Mon Nov 10 21:57:49 2025
+Fri Dec 19 18:56:39 2025
+-----------------------------------------------------------------------------------------+
-| NVIDIA-SMI 580.95.05 Driver Version: 580.95.05 CUDA Version: 13.0 |
+| NVIDIA-SMI 580.105.08 Driver Version: 580.105.08 CUDA Version: 13.0 |
+-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA L40S On | 00000000:4D:00.0 Off | 0 |
-| N/A 27C P0 77W / 350W | 0MiB / 46068MiB | 18% Default |
+| N/A 33C P0 78W / 350W | 0MiB / 46068MiB | 11% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
@@ -3937,7 +3937,7 @@ Cell: nv | 0.22s
▼ output
▶ uv-logs
|
-Cell: benchmark | 3.89s
+Cell: benchmark | 3.84s
| ▶ run
Copy
Raw
@@ -3999,29 +3999,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D64_S128_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 411.136us 2127.15% 411.136us 411.136us 1
- torch_eager 8.60% 205.173us 99.40% 2.372ms 2.372ms 0.000us 0.00% 21.632us 21.632us 1
- aten::to 0.40% 9.649us 83.06% 1.982ms 330.358us 0.000us 0.00% 14.272us 2.379us 6
- aten::_to_copy 1.47% 35.141us 82.65% 1.973ms 328.750us 0.000us 0.00% 14.272us 2.379us 6
- aten::copy_ 2.42% 57.830us 79.13% 1.889ms 314.753us 11.968us 61.92% 14.272us 2.379us 6
- aten::conv1d 0.32% 7.640us 6.22% 148.384us 49.461us 0.000us 0.00% 7.360us 2.453us 3
- aten::convolution 0.55% 13.222us 5.90% 140.744us 46.915us 0.000us 0.00% 7.360us 2.453us 3
- aten::_convolution 1.23% 29.427us 5.34% 127.522us 42.507us 0.000us 0.00% 7.360us 2.453us 3
- aten::_conv_depthwise2d 1.41% 33.690us 3.44% 82.073us 27.358us 7.360us 38.08% 7.360us 2.453us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.360us 38.08% 7.360us 2.453us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.304us 32.62% 6.304us 2.101us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.664us 29.30% 5.664us 1.888us 3
- Activity Buffer Request 73.85% 1.762ms 73.85% 1.762ms 1.762ms 2.304us 11.92% 2.304us 2.304us 1
- aten::empty_strided 2.05% 48.841us 2.05% 48.841us 8.140us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 3.88% 92.484us 3.88% 92.484us 10.276us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.94% 22.551us 1.23% 29.352us 3.261us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.46% 10.991us 0.46% 10.991us 0.733us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.53% 12.660us 0.53% 12.660us 4.220us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.49% 11.631us 0.49% 11.631us 3.877us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.27% 6.340us 0.32% 7.570us 2.523us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 420.863us 2177.59% 420.863us 420.863us 1
+ torch_eager 8.44% 198.071us 99.39% 2.333ms 2.333ms 0.000us 0.00% 21.695us 21.695us 1
+ aten::to 0.42% 9.821us 82.68% 1.941ms 323.474us 0.000us 0.00% 14.367us 2.395us 6
+ aten::_to_copy 1.47% 34.509us 82.26% 1.931ms 321.837us 0.000us 0.00% 14.367us 2.395us 6
+ aten::copy_ 2.58% 60.614us 78.68% 1.847ms 307.834us 11.999us 62.08% 14.367us 2.395us 6
+ aten::conv1d 0.31% 7.291us 6.62% 155.424us 51.808us 0.000us 0.00% 7.328us 2.443us 3
+ aten::convolution 0.56% 13.171us 6.31% 148.133us 49.378us 0.000us 0.00% 7.328us 2.443us 3
+ aten::_convolution 1.45% 33.972us 5.75% 134.962us 44.987us 0.000us 0.00% 7.328us 2.443us 3
+ aten::_conv_depthwise2d 1.44% 33.841us 3.61% 84.652us 28.217us 7.328us 37.92% 7.328us 2.443us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.328us 37.92% 7.328us 2.443us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.335us 32.78% 6.335us 2.112us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.664us 29.31% 5.664us 1.888us 3
+ Activity Buffer Request 73.26% 1.720ms 73.26% 1.720ms 1.720ms 2.368us 12.25% 2.368us 2.368us 1
+ aten::empty_strided 2.11% 49.511us 2.11% 49.511us 8.252us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 3.97% 93.272us 3.97% 93.272us 10.364us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 1.01% 23.716us 1.33% 31.137us 3.460us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.50% 11.722us 0.50% 11.722us 0.781us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.50% 11.630us 0.50% 11.630us 3.877us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.53% 12.550us 0.53% 12.550us 4.183us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.29% 6.740us 0.34% 8.071us 2.690us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.386ms
-Self CUDA time total: 19.328us
+Self CPU time total: 2.348ms
+Self CUDA time total: 19.327us
@@ -4031,29 +4031,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D64_S128_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 320.094us 1629.14% 320.094us 320.094us 1
- torch_eager 6.61% 147.267us 99.75% 2.222ms 2.222ms 0.000us 0.00% 21.856us 21.856us 1
- aten::to 0.28% 6.328us 86.86% 1.935ms 322.525us 0.000us 0.00% 13.888us 2.315us 6
- aten::_to_copy 0.99% 22.058us 86.58% 1.929ms 321.470us 0.000us 0.00% 13.888us 2.315us 6
- aten::copy_ 2.09% 46.581us 84.13% 1.874ms 312.384us 11.680us 59.45% 13.888us 2.315us 6
- aten::conv1d 0.26% 5.880us 5.20% 115.901us 38.634us 0.000us 0.00% 7.968us 2.656us 3
- aten::convolution 0.41% 9.201us 4.94% 110.021us 36.674us 0.000us 0.00% 7.968us 2.656us 3
- aten::_convolution 0.99% 22.029us 4.53% 100.820us 33.607us 0.000us 0.00% 7.968us 2.656us 3
- aten::_conv_depthwise2d 0.98% 21.809us 2.84% 63.210us 21.070us 7.968us 40.55% 7.968us 2.656us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.968us 40.55% 7.968us 2.656us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.112us 31.11% 6.112us 2.037us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.568us 28.34% 5.568us 1.856us 3
- Activity Buffer Request 79.89% 1.780ms 79.89% 1.780ms 1.780ms 2.208us 11.24% 2.208us 2.208us 1
- aten::empty_strided 1.46% 32.461us 1.46% 32.461us 5.410us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 3.22% 71.802us 3.22% 71.802us 7.978us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.71% 15.809us 0.93% 20.750us 2.306us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.38% 8.492us 0.38% 8.492us 0.566us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.41% 9.081us 0.41% 9.081us 3.027us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.38% 8.530us 0.38% 8.530us 2.843us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.26% 5.730us 0.32% 7.140us 2.380us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 313.250us 1596.75% 313.250us 313.250us 1
+ torch_eager 6.72% 145.881us 99.75% 2.164ms 2.164ms 0.000us 0.00% 21.826us 21.826us 1
+ aten::to 0.27% 5.881us 86.54% 1.878ms 312.937us 0.000us 0.00% 13.921us 2.320us 6
+ aten::_to_copy 1.01% 21.932us 86.27% 1.872ms 311.957us 0.000us 0.00% 13.921us 2.320us 6
+ aten::copy_ 2.19% 47.440us 83.99% 1.822ms 303.712us 11.713us 59.71% 13.921us 2.320us 6
+ aten::conv1d 0.28% 6.160us 5.31% 115.143us 38.381us 0.000us 0.00% 7.905us 2.635us 3
+ aten::convolution 0.41% 8.901us 5.02% 108.983us 36.328us 0.000us 0.00% 7.905us 2.635us 3
+ aten::_convolution 1.01% 21.821us 4.61% 100.082us 33.361us 0.000us 0.00% 7.905us 2.635us 3
+ aten::_conv_depthwise2d 0.97% 21.100us 2.86% 62.101us 20.700us 7.905us 40.29% 7.905us 2.635us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.905us 40.29% 7.905us 2.635us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.113us 31.16% 6.113us 2.038us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.600us 28.55% 5.600us 1.867us 3
+ Activity Buffer Request 79.65% 1.728ms 79.65% 1.728ms 1.728ms 2.208us 11.25% 2.208us 2.208us 1
+ aten::empty_strided 1.27% 27.539us 1.27% 27.539us 4.590us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 3.21% 69.681us 3.21% 69.681us 7.742us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.78% 16.828us 1.04% 22.610us 2.512us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.44% 9.512us 0.44% 9.512us 0.634us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.43% 9.331us 0.43% 9.331us 3.110us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.41% 8.800us 0.41% 8.800us 2.933us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.26% 5.580us 0.33% 7.220us 2.407us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.228ms
-Self CUDA time total: 19.648us
+Self CPU time total: 2.170ms
+Self CUDA time total: 19.618us
@@ -4063,29 +4063,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D64_S512_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 322.750us 1724.09% 322.750us 322.750us 1
- torch_eager 6.97% 154.353us 99.74% 2.208ms 2.208ms 0.000us 0.00% 20.736us 20.736us 1
- aten::to 0.30% 6.580us 86.44% 1.913ms 318.849us 0.000us 0.00% 13.791us 2.299us 6
- aten::_to_copy 1.09% 24.161us 86.14% 1.907ms 317.752us 0.000us 0.00% 13.791us 2.299us 6
- aten::copy_ 2.12% 46.909us 83.64% 1.851ms 308.533us 11.775us 62.90% 13.791us 2.299us 6
- aten::conv1d 0.30% 6.591us 5.18% 114.662us 38.221us 0.000us 0.00% 6.945us 2.315us 3
- aten::convolution 0.40% 8.811us 4.88% 108.071us 36.024us 0.000us 0.00% 6.945us 2.315us 3
- aten::_convolution 0.96% 21.188us 4.48% 99.260us 33.087us 0.000us 0.00% 6.945us 2.315us 3
- aten::_conv_depthwise2d 0.97% 21.520us 2.82% 62.461us 20.820us 6.945us 37.10% 6.945us 2.315us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 6.945us 37.10% 6.945us 2.315us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.047us 32.30% 6.047us 2.016us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.728us 30.60% 5.728us 1.909us 3
