # of Calls |
| 225 | +# ------------------------------ ------------ ------------ ------------ ------------ ------------ |
| 226 | +# model_inference 0.000us 0.00% 2.567ms 2.567ms 1 |
| 227 | +# aten::conv2d 0.000us 0.00% 1.871ms 93.560us 20 |
| 228 | +# aten::convolution 0.000us 0.00% 1.871ms 93.560us 20 |
| 229 | +# aten::_convolution 0.000us 0.00% 1.871ms 93.560us 20 |
| 230 | +# aten::convolution_overrideable 1.871ms 72.89% 1.871ms 93.560us 20 |
| 231 | +# gen_conv 1.484ms 57.82% 1.484ms 74.216us 20 |
| 232 | +# aten::batch_norm 0.000us 0.00% 432.640us 21.632us 20 |
| 233 | +# aten::_batch_norm_impl_index 0.000us 0.00% 432.640us 21.632us 20 |
| 234 | +# aten::native_batch_norm 432.640us 16.85% 432.640us 21.632us 20 |
| 235 | +# conv_reorder 386.880us 15.07% 386.880us 6.448us 60 |
| 236 | +# ------------------------------ ------------ ------------ ------------ ------------ ------------ |
| 237 | +# Self CPU time total: 712.486ms |
| 238 | +# Self XPU time total: 2.567ms |
237 | 239 | # |
238 | 240 |
|
239 | 241 | ###################################################################### |
240 | | -# Note the occurrence of on-device kernels in the output (e.g. ``sgemm_32x32x32_NN``). |
| 242 | +# Note the occurrence of on-device kernels in the output (e.g. ``sgemm_32x32x32_NN`` for CUDA or ``gen_conv`` for XPU). |
241 | 243 |
|
242 | 244 | ###################################################################### |
243 | 245 | # 4. Using profiler to analyze memory consumption |
|
0 commit comments