- Activity Buffer Request 79.41% 1.758ms 79.41% 1.758ms 1.758ms 2.016us 10.77% 2.016us 2.016us 1
- aten::empty_strided 1.41% 31.151us 1.41% 31.151us 5.192us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 3.17% 70.153us 3.17% 70.153us 7.795us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.77% 17.060us 1.01% 22.310us 2.479us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.39% 8.641us 0.39% 8.641us 0.576us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.42% 9.380us 0.42% 9.380us 3.127us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.37% 8.090us 0.37% 8.090us 2.697us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.25% 5.450us 0.31% 6.801us 2.267us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 311.776us 1662.54% 311.776us 311.776us 1
+ torch_eager 6.35% 136.063us 99.73% 2.136ms 2.136ms 0.000us 0.00% 20.769us 20.769us 1
+ aten::to 0.26% 5.629us 86.85% 1.860ms 310.039us 0.000us 0.00% 13.823us 2.304us 6
+ aten::_to_copy 0.98% 21.032us 86.59% 1.855ms 309.100us 0.000us 0.00% 13.823us 2.304us 6
+ aten::copy_ 2.10% 44.987us 84.33% 1.806ms 301.043us 11.807us 62.96% 13.823us 2.304us 6
+ aten::conv1d 0.27% 5.831us 5.33% 114.232us 38.077us 0.000us 0.00% 6.946us 2.315us 3
+ aten::convolution 0.40% 8.671us 5.06% 108.401us 36.134us 0.000us 0.00% 6.946us 2.315us 3
+ aten::_convolution 0.98% 20.999us 4.66% 99.730us 33.243us 0.000us 0.00% 6.946us 2.315us 3
+ aten::_conv_depthwise2d 0.97% 20.860us 2.94% 63.051us 21.017us 6.946us 37.04% 6.946us 2.315us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 6.946us 37.04% 6.946us 2.315us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.047us 32.25% 6.047us 2.016us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.760us 30.72% 5.760us 1.920us 3
+ Activity Buffer Request 79.94% 1.712ms 79.94% 1.712ms 1.712ms 2.016us 10.75% 2.016us 2.016us 1
+ aten::empty_strided 1.28% 27.310us 1.28% 27.310us 4.552us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 3.44% 73.575us 3.44% 73.575us 8.175us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.79% 16.901us 1.05% 22.540us 2.504us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.42% 9.019us 0.42% 9.019us 0.601us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.43% 9.230us 0.43% 9.230us 3.077us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.40% 8.580us 0.40% 8.580us 2.860us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.25% 5.250us 0.31% 6.610us 2.203us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.213ms
-Self CUDA time total: 18.720us
+Self CPU time total: 2.142ms
+Self CUDA time total: 18.753us
@@ -4095,29 +4095,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D64_S512_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 328.254us 1673.31% 328.254us 328.254us 1
- torch_eager 6.02% 146.742us 99.79% 2.431ms 2.431ms 0.000us 0.00% 21.729us 21.729us 1
- aten::to 0.25% 6.201us 87.89% 2.141ms 356.794us 0.000us 0.00% 14.048us 2.341us 6
- aten::_to_copy 0.95% 23.051us 87.64% 2.135ms 355.761us 0.000us 0.00% 14.048us 2.341us 6
- aten::copy_ 1.93% 46.899us 85.39% 2.080ms 346.662us 11.936us 60.85% 14.048us 2.341us 6
- aten::conv1d 0.28% 6.941us 4.83% 117.552us 39.184us 0.000us 0.00% 7.681us 2.560us 3
- aten::convolution 0.38% 9.320us 4.54% 110.611us 36.870us 0.000us 0.00% 7.681us 2.560us 3
- aten::_convolution 0.86% 20.861us 4.16% 101.291us 33.764us 0.000us 0.00% 7.681us 2.560us 3
- aten::_conv_depthwise2d 0.93% 22.752us 2.67% 64.991us 21.664us 7.681us 39.15% 7.681us 2.560us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.681us 39.15% 7.681us 2.560us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.208us 31.65% 6.208us 2.069us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.728us 29.20% 5.728us 1.909us 3
- Activity Buffer Request 75.50% 1.839ms 75.50% 1.839ms 1.839ms 2.112us 10.77% 2.112us 2.112us 1
- aten::empty_strided 1.29% 31.540us 1.29% 31.540us 5.257us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.87% 216.103us 8.87% 216.103us 24.011us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.70% 16.989us 0.90% 21.970us 2.441us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.35% 8.601us 0.35% 8.601us 0.573us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.43% 10.359us 0.43% 10.359us 3.453us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.40% 9.840us 0.40% 9.840us 3.280us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.22% 5.410us 0.28% 6.920us 2.307us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 315.647us 1619.78% 315.647us 315.647us 1
+ torch_eager 6.63% 154.700us 99.79% 2.328ms 2.328ms 0.000us 0.00% 21.599us 21.599us 1
+ aten::to 0.26% 6.031us 87.19% 2.034ms 338.971us 0.000us 0.00% 13.983us 2.331us 6
+ aten::_to_copy 0.93% 21.742us 86.93% 2.028ms 337.966us 0.000us 0.00% 13.983us 2.331us 6
+ aten::copy_ 2.02% 47.211us 84.82% 1.979ms 329.757us 11.871us 60.92% 13.983us 2.331us 6
+ aten::conv1d 0.25% 5.920us 4.90% 114.312us 38.104us 0.000us 0.00% 7.616us 2.539us 3
+ aten::convolution 0.39% 8.981us 4.65% 108.392us 36.131us 0.000us 0.00% 7.616us 2.539us 3
+ aten::_convolution 0.89% 20.821us 4.26% 99.411us 33.137us 0.000us 0.00% 7.616us 2.539us 3
+ aten::_conv_depthwise2d 0.89% 20.790us 2.71% 63.250us 21.083us 7.616us 39.08% 7.616us 2.539us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.616us 39.08% 7.616us 2.539us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.143us 31.52% 6.143us 2.048us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.728us 29.39% 5.728us 1.909us 3
+ Activity Buffer Request 71.63% 1.671ms 71.63% 1.671ms 1.671ms 2.112us 10.84% 2.112us 2.112us 1
+ aten::empty_strided 1.18% 27.509us 1.18% 27.509us 4.585us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 12.15% 283.514us 12.15% 283.514us 31.502us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.70% 16.421us 0.91% 21.332us 2.370us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.37% 8.571us 0.37% 8.571us 0.571us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.43% 10.140us 0.43% 10.140us 3.380us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.39% 9.150us 0.39% 9.150us 3.050us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.23% 5.450us 0.30% 7.100us 2.367us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.436ms
-Self CUDA time total: 19.617us
+Self CPU time total: 2.333ms
+Self CUDA time total: 19.487us
@@ -4127,29 +4127,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D64_S2048_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 325.374us 1318.69% 325.374us 325.374us 1
- torch_eager 6.23% 145.210us 99.78% 2.326ms 2.326ms 0.000us 0.00% 26.978us 26.978us 1
- aten::to 0.28% 6.471us 87.58% 2.041ms 340.232us 0.000us 0.00% 15.298us 2.550us 6
- aten::_to_copy 1.01% 23.559us 87.30% 2.035ms 339.154us 0.000us 0.00% 15.298us 2.550us 6
- aten::copy_ 2.04% 47.563us 85.03% 1.982ms 330.320us 12.994us 52.66% 15.298us 2.550us 6
- aten::conv1d 0.26% 6.060us 4.91% 114.341us 38.114us 0.000us 0.00% 11.680us 3.893us 3
- aten::convolution 0.40% 9.250us 4.65% 108.281us 36.094us 0.000us 0.00% 11.680us 3.893us 3
- aten::_convolution 0.89% 20.669us 4.25% 99.031us 33.010us 0.000us 0.00% 11.680us 3.893us 3
- aten::_conv_depthwise2d 0.95% 22.039us 2.73% 63.550us 21.183us 11.680us 47.34% 11.680us 3.893us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 11.680us 47.34% 11.680us 3.893us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.657us 26.98% 6.657us 2.219us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 6.337us 25.68% 6.337us 2.112us 3
- Activity Buffer Request 74.59% 1.739ms 74.59% 1.739ms 1.739ms 2.304us 9.34% 2.304us 2.304us 1
- aten::empty_strided 1.26% 29.442us 1.26% 29.442us 4.907us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 9.39% 218.802us 9.39% 218.802us 24.311us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.69% 16.041us 0.91% 21.173us 2.353us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.37% 8.602us 0.37% 8.602us 0.573us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.40% 9.341us 0.40% 9.341us 3.114us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.39% 8.990us 0.39% 8.990us 2.997us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.23% 5.290us 0.28% 6.580us 2.193us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 316.162us 1286.41% 316.162us 316.162us 1
+ torch_eager 6.08% 142.813us 99.78% 2.344ms 2.344ms 0.000us 0.00% 26.881us 26.881us 1
+ aten::to 0.26% 6.099us 87.82% 2.063ms 343.816us 0.000us 0.00% 15.329us 2.555us 6
+ aten::_to_copy 0.96% 22.501us 87.56% 2.057ms 342.799us 0.000us 0.00% 15.329us 2.555us 6
+ aten::copy_ 2.00% 46.941us 85.41% 2.006ms 334.376us 13.025us 53.00% 15.329us 2.555us 6
+ aten::conv1d 0.27% 6.279us 4.78% 112.281us 37.427us 0.000us 0.00% 11.552us 3.851us 3
+ aten::convolution 0.37% 8.701us 4.51% 106.002us 35.334us 0.000us 0.00% 11.552us 3.851us 3
+ aten::_convolution 0.88% 20.659us 4.14% 97.301us 32.434us 0.000us 0.00% 11.552us 3.851us 3
+ aten::_conv_depthwise2d 0.89% 20.940us 2.63% 61.791us 20.597us 11.552us 47.00% 11.552us 3.851us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 11.552us 47.00% 11.552us 3.851us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.689us 27.22% 6.689us 2.230us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 6.336us 25.78% 6.336us 2.112us 3
+ Activity Buffer Request 74.92% 1.760ms 74.92% 1.760ms 1.760ms 2.304us 9.37% 2.304us 2.304us 1
+ aten::empty_strided 1.19% 28.040us 1.19% 28.040us 4.673us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 9.45% 222.075us 9.45% 222.075us 24.675us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.70% 16.419us 0.91% 21.470us 2.386us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.36% 8.480us 0.36% 8.480us 0.565us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.40% 9.310us 0.40% 9.310us 3.103us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.38% 8.880us 0.38% 8.880us 2.960us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.24% 5.612us 0.29% 6.911us 2.304us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.331ms
-Self CUDA time total: 24.674us
+Self CPU time total: 2.349ms
+Self CUDA time total: 24.577us
@@ -4159,29 +4159,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D64_S2048_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 325.853us 1241.91% 325.853us 325.853us 1
- torch_eager 6.02% 142.382us 99.78% 2.359ms 2.359ms 0.000us 0.00% 28.510us 28.510us 1
- aten::to 0.27% 6.279us 87.80% 2.076ms 345.959us 0.000us 0.00% 15.262us 2.544us 6
- aten::_to_copy 0.97% 22.980us 87.54% 2.069ms 344.912us 0.000us 0.00% 15.262us 2.544us 6
- aten::copy_ 2.02% 47.672us 85.33% 2.017ms 336.189us 12.990us 49.51% 15.262us 2.544us 6
- aten::conv1d 0.27% 6.391us 4.88% 115.262us 38.421us 0.000us 0.00% 13.248us 4.416us 3
- aten::convolution 0.41% 9.629us 4.61% 108.871us 36.290us 0.000us 0.00% 13.248us 4.416us 3
- aten::_convolution 0.88% 20.800us 4.20% 99.242us 33.081us 0.000us 0.00% 13.248us 4.416us 3
- aten::_conv_depthwise2d 0.93% 21.882us 2.62% 62.041us 20.680us 13.248us 50.49% 13.248us 4.416us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 13.248us 50.49% 13.248us 4.416us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.622us 25.24% 6.622us 2.207us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 6.368us 24.27% 6.368us 2.123us 3
- Activity Buffer Request 75.21% 1.778ms 75.21% 1.778ms 1.778ms 2.272us 8.66% 2.272us 2.272us 1
- aten::empty_strided 1.24% 29.361us 1.24% 29.361us 4.893us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.97% 212.032us 8.97% 212.032us 23.559us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.75% 17.821us 0.98% 23.130us 2.570us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.37% 8.699us 0.37% 8.699us 0.580us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.38% 9.090us 0.38% 9.090us 3.030us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.44% 10.480us 0.44% 10.480us 3.493us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.24% 5.631us 0.30% 7.011us 2.337us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 315.548us 1212.95% 315.548us 315.548us 1
+ torch_eager 6.12% 137.996us 99.74% 2.248ms 2.248ms 0.000us 0.00% 28.287us 28.287us 1
+ aten::to 0.25% 5.659us 87.43% 1.970ms 328.377us 0.000us 0.00% 15.295us 2.549us 6
+ aten::_to_copy 0.93% 21.030us 87.17% 1.965ms 327.434us 0.000us 0.00% 15.295us 2.549us 6
+ aten::copy_ 2.11% 47.618us 85.03% 1.916ms 319.389us 13.023us 50.06% 15.295us 2.549us 6
+ aten::conv1d 0.25% 5.679us 5.07% 114.160us 38.053us 0.000us 0.00% 12.992us 4.331us 3
+ aten::convolution 0.38% 8.670us 4.81% 108.481us 36.160us 0.000us 0.00% 12.992us 4.331us 3
+ aten::_convolution 0.97% 21.770us 4.43% 99.811us 33.270us 0.000us 0.00% 12.992us 4.331us 3
+ aten::_conv_depthwise2d 0.95% 21.450us 2.80% 63.011us 21.004us 12.992us 49.94% 12.992us 4.331us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 12.992us 49.94% 12.992us 4.331us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.687us 25.70% 6.687us 2.229us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 6.336us 24.36% 6.336us 2.112us 3
+ Activity Buffer Request 74.56% 1.680ms 74.56% 1.680ms 1.680ms 2.272us 8.73% 2.272us 2.272us 1
+ aten::empty_strided 1.21% 27.241us 1.21% 27.241us 4.540us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 9.39% 211.596us 9.39% 211.596us 23.511us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.73% 16.539us 0.97% 21.840us 2.427us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.39% 8.730us 0.39% 8.730us 0.582us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.41% 9.190us 0.41% 9.190us 3.063us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.40% 9.090us 0.40% 9.090us 3.030us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.24% 5.411us 0.30% 6.770us 2.257us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.364ms
-Self CUDA time total: 26.238us
+Self CPU time total: 2.254ms
+Self CUDA time total: 26.015us
@@ -4191,29 +4191,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D2048_S128_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 331.328us 858.50% 331.328us 331.328us 1
- torch_eager 5.97% 146.471us 99.79% 2.446ms 2.446ms 0.000us 0.00% 41.186us 41.186us 1
- aten::conv1d 0.25% 6.210us 4.77% 116.961us 38.987us 0.000us 0.00% 22.849us 7.616us 3
- aten::convolution 0.40% 9.740us 4.52% 110.751us 36.917us 0.000us 0.00% 22.849us 7.616us 3
- aten::_convolution 0.89% 21.911us 4.12% 101.011us 33.670us 0.000us 0.00% 22.849us 7.616us 3
- aten::_conv_depthwise2d 0.92% 22.550us 2.59% 63.530us 21.177us 22.849us 59.20% 22.849us 7.616us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 22.849us 59.20% 22.849us 7.616us 3
- aten::to 0.25% 6.228us 88.01% 2.158ms 359.617us 0.000us 0.00% 18.337us 3.056us 6
- aten::_to_copy 1.00% 24.602us 87.76% 2.151ms 358.579us 0.000us 0.00% 18.337us 3.056us 6
- aten::copy_ 1.98% 48.619us 85.49% 2.096ms 349.334us 15.745us 40.80% 18.337us 3.056us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.385us 21.73% 8.385us 2.795us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.360us 19.07% 7.360us 2.453us 3
- Activity Buffer Request 75.73% 1.857ms 75.73% 1.857ms 1.857ms 2.592us 6.72% 2.592us 2.592us 1
- aten::empty_strided 1.26% 30.871us 1.26% 30.871us 5.145us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.69% 213.074us 8.69% 213.074us 23.675us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.69% 16.899us 0.91% 22.302us 2.478us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.35% 8.674us 0.35% 8.674us 0.578us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.39% 9.670us 0.39% 9.670us 3.223us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.37% 9.000us 0.37% 9.000us 3.000us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.23% 5.570us 0.28% 6.790us 2.263us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 309.088us 806.93% 309.088us 309.088us 1
+ torch_eager 6.25% 144.584us 99.79% 2.308ms 2.308ms 0.000us 0.00% 40.896us 40.896us 1
+ aten::conv1d 0.25% 5.670us 4.80% 111.051us 37.017us 0.000us 0.00% 22.560us 7.520us 3
+ aten::convolution 0.37% 8.509us 4.56% 105.381us 35.127us 0.000us 0.00% 22.560us 7.520us 3
+ aten::_convolution 0.91% 21.140us 4.19% 96.872us 32.291us 0.000us 0.00% 22.560us 7.520us 3
+ aten::_conv_depthwise2d 0.90% 20.721us 2.62% 60.611us 20.204us 22.560us 58.90% 22.560us 7.520us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 22.560us 58.90% 22.560us 7.520us 3
+ aten::to 0.25% 5.820us 87.66% 2.027ms 337.907us 0.000us 0.00% 18.336us 3.056us 6
+ aten::_to_copy 0.93% 21.482us 87.41% 2.022ms 336.937us 0.000us 0.00% 18.336us 3.056us 6
+ aten::copy_ 1.96% 45.321us 85.30% 1.973ms 328.809us 15.744us 41.10% 18.336us 3.056us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.416us 21.97% 8.416us 2.805us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.328us 19.13% 7.328us 2.443us 3
+ Activity Buffer Request 75.52% 1.747ms 75.52% 1.747ms 1.747ms 2.592us 6.77% 2.592us 2.592us 1
+ aten::empty_strided 1.18% 27.288us 1.18% 27.288us 4.548us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 8.77% 202.733us 8.77% 202.733us 22.526us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.70% 16.129us 0.91% 20.960us 2.329us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.36% 8.361us 0.36% 8.361us 0.557us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.40% 9.150us 0.40% 9.150us 3.050us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.39% 8.930us 0.39% 8.930us 2.977us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.24% 5.551us 0.30% 6.831us 2.277us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.452ms
-Self CUDA time total: 38.594us
+Self CPU time total: 2.313ms
+Self CUDA time total: 38.304us
@@ -4223,29 +4223,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D2048_S128_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 324.382us 781.00% 324.382us 324.382us 1
- torch_eager 6.15% 143.693us 99.76% 2.329ms 2.329ms 0.000us 0.00% 44.158us 44.158us 1
- aten::conv1d 0.25% 5.870us 4.90% 114.381us 38.127us 0.000us 0.00% 25.694us 8.565us 3
- aten::convolution 0.39% 9.129us 4.65% 108.511us 36.170us 0.000us 0.00% 25.694us 8.565us 3
- aten::_convolution 0.92% 21.560us 4.26% 99.382us 33.127us 0.000us 0.00% 25.694us 8.565us 3
- aten::_conv_depthwise2d 0.91% 21.251us 2.67% 62.331us 20.777us 25.694us 61.86% 25.694us 8.565us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 25.694us 61.86% 25.694us 8.565us 3
- aten::to 0.26% 6.051us 87.64% 2.046ms 341.007us 0.000us 0.00% 18.464us 3.077us 6
- aten::_to_copy 0.99% 23.033us 87.38% 2.040ms 339.999us 0.000us 0.00% 18.464us 3.077us 6
- aten::copy_ 2.09% 48.709us 85.05% 1.985ms 330.910us 15.840us 38.14% 18.464us 3.077us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.448us 20.34% 8.448us 2.816us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.392us 17.80% 7.392us 2.464us 3
- Activity Buffer Request 74.80% 1.746ms 74.80% 1.746ms 1.746ms 2.624us 6.32% 2.624us 2.624us 1
- aten::empty_strided 1.35% 31.498us 1.35% 31.498us 5.250us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 9.10% 212.334us 9.10% 212.334us 23.593us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.70% 16.311us 0.92% 21.550us 2.394us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.38% 8.780us 0.38% 8.780us 0.585us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.39% 9.170us 0.39% 9.170us 3.057us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.44% 10.170us 0.44% 10.170us 3.390us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.24% 5.530us 0.30% 6.891us 2.297us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 314.049us 763.16% 314.049us 314.049us 1
+ torch_eager 6.27% 140.693us 99.78% 2.240ms 2.240ms 0.000us 0.00% 43.775us 43.775us 1
+ aten::conv1d 0.27% 6.020us 5.00% 112.161us 37.387us 0.000us 0.00% 25.408us 8.469us 3
+ aten::convolution 0.39% 8.730us 4.73% 106.141us 35.380us 0.000us 0.00% 25.408us 8.469us 3
+ aten::_convolution 0.96% 21.462us 4.34% 97.411us 32.470us 0.000us 0.00% 25.408us 8.469us 3
+ aten::_conv_depthwise2d 0.92% 20.739us 2.73% 61.250us 20.417us 25.408us 61.74% 25.408us 8.469us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 25.408us 61.74% 25.408us 8.469us 3
+ aten::to 0.26% 5.842us 87.39% 1.962ms 327.034us 0.000us 0.00% 18.367us 3.061us 6
+ aten::_to_copy 0.95% 21.318us 87.13% 1.956ms 326.060us 0.000us 0.00% 18.367us 3.061us 6
+ aten::copy_ 2.12% 47.620us 84.95% 1.908ms 317.917us 15.743us 38.26% 18.367us 3.061us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.384us 20.37% 8.384us 2.795us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.359us 17.88% 7.359us 2.453us 3
+ Activity Buffer Request 74.98% 1.684ms 74.98% 1.684ms 1.684ms 2.624us 6.38% 2.624us 2.624us 1
+ aten::empty_strided 1.23% 27.541us 1.23% 27.541us 4.590us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 8.81% 197.723us 8.81% 197.723us 21.969us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.72% 16.172us 0.94% 21.010us 2.334us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.37% 8.238us 0.37% 8.238us 0.549us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.46% 10.261us 0.46% 10.261us 3.420us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.39% 8.800us 0.39% 8.800us 2.933us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.24% 5.440us 0.30% 6.680us 2.227us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.335ms
-Self CUDA time total: 41.534us
+Self CPU time total: 2.245ms
+Self CUDA time total: 41.151us
@@ -4255,29 +4255,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D2048_S512_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 319.038us 307.34% 319.038us 319.038us 1
- torch_eager 4.95% 115.620us 99.75% 2.329ms 2.329ms 0.000us 0.00% 109.886us 109.886us 1
- aten::conv1d 0.24% 5.500us 4.79% 111.722us 37.241us 0.000us 0.00% 71.360us 23.787us 3
- aten::convolution 0.38% 8.820us 4.55% 106.222us 35.407us 0.000us 0.00% 71.360us 23.787us 3
- aten::_convolution 0.86% 20.169us 4.17% 97.402us 32.467us 0.000us 0.00% 71.360us 23.787us 3
- aten::_conv_depthwise2d 0.88% 20.499us 2.70% 62.992us 20.997us 71.360us 68.74% 71.360us 23.787us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 71.360us 68.74% 71.360us 23.787us 3
- aten::to 0.25% 5.942us 88.99% 2.078ms 346.257us 0.000us 0.00% 38.526us 6.421us 6
- aten::_to_copy 0.97% 22.531us 88.74% 2.072ms 345.267us 0.000us 0.00% 38.526us 6.421us 6
- aten::copy_ 1.95% 45.459us 86.50% 2.019ms 336.557us 32.447us 31.26% 38.526us 6.421us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 17.791us 17.14% 17.791us 5.930us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 14.656us 14.12% 14.656us 4.885us 3
- Activity Buffer Request 76.44% 1.784ms 76.44% 1.784ms 1.784ms 6.079us 5.86% 6.079us 6.079us 1
- aten::empty_strided 1.27% 29.730us 1.27% 29.730us 4.955us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 9.13% 213.066us 9.13% 213.066us 23.674us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.66% 15.410us 0.85% 19.870us 2.208us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.33% 7.790us 0.33% 7.790us 0.519us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.44% 10.351us 0.44% 10.351us 3.450us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.36% 8.461us 0.36% 8.461us 2.820us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.23% 5.401us 0.29% 6.691us 2.230us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 321.281us 312.48% 321.281us 321.281us 1
+ torch_eager 5.86% 138.423us 99.79% 2.355ms 2.355ms 0.000us 0.00% 108.833us 108.833us 1
+ aten::conv1d 0.25% 5.850us 4.76% 112.341us 37.447us 0.000us 0.00% 70.720us 23.573us 3
+ aten::convolution 0.36% 8.520us 4.51% 106.491us 35.497us 0.000us 0.00% 70.720us 23.573us 3
+ aten::_convolution 0.88% 20.802us 4.15% 97.971us 32.657us 0.000us 0.00% 70.720us 23.573us 3
+ aten::_conv_depthwise2d 0.89% 20.951us 2.63% 62.101us 20.700us 70.720us 68.78% 70.720us 23.573us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 70.720us 68.78% 70.720us 23.573us 3
+ aten::to 0.25% 6.012us 88.03% 2.078ms 346.293us 0.000us 0.00% 38.113us 6.352us 6
+ aten::_to_copy 0.93% 22.001us 87.78% 2.072ms 345.291us 0.000us 0.00% 38.113us 6.352us 6
+ aten::copy_ 1.93% 45.601us 85.70% 2.023ms 337.137us 32.097us 31.22% 38.113us 6.352us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 17.601us 17.12% 17.601us 5.867us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 14.496us 14.10% 14.496us 4.832us 3
+ Activity Buffer Request 76.10% 1.796ms 76.10% 1.796ms 1.796ms 6.016us 5.85% 6.016us 6.016us 1
+ aten::empty_strided 1.14% 26.919us 1.14% 26.919us 4.487us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 8.62% 203.563us 8.62% 203.563us 22.618us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.74% 17.469us 0.97% 22.809us 2.534us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.37% 8.760us 0.37% 8.760us 0.584us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.40% 9.490us 0.40% 9.490us 3.163us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.39% 9.140us 0.39% 9.140us 3.047us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.23% 5.370us 0.28% 6.650us 2.217us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.335ms
-Self CUDA time total: 103.807us
+Self CPU time total: 2.360ms
+Self CUDA time total: 102.817us
@@ -4287,29 +4287,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D2048_S512_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 320.032us 281.56% 320.032us 320.032us 1
- torch_eager 4.89% 112.502us 99.77% 2.297ms 2.297ms 0.000us 0.00% 119.649us 119.649us 1
- aten::conv1d 0.24% 5.540us 4.86% 111.980us 37.327us 0.000us 0.00% 81.407us 27.136us 3
- aten::convolution 0.38% 8.839us 4.62% 106.440us 35.480us 0.000us 0.00% 81.407us 27.136us 3
- aten::_convolution 0.90% 20.821us 4.24% 97.601us 32.534us 0.000us 0.00% 81.407us 27.136us 3
- aten::_conv_depthwise2d 0.94% 21.639us 2.69% 61.990us 20.663us 81.407us 71.62% 81.407us 27.136us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 81.407us 71.62% 81.407us 27.136us 3
- aten::to 0.26% 5.912us 88.93% 2.047ms 341.211us 0.000us 0.00% 38.242us 6.374us 6
- aten::_to_copy 0.96% 22.099us 88.68% 2.041ms 340.225us 0.000us 0.00% 38.242us 6.374us 6
- aten::copy_ 2.13% 49.062us 86.51% 1.991ms 331.902us 32.257us 28.38% 38.242us 6.374us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 17.665us 15.54% 17.665us 5.888us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 14.592us 12.84% 14.592us 4.864us 3
- Activity Buffer Request 76.05% 1.751ms 76.05% 1.751ms 1.751ms 5.985us 5.27% 5.985us 5.985us 1
- aten::empty_strided 1.21% 27.841us 1.21% 27.841us 4.640us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 9.26% 213.213us 9.26% 213.213us 23.690us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.70% 16.150us 0.91% 21.061us 2.340us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.36% 8.381us 0.36% 8.381us 0.559us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.40% 9.130us 0.40% 9.130us 3.043us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.42% 9.600us 0.42% 9.600us 3.200us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.24% 5.419us 0.29% 6.669us 2.223us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 314.234us 278.42% 314.234us 314.234us 1
+ torch_eager 14.56% 110.034us 99.36% 750.943us 750.943us 0.000us 0.00% 118.847us 118.847us 1
+ aten::conv1d 0.76% 5.759us 14.83% 112.060us 37.353us 0.000us 0.00% 80.768us 26.923us 3
+ aten::convolution 1.14% 8.591us 14.06% 106.301us 35.434us 0.000us 0.00% 80.768us 26.923us 3
+ aten::_convolution 2.73% 20.639us 12.93% 97.710us 32.570us 0.000us 0.00% 80.768us 26.923us 3
+ aten::_conv_depthwise2d 2.77% 20.950us 8.22% 62.160us 20.720us 80.768us 71.56% 80.768us 26.923us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 80.768us 71.56% 80.768us 26.923us 3
+ aten::to 0.75% 5.692us 66.62% 503.480us 83.913us 0.000us 0.00% 38.079us 6.346us 6
+ aten::_to_copy 2.84% 21.440us 65.86% 497.788us 82.965us 0.000us 0.00% 38.079us 6.346us 6
+ aten::copy_ 6.27% 47.370us 59.47% 449.478us 74.913us 32.095us 28.44% 38.079us 6.346us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 17.567us 15.56% 17.567us 5.856us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 14.528us 12.87% 14.528us 4.843us 3
+ Activity Buffer Request 28.71% 217.014us 28.71% 217.014us 217.014us 5.984us 5.30% 5.984us 5.984us 1
+ aten::empty_strided 3.56% 26.870us 3.56% 26.870us 4.478us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 27.58% 208.434us 27.58% 208.434us 23.159us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.15% 16.260us 2.82% 21.342us 2.371us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.12% 8.483us 1.12% 8.483us 0.566us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.19% 8.980us 1.19% 8.980us 2.993us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.18% 8.890us 1.18% 8.890us 2.963us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.73% 5.530us 0.90% 6.840us 2.280us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.302ms
-Self CUDA time total: 113.664us
+Self CPU time total: 755.803us
+Self CUDA time total: 112.863us
@@ -4319,29 +4319,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D2048_S2048_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 4.70% 113.641us 96.03% 2.320ms 2.320ms 0.000us 0.00% 464.763us 464.763us 1
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 453.786us 106.62% 453.786us 453.786us 1
- aten::conv1d 0.23% 5.630us 4.62% 111.673us 37.224us 0.000us 0.00% 278.940us 92.980us 3
- aten::convolution 0.36% 8.651us 4.39% 106.043us 35.348us 0.000us 0.00% 278.940us 92.980us 3
- aten::_convolution 0.86% 20.739us 4.03% 97.392us 32.464us 0.000us 0.00% 278.940us 92.980us 3
- aten::_conv_depthwise2d 0.90% 21.710us 2.57% 62.062us 20.687us 278.940us 65.54% 278.940us 92.980us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 278.940us 65.54% 278.940us 92.980us 3
- aten::to 0.24% 5.880us 85.69% 2.071ms 345.102us 0.000us 0.00% 185.823us 30.970us 6
- aten::_to_copy 0.90% 21.820us 85.45% 2.065ms 344.122us 0.000us 0.00% 185.823us 30.970us 6
- aten::copy_ 1.99% 48.071us 83.40% 2.015ms 335.882us 146.655us 34.46% 185.823us 30.970us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 105.919us 24.89% 105.919us 35.306us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 40.736us 9.57% 40.736us 13.579us 3
- Activity Buffer Request 72.26% 1.746ms 72.26% 1.746ms 1.746ms 39.168us 9.20% 39.168us 39.168us 1
- aten::empty_strided 1.14% 27.621us 1.14% 27.621us 4.604us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 10.07% 243.344us 10.07% 243.344us 27.038us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.66% 15.908us 0.86% 20.760us 2.307us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.34% 8.262us 0.34% 8.262us 0.551us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.37% 8.921us 0.37% 8.921us 2.974us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.38% 9.260us 0.38% 9.260us 3.087us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.22% 5.361us 0.27% 6.641us 2.214us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 13.44% 111.290us 87.96% 728.462us 728.462us 0.000us 0.00% 465.598us 465.598us 1
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 451.198us 105.76% 451.198us 451.198us 1
+ aten::conv1d 0.71% 5.852us 13.50% 111.822us 37.274us 0.000us 0.00% 279.070us 93.023us 3
+ aten::convolution 1.02% 8.470us 12.80% 105.970us 35.323us 0.000us 0.00% 279.070us 93.023us 3
+ aten::_convolution 2.46% 20.349us 11.77% 97.500us 32.500us 0.000us 0.00% 279.070us 93.023us 3
+ aten::_conv_depthwise2d 2.50% 20.731us 7.53% 62.371us 20.790us 279.070us 65.41% 279.070us 93.023us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 279.070us 65.41% 279.070us 93.023us 3
+ aten::to 0.69% 5.693us 58.10% 481.149us 80.192us 0.000us 0.00% 186.528us 31.088us 6
+ aten::_to_copy 2.52% 20.868us 57.41% 475.456us 79.243us 0.000us 0.00% 186.528us 31.088us 6
+ aten::copy_ 5.92% 49.010us 51.66% 427.857us 71.310us 147.552us 34.59% 186.528us 31.088us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 107.296us 25.15% 107.296us 35.765us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 40.256us 9.44% 40.256us 13.419us 3
+ Activity Buffer Request 24.18% 200.274us 24.18% 200.274us 200.274us 38.976us 9.14% 38.976us 38.976us 1
+ aten::empty_strided 3.23% 26.731us 3.23% 26.731us 4.455us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 24.27% 200.993us 24.27% 200.993us 22.333us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 1.86% 15.430us 2.41% 19.970us 2.219us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.00% 8.282us 1.00% 8.282us 0.552us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.22% 10.070us 1.22% 10.070us 3.357us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.10% 9.150us 1.10% 9.150us 3.050us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.65% 5.408us 0.83% 6.890us 2.297us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.416ms
-Self CUDA time total: 425.595us
+Self CPU time total: 828.193us
+Self CUDA time total: 426.622us
@@ -4351,29 +4351,29 @@ PROFILE TRACE: torch_eager | cuda_B2_D2048_S2048_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 4.81% 115.230us 95.51% 2.289ms 2.289ms 0.000us 0.00% 473.560us 473.560us 1
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 466.268us 106.59% 466.268us 466.268us 1
- aten::conv1d 0.23% 5.540us 4.63% 111.002us 37.001us 0.000us 0.00% 298.430us 99.477us 3
- aten::convolution 0.37% 8.900us 4.40% 105.462us 35.154us 0.000us 0.00% 298.430us 99.477us 3
- aten::_convolution 0.85% 20.430us 4.03% 96.562us 32.187us 0.000us 0.00% 298.430us 99.477us 3
- aten::_conv_depthwise2d 0.86% 20.562us 2.57% 61.592us 20.531us 298.430us 68.22% 298.430us 99.477us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 298.430us 68.22% 298.430us 99.477us 3
- aten::to 0.24% 5.669us 85.05% 2.039ms 339.802us 0.000us 0.00% 175.130us 29.188us 6
- aten::_to_copy 0.96% 22.942us 84.82% 2.033ms 338.857us 0.000us 0.00% 175.130us 29.188us 6
- aten::copy_ 2.01% 48.190us 82.64% 1.981ms 330.170us 139.003us 31.78% 175.130us 29.188us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 98.430us 22.50% 98.430us 32.810us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 40.573us 9.28% 40.573us 13.524us 3
- Activity Buffer Request 72.81% 1.745ms 72.81% 1.745ms 1.745ms 36.127us 8.26% 36.127us 36.127us 1
- aten::empty_strided 1.22% 29.180us 1.22% 29.180us 4.863us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.73% 209.224us 8.73% 209.224us 23.247us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.66% 15.770us 0.87% 20.750us 2.306us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.35% 8.340us 0.35% 8.340us 0.556us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.43% 10.290us 0.43% 10.290us 3.430us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.37% 8.960us 0.37% 8.960us 2.987us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.22% 5.340us 0.28% 6.610us 2.203us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 12.75% 112.781us 87.94% 777.793us 777.793us 0.000us 0.00% 472.535us 472.535us 1
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 465.081us 106.66% 465.081us 465.081us 1
+ aten::conv1d 0.64% 5.670us 12.88% 113.921us 37.974us 0.000us 0.00% 298.235us 99.412us 3
+ aten::convolution 0.99% 8.800us 12.24% 108.251us 36.084us 0.000us 0.00% 298.235us 99.412us 3
+ aten::_convolution 2.35% 20.829us 11.24% 99.451us 33.150us 0.000us 0.00% 298.235us 99.412us 3
+ aten::_conv_depthwise2d 2.33% 20.611us 7.13% 63.071us 21.024us 298.235us 68.40% 298.235us 99.412us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 298.235us 68.40% 298.235us 99.412us 3
+ aten::to 0.69% 6.129us 59.45% 525.809us 87.635us 0.000us 0.00% 174.300us 29.050us 6
+ aten::_to_copy 2.38% 21.040us 58.76% 519.680us 86.613us 0.000us 0.00% 174.300us 29.050us 6
+ aten::copy_ 5.38% 47.582us 53.27% 471.169us 78.528us 137.789us 31.60% 174.300us 29.050us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 97.534us 22.37% 97.534us 32.511us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 40.255us 9.23% 40.255us 13.418us 3
+ Activity Buffer Request 27.57% 243.834us 27.57% 243.834us 243.834us 36.511us 8.37% 36.511us 36.511us 1
+ aten::empty_strided 3.11% 27.471us 3.11% 27.471us 4.579us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 23.03% 203.723us 23.03% 203.723us 22.636us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 1.93% 17.041us 2.45% 21.672us 2.408us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.93% 8.218us 0.93% 8.218us 0.548us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.01% 8.950us 1.01% 8.950us 2.983us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.08% 9.540us 1.08% 9.540us 3.180us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.64% 5.653us 0.81% 7.121us 2.374us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.397ms
-Self CUDA time total: 437.433us
+Self CPU time total: 884.464us
+Self CUDA time total: 436.024us
@@ -4383,29 +4383,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D64_S128_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 325.149us 1725.02% 325.149us 325.149us 1
- torch_eager 4.86% 112.628us 99.78% 2.311ms 2.311ms 0.000us 0.00% 20.769us 20.769us 1
- aten::to 0.26% 5.932us 88.67% 2.054ms 342.251us 0.000us 0.00% 13.536us 2.256us 6
- aten::_to_copy 1.00% 23.270us 88.41% 2.048ms 341.262us 0.000us 0.00% 13.536us 2.256us 6
- aten::copy_ 2.14% 49.511us 86.15% 1.995ms 332.552us 11.616us 61.63% 13.536us 2.256us 6
- aten::conv1d 0.24% 5.480us 5.19% 120.221us 40.074us 0.000us 0.00% 7.233us 2.411us 3
- aten::convolution 0.37% 8.641us 4.95% 114.741us 38.247us 0.000us 0.00% 7.233us 2.411us 3
- aten::_convolution 0.88% 20.361us 4.58% 106.100us 35.367us 0.000us 0.00% 7.233us 2.411us 3
- aten::_conv_depthwise2d 0.96% 22.180us 3.05% 70.680us 23.560us 7.233us 38.37% 7.233us 2.411us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.233us 38.37% 7.233us 2.411us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 5.920us 31.41% 5.920us 1.973us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.696us 30.22% 5.696us 1.899us 3
- Activity Buffer Request 75.90% 1.758ms 75.90% 1.758ms 1.758ms 1.920us 10.19% 1.920us 1.920us 1
- aten::empty_strided 1.25% 28.990us 1.25% 28.990us 4.832us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 9.42% 218.162us 9.42% 218.162us 24.240us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.68% 15.833us 0.90% 20.731us 2.303us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.37% 8.468us 0.37% 8.468us 0.565us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.40% 9.220us 0.40% 9.220us 3.073us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.39% 8.980us 0.39% 8.980us 2.993us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.24% 5.550us 0.30% 7.000us 2.333us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 322.331us 1707.26% 322.331us 322.331us 1
+ torch_eager 13.79% 111.240us 99.37% 801.573us 801.573us 0.000us 0.00% 20.864us 20.864us 1
+ aten::to 0.71% 5.762us 67.35% 543.311us 90.552us 0.000us 0.00% 13.696us 2.283us 6
+ aten::_to_copy 2.83% 22.852us 66.64% 537.549us 89.592us 0.000us 0.00% 13.696us 2.283us 6
+ aten::copy_ 5.86% 47.269us 60.36% 486.896us 81.149us 11.712us 62.03% 13.696us 2.283us 6
+ aten::conv1d 0.71% 5.710us 14.24% 114.861us 38.287us 0.000us 0.00% 7.168us 2.389us 3
+ aten::convolution 1.09% 8.781us 13.53% 109.151us 36.384us 0.000us 0.00% 7.168us 2.389us 3
+ aten::_convolution 2.85% 22.969us 12.44% 100.370us 33.457us 0.000us 0.00% 7.168us 2.389us 3
+ aten::_conv_depthwise2d 2.60% 20.991us 7.73% 62.361us 20.787us 7.168us 37.97% 7.168us 2.389us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.168us 37.97% 7.168us 2.389us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.016us 31.86% 6.016us 2.005us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.696us 30.17% 5.696us 1.899us 3
+ Activity Buffer Request 32.81% 264.695us 32.81% 264.695us 264.695us 1.984us 10.51% 1.984us 1.984us 1
+ aten::empty_strided 3.45% 27.801us 3.45% 27.801us 4.633us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 24.60% 198.402us 24.60% 198.402us 22.045us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.08% 16.740us 3.51% 28.341us 3.149us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.86% 15.030us 1.86% 15.030us 1.002us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.16% 9.330us 1.16% 9.330us 3.110us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.06% 8.570us 1.06% 8.570us 2.857us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.67% 5.391us 0.83% 6.660us 2.220us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.316ms
-Self CUDA time total: 18.849us
+Self CPU time total: 806.653us
+Self CUDA time total: 18.880us
@@ -4415,29 +4415,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D64_S128_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 320.511us 1636.76% 320.511us 320.511us 1
- torch_eager 5.91% 139.372us 99.79% 2.353ms 2.353ms 0.000us 0.00% 21.598us 21.598us 1
- aten::to 0.25% 6.010us 87.93% 2.073ms 345.496us 0.000us 0.00% 13.663us 2.277us 6
- aten::_to_copy 0.96% 22.549us 87.67% 2.067ms 344.494us 0.000us 0.00% 13.663us 2.277us 6
- aten::copy_ 2.09% 49.251us 85.51% 2.016ms 335.977us 11.647us 59.48% 13.663us 2.277us 6
- aten::conv1d 0.26% 6.081us 4.89% 115.321us 38.440us 0.000us 0.00% 7.935us 2.645us 3
- aten::convolution 0.40% 9.450us 4.63% 109.240us 36.413us 0.000us 0.00% 7.935us 2.645us 3
- aten::_convolution 0.90% 21.168us 4.23% 99.790us 33.263us 0.000us 0.00% 7.935us 2.645us 3
- aten::_conv_depthwise2d 0.87% 20.610us 2.67% 62.871us 20.957us 7.935us 40.52% 7.935us 2.645us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.935us 40.52% 7.935us 2.645us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 5.983us 30.55% 5.983us 1.994us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.664us 28.92% 5.664us 1.888us 3
- Activity Buffer Request 75.47% 1.779ms 75.47% 1.779ms 1.779ms 2.016us 10.30% 2.016us 2.016us 1
- aten::empty_strided 1.21% 28.551us 1.21% 28.551us 4.759us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.91% 210.105us 8.91% 210.105us 23.345us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.72% 16.961us 0.93% 21.872us 2.430us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.36% 8.422us 0.36% 8.422us 0.561us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.46% 10.910us 0.46% 10.910us 3.637us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.37% 8.650us 0.37% 8.650us 2.883us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.24% 5.579us 0.30% 6.970us 2.323us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 310.011us 1593.40% 310.011us 310.011us 1
+ torch_eager 14.19% 114.063us 99.41% 798.944us 798.944us 0.000us 0.00% 21.440us 21.440us 1
+ aten::to 0.69% 5.570us 68.28% 548.719us 91.453us 0.000us 0.00% 13.535us 2.256us 6
+ aten::_to_copy 2.58% 20.749us 67.58% 543.149us 90.525us 0.000us 0.00% 13.535us 2.256us 6
+ aten::copy_ 5.87% 47.160us 61.68% 495.718us 82.620us 11.551us 59.37% 13.535us 2.256us 6
+ aten::conv1d 0.70% 5.640us 13.72% 110.252us 36.751us 0.000us 0.00% 7.905us 2.635us 3
+ aten::convolution 1.10% 8.879us 13.02% 104.612us 34.871us 0.000us 0.00% 7.905us 2.635us 3
+ aten::_convolution 2.60% 20.881us 11.91% 95.733us 31.911us 0.000us 0.00% 7.905us 2.635us 3
+ aten::_conv_depthwise2d 2.57% 20.619us 7.48% 60.111us 20.037us 7.905us 40.63% 7.905us 2.635us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.905us 40.63% 7.905us 2.635us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 5.888us 30.26% 5.888us 1.963us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.663us 29.11% 5.663us 1.888us 3
+ Activity Buffer Request 34.15% 274.474us 34.15% 274.474us 274.474us 1.984us 10.20% 1.984us 1.984us 1
+ aten::empty_strided 3.32% 26.682us 3.32% 26.682us 4.447us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 24.32% 195.444us 24.32% 195.444us 21.716us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.12% 17.050us 2.77% 22.260us 2.473us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.06% 8.520us 1.06% 8.520us 0.568us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.13% 9.091us 1.13% 9.091us 3.030us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.12% 9.041us 1.12% 9.041us 3.014us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.64% 5.131us 0.80% 6.391us 2.130us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.358ms
-Self CUDA time total: 19.582us
+Self CPU time total: 803.684us
+Self CUDA time total: 19.456us
@@ -4447,29 +4447,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D64_S512_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 310.009us 1591.01% 310.009us 310.009us 1
- torch_eager 14.85% 113.881us 99.35% 762.102us 762.102us 0.000us 0.00% 21.693us 21.693us 1
- aten::to 0.75% 5.742us 67.36% 516.710us 86.118us 0.000us 0.00% 14.398us 2.400us 6
- aten::_to_copy 2.84% 21.798us 66.61% 510.968us 85.161us 0.000us 0.00% 14.398us 2.400us 6
- aten::copy_ 6.26% 48.021us 59.81% 458.808us 76.468us 12.190us 62.56% 14.398us 2.400us 6
- aten::conv1d 0.69% 5.290us 14.07% 107.951us 35.984us 0.000us 0.00% 7.295us 2.432us 3
- aten::convolution 1.14% 8.770us 13.38% 102.661us 34.220us 0.000us 0.00% 7.295us 2.432us 3
- aten::_convolution 2.56% 19.629us 12.24% 93.891us 31.297us 0.000us 0.00% 7.295us 2.432us 3
- aten::_conv_depthwise2d 2.72% 20.851us 7.84% 60.152us 20.051us 7.295us 37.44% 7.295us 2.432us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.295us 37.44% 7.295us 2.432us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.271us 32.18% 6.271us 2.090us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.919us 30.38% 5.919us 1.973us 3
- Activity Buffer Request 29.70% 227.833us 29.70% 227.833us 227.833us 2.208us 11.33% 2.208us 2.208us 1
- aten::empty_strided 3.96% 30.362us 3.96% 30.362us 5.060us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 26.62% 204.185us 26.62% 204.185us 22.687us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 2.01% 15.431us 2.57% 19.700us 2.189us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.98% 7.520us 0.98% 7.520us 0.501us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 1.29% 9.930us 1.29% 9.930us 3.310us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 1.06% 8.140us 1.06% 8.140us 2.713us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.67% 5.119us 0.83% 6.400us 2.133us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 314.048us 1627.61% 314.048us 314.048us 1
+ torch_eager 6.09% 141.071us 99.78% 2.312ms 2.312ms 0.000us 0.00% 21.471us 21.471us 1
+ aten::to 0.25% 5.868us 87.74% 2.033ms 338.847us 0.000us 0.00% 14.239us 2.373us 6
+ aten::_to_copy 0.91% 21.009us 87.49% 2.027ms 337.869us 0.000us 0.00% 14.239us 2.373us 6
+ aten::copy_ 2.05% 47.413us 85.30% 1.977ms 329.427us 12.063us 62.52% 14.239us 2.373us 6
+ aten::conv1d 0.24% 5.651us 4.86% 112.612us 37.537us 0.000us 0.00% 7.232us 2.411us 3
+ aten::convolution 0.38% 8.789us 4.62% 106.961us 35.654us 0.000us 0.00% 7.232us 2.411us 3
+ aten::_convolution 0.89% 20.661us 4.24% 98.172us 32.724us 0.000us 0.00% 7.232us 2.411us 3
+ aten::_conv_depthwise2d 0.90% 20.862us 2.69% 62.371us 20.790us 7.232us 37.48% 7.232us 2.411us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.232us 37.48% 7.232us 2.411us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.208us 32.17% 6.208us 2.069us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.855us 30.34% 5.855us 1.952us 3
+ Activity Buffer Request 75.73% 1.755ms 75.73% 1.755ms 1.755ms 2.176us 11.28% 2.176us 2.176us 1
+ aten::empty_strided 1.28% 29.642us 1.28% 29.642us 4.940us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 8.54% 197.852us 8.54% 197.852us 21.984us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.71% 16.502us 0.92% 21.412us 2.379us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.36% 8.400us 0.36% 8.400us 0.560us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.40% 9.250us 0.40% 9.250us 3.083us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.37% 8.629us 0.37% 8.629us 2.876us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.24% 5.530us 0.30% 6.920us 2.307us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 767.122us
-Self CUDA time total: 19.485us
+Self CPU time total: 2.317ms
+Self CUDA time total: 19.295us
@@ -4479,29 +4479,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D64_S512_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 312.058us 1547.83% 312.058us 312.058us 1
- torch_eager 19.84% 167.701us 99.34% 839.603us 839.603us 0.000us 0.00% 22.369us 22.369us 1
- aten::to 0.69% 5.791us 63.55% 537.169us 89.528us 0.000us 0.00% 14.400us 2.400us 6
- aten::_to_copy 2.59% 21.910us 62.87% 531.378us 88.563us 0.000us 0.00% 14.400us 2.400us 6
- aten::copy_ 5.79% 48.970us 56.91% 481.028us 80.171us 12.192us 60.47% 14.400us 2.400us 6
- aten::conv1d 0.65% 5.520us 13.10% 110.752us 36.917us 0.000us 0.00% 7.969us 2.656us 3
- aten::convolution 1.03% 8.700us 12.45% 105.232us 35.077us 0.000us 0.00% 7.969us 2.656us 3
- aten::_convolution 2.40% 20.311us 11.42% 96.532us 32.177us 0.000us 0.00% 7.969us 2.656us 3
- aten::_conv_depthwise2d 2.39% 20.240us 7.28% 61.521us 20.507us 7.969us 39.53% 7.969us 2.656us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.969us 39.53% 7.969us 2.656us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.272us 31.11% 6.272us 2.091us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.920us 29.36% 5.920us 1.973us 3
- Activity Buffer Request 29.19% 246.714us 29.19% 246.714us 246.714us 2.208us 10.95% 2.208us 2.208us 1
- aten::empty_strided 3.36% 28.440us 3.36% 28.440us 4.740us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 24.70% 208.775us 24.70% 208.775us 23.197us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 1.84% 15.580us 2.41% 20.350us 2.261us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.95% 8.049us 0.95% 8.049us 0.537us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 1.07% 9.050us 1.07% 9.050us 3.017us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 1.04% 8.800us 1.04% 8.800us 2.933us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.63% 5.361us 0.79% 6.650us 2.217us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 312.955us 1552.43% 312.955us 312.955us 1
+ torch_eager 14.84% 110.223us 99.32% 737.732us 737.732us 0.000us 0.00% 22.399us 22.399us 1
+ aten::to 0.81% 6.020us 65.77% 488.537us 81.423us 0.000us 0.00% 14.464us 2.411us 6
+ aten::_to_copy 2.95% 21.919us 64.96% 482.517us 80.420us 0.000us 0.00% 14.464us 2.411us 6
+ aten::copy_ 6.27% 46.582us 58.16% 432.008us 72.001us 12.224us 60.64% 14.464us 2.411us 6
+ aten::conv1d 0.73% 5.430us 15.33% 113.861us 37.954us 0.000us 0.00% 7.935us 2.645us 3
+ aten::convolution 1.23% 9.170us 14.60% 108.431us 36.144us 0.000us 0.00% 7.935us 2.645us 3
+ aten::_convolution 2.83% 21.009us 13.36% 99.261us 33.087us 0.000us 0.00% 7.935us 2.645us 3
+ aten::_conv_depthwise2d 2.83% 21.021us 8.51% 63.211us 21.070us 7.935us 39.36% 7.935us 2.645us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.935us 39.36% 7.935us 2.645us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.304us 31.27% 6.304us 2.101us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.920us 29.37% 5.920us 1.973us 3
+ Activity Buffer Request 28.04% 208.284us 28.04% 208.284us 208.284us 2.240us 11.11% 2.240us 2.240us 1
+ aten::empty_strided 3.85% 28.590us 3.85% 28.590us 4.765us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 26.94% 200.092us 26.94% 200.092us 22.232us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.19% 16.299us 2.86% 21.260us 2.362us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.14% 8.481us 1.14% 8.481us 0.565us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.35% 10.000us 1.35% 10.000us 3.333us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.24% 9.240us 1.24% 9.240us 3.080us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.74% 5.520us 0.94% 6.960us 2.320us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 845.213us
-Self CUDA time total: 20.161us
+Self CPU time total: 742.812us
+Self CUDA time total: 20.159us
@@ -4511,29 +4511,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D64_S2048_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 312.867us 859.95% 312.867us 312.867us 1
- torch_eager 14.44% 112.752us 99.36% 776.042us 776.042us 0.000us 0.00% 39.006us 39.006us 1
- aten::conv1d 0.71% 5.580us 13.99% 109.252us 36.417us 0.000us 0.00% 20.512us 6.837us 3
- aten::convolution 1.09% 8.531us 13.27% 103.672us 34.557us 0.000us 0.00% 20.512us 6.837us 3
- aten::_convolution 2.62% 20.459us 12.18% 95.141us 31.714us 0.000us 0.00% 20.512us 6.837us 3
- aten::_conv_depthwise2d 2.59% 20.222us 7.70% 60.162us 20.054us 20.512us 56.38% 20.512us 6.837us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 20.512us 56.38% 20.512us 6.837us 3
- aten::to 0.75% 5.821us 67.81% 529.608us 88.268us 0.000us 0.00% 18.494us 3.082us 6
- aten::_to_copy 2.86% 22.338us 67.06% 523.787us 87.298us 0.000us 0.00% 18.494us 3.082us 6
- aten::copy_ 6.02% 47.020us 60.45% 472.148us 78.691us 15.870us 43.62% 18.494us 3.082us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.447us 23.22% 8.447us 2.816us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.423us 20.40% 7.423us 2.474us 3
- Activity Buffer Request 30.80% 240.594us 30.80% 240.594us 240.594us 2.624us 7.21% 2.624us 2.624us 1
- aten::empty_strided 3.75% 29.301us 3.75% 29.301us 4.884us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 26.46% 206.633us 26.46% 206.633us 22.959us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 2.01% 15.720us 2.61% 20.410us 2.268us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 1.02% 7.981us 1.02% 7.981us 0.532us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 1.13% 8.841us 1.13% 8.841us 2.947us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 1.15% 9.000us 1.15% 9.000us 3.000us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.68% 5.329us 0.84% 6.560us 2.187us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 313.245us 871.70% 313.245us 313.245us 1
+ torch_eager 14.61% 110.582us 99.35% 752.253us 752.253us 0.000us 0.00% 38.527us 38.527us 1
+ aten::conv1d 0.75% 5.651us 14.90% 112.841us 37.614us 0.000us 0.00% 20.160us 6.720us 3
+ aten::convolution 1.16% 8.759us 14.16% 107.190us 35.730us 0.000us 0.00% 20.160us 6.720us 3
+ aten::_convolution 2.72% 20.612us 13.00% 98.431us 32.810us 0.000us 0.00% 20.160us 6.720us 3
+ aten::_conv_depthwise2d 2.88% 21.780us 8.26% 62.540us 20.847us 20.160us 56.10% 20.160us 6.720us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 20.160us 56.10% 20.160us 6.720us 3
+ aten::to 0.75% 5.701us 66.54% 503.830us 83.972us 0.000us 0.00% 18.367us 3.061us 6
+ aten::_to_copy 2.88% 21.780us 65.79% 498.129us 83.021us 0.000us 0.00% 18.367us 3.061us 6
+ aten::copy_ 6.12% 46.334us 59.23% 448.449us 74.742us 15.775us 43.90% 18.367us 3.061us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.415us 23.42% 8.415us 2.805us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.360us 20.48% 7.360us 2.453us 3
+ Activity Buffer Request 29.82% 225.793us 29.82% 225.793us 225.793us 2.592us 7.21% 2.592us 2.592us 1
+ aten::empty_strided 3.68% 27.900us 3.68% 27.900us 4.650us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 26.28% 198.992us 26.28% 198.992us 22.110us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.15% 16.291us 2.79% 21.110us 2.346us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.09% 8.258us 1.09% 8.258us 0.551us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.19% 9.040us 1.19% 9.040us 3.013us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.20% 9.050us 1.20% 9.050us 3.017us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.73% 5.530us 0.90% 6.810us 2.270us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 781.073us
-Self CUDA time total: 36.382us
+Self CPU time total: 757.144us
+Self CUDA time total: 35.935us
@@ -4543,29 +4543,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D64_S2048_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 353.311us 916.31% 353.311us 353.311us 1
- torch_eager 17.31% 144.171us 99.40% 827.943us 827.943us 0.000us 0.00% 41.150us 41.150us 1
- aten::conv1d 0.66% 5.470us 14.12% 117.601us 39.200us 0.000us 0.00% 22.624us 7.541us 3
- aten::convolution 1.09% 9.120us 13.46% 112.131us 37.377us 0.000us 0.00% 22.624us 7.541us 3
- aten::_convolution 2.77% 23.100us 12.37% 103.011us 34.337us 0.000us 0.00% 22.624us 7.541us 3
- aten::_conv_depthwise2d 2.63% 21.901us 7.78% 64.791us 21.597us 22.624us 58.68% 22.624us 7.541us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 22.624us 58.68% 22.624us 7.541us 3
- aten::to 0.71% 5.920us 64.88% 540.450us 90.075us 0.000us 0.00% 18.526us 3.088us 6
- aten::_to_copy 2.59% 21.613us 64.17% 534.530us 89.088us 0.000us 0.00% 18.526us 3.088us 6
- aten::copy_ 5.88% 48.990us 58.06% 483.646us 80.608us 15.934us 41.32% 18.526us 3.088us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.575us 22.24% 8.575us 2.858us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.359us 19.09% 7.359us 2.453us 3
- Activity Buffer Request 29.91% 249.164us 29.91% 249.164us 249.164us 2.592us 6.72% 2.592us 2.592us 1
- aten::empty_strided 3.51% 29.271us 3.51% 29.271us 4.879us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 25.18% 209.712us 25.18% 209.712us 23.301us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 1.99% 16.542us 2.59% 21.611us 2.401us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 1.04% 8.638us 1.04% 8.638us 0.576us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 1.16% 9.650us 1.16% 9.650us 3.217us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 1.08% 9.020us 1.08% 9.020us 3.007us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.68% 5.681us 0.85% 7.060us 2.353us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 311.265us 814.68% 311.265us 311.265us 1
+ torch_eager 15.57% 112.202us 99.29% 715.482us 715.482us 0.000us 0.00% 40.798us 40.798us 1
+ aten::conv1d 0.79% 5.710us 15.58% 112.241us 37.414us 0.000us 0.00% 22.336us 7.445us 3
+ aten::convolution 1.20% 8.631us 14.78% 106.531us 35.510us 0.000us 0.00% 22.336us 7.445us 3
+ aten::_convolution 2.96% 21.310us 13.59% 97.900us 32.633us 0.000us 0.00% 22.336us 7.445us 3
+ aten::_conv_depthwise2d 2.84% 20.470us 8.57% 61.720us 20.573us 22.336us 58.46% 22.336us 7.445us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 22.336us 58.46% 22.336us 7.445us 3
+ aten::to 0.77% 5.551us 64.63% 465.749us 77.625us 0.000us 0.00% 18.462us 3.077us 6
+ aten::_to_copy 2.88% 20.749us 63.86% 460.198us 76.700us 0.000us 0.00% 18.462us 3.077us 6
+ aten::copy_ 6.44% 46.412us 57.18% 412.048us 68.675us 15.871us 41.54% 18.462us 3.077us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.511us 22.28% 8.511us 2.837us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.360us 19.26% 7.360us 2.453us 3
+ Activity Buffer Request 26.39% 190.134us 26.39% 190.134us 190.134us 2.591us 6.78% 2.591us 2.591us 1
+ aten::empty_strided 3.80% 27.401us 3.80% 27.401us 4.567us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 27.38% 197.312us 27.38% 197.312us 21.924us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.26% 16.250us 2.97% 21.400us 2.378us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.19% 8.579us 1.19% 8.579us 0.572us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.47% 10.600us 1.47% 10.600us 3.533us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.23% 8.840us 1.23% 8.840us 2.947us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.77% 5.530us 0.95% 6.840us 2.280us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 832.973us
-Self CUDA time total: 38.558us
+Self CPU time total: 720.602us
+Self CUDA time total: 38.207us
@@ -4575,29 +4575,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D2048_S128_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 316.829us 488.45% 316.829us 316.829us 1
- torch_eager 14.19% 114.002us 99.33% 798.183us 798.183us 0.000us 0.00% 68.991us 68.991us 1
- aten::conv1d 0.68% 5.460us 13.80% 110.892us 36.964us 0.000us 0.00% 42.304us 14.101us 3
- aten::convolution 1.10% 8.859us 13.12% 105.432us 35.144us 0.000us 0.00% 42.304us 14.101us 3
- aten::_convolution 2.59% 20.821us 12.02% 96.573us 32.191us 0.000us 0.00% 42.304us 14.101us 3
- aten::_conv_depthwise2d 2.64% 21.190us 7.50% 60.251us 20.084us 42.304us 65.22% 42.304us 14.101us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 42.304us 65.22% 42.304us 14.101us 3
- aten::to 0.75% 6.059us 68.35% 549.177us 91.530us 0.000us 0.00% 26.687us 4.448us 6
- aten::_to_copy 2.76% 22.169us 67.59% 543.118us 90.520us 0.000us 0.00% 26.687us 4.448us 6
- aten::copy_ 6.74% 54.161us 61.27% 492.308us 82.051us 22.560us 34.78% 26.687us 4.448us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 12.095us 18.65% 12.095us 4.032us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 10.465us 16.13% 10.465us 3.488us 3
- Activity Buffer Request 31.75% 255.134us 31.75% 255.134us 255.134us 4.127us 6.36% 4.127us 4.127us 1
- aten::empty_strided 3.56% 28.641us 3.56% 28.641us 4.773us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 25.49% 204.843us 25.49% 204.843us 22.760us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 2.06% 16.521us 2.65% 21.322us 2.369us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 1.02% 8.171us 1.02% 8.171us 0.545us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 1.14% 9.170us 1.14% 9.170us 3.057us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 1.00% 8.061us 1.00% 8.061us 2.687us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.66% 5.330us 0.81% 6.520us 2.173us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 314.752us 491.08% 314.752us 314.752us 1
+ torch_eager 14.41% 112.890us 99.37% 778.613us 778.613us 0.000us 0.00% 68.189us 68.189us 1
+ aten::conv1d 0.75% 5.840us 14.26% 111.761us 37.254us 0.000us 0.00% 41.759us 13.920us 3
+ aten::convolution 1.10% 8.641us 13.52% 105.921us 35.307us 0.000us 0.00% 41.759us 13.920us 3
+ aten::_convolution 2.66% 20.830us 12.41% 97.280us 32.427us 0.000us 0.00% 41.759us 13.920us 3
+ aten::_conv_depthwise2d 2.71% 21.201us 7.84% 61.431us 20.477us 41.759us 65.15% 41.759us 13.920us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 41.759us 65.15% 41.759us 13.920us 3
+ aten::to 0.73% 5.701us 67.43% 528.381us 88.063us 0.000us 0.00% 26.430us 4.405us 6
+ aten::_to_copy 2.70% 21.142us 66.70% 522.680us 87.113us 0.000us 0.00% 26.430us 4.405us 6
+ aten::copy_ 6.06% 47.459us 59.50% 466.248us 77.708us 22.335us 34.85% 26.430us 4.405us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 11.903us 18.57% 11.903us 3.968us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 10.432us 16.28% 10.432us 3.477us 3
+ Activity Buffer Request 30.62% 239.904us 30.62% 239.904us 239.904us 4.095us 6.39% 4.095us 4.095us 1
+ aten::empty_strided 4.50% 35.290us 4.50% 35.290us 5.882us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 25.58% 200.475us 25.58% 200.475us 22.275us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.10% 16.441us 2.75% 21.561us 2.396us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.10% 8.620us 1.10% 8.620us 0.575us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.18% 9.220us 1.18% 9.220us 3.073us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.20% 9.420us 1.20% 9.420us 3.140us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.71% 5.569us 0.87% 6.849us 2.283us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 803.533us
-Self CUDA time total: 64.864us
+Self CPU time total: 783.583us
+Self CUDA time total: 64.094us
@@ -4607,29 +4607,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D2048_S128_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 328.383us 466.25% 328.383us 328.383us 1
- torch_eager 5.82% 138.672us 99.78% 2.376ms 2.376ms 0.000us 0.00% 74.527us 74.527us 1
- aten::conv1d 0.24% 5.689us 4.87% 115.970us 38.657us 0.000us 0.00% 47.969us 15.990us 3
- aten::convolution 0.43% 10.191us 4.63% 110.281us 36.760us 0.000us 0.00% 47.969us 15.990us 3
- aten::_convolution 0.91% 21.579us 4.20% 100.090us 33.363us 0.000us 0.00% 47.969us 15.990us 3
- aten::_conv_depthwise2d 0.87% 20.670us 2.63% 62.670us 20.890us 47.969us 68.11% 47.969us 15.990us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 47.969us 68.11% 47.969us 15.990us 3
- aten::to 0.27% 6.430us 88.04% 2.097ms 349.464us 0.000us 0.00% 26.558us 4.426us 6
- aten::_to_copy 0.99% 23.642us 87.77% 2.090ms 348.392us 0.000us 0.00% 26.558us 4.426us 6
- aten::copy_ 2.06% 49.120us 85.54% 2.037ms 339.525us 22.462us 31.89% 26.558us 4.426us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 11.999us 17.04% 11.999us 4.000us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 10.463us 14.86% 10.463us 3.488us 3
- Activity Buffer Request 75.66% 1.802ms 75.66% 1.802ms 1.802ms 4.096us 5.82% 4.096us 4.096us 1
- aten::empty_strided 1.24% 29.560us 1.24% 29.560us 4.927us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.75% 208.373us 8.75% 208.373us 23.153us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.70% 16.782us 0.92% 21.972us 2.441us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.36% 8.520us 0.36% 8.520us 0.568us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.38% 9.160us 0.38% 9.160us 3.053us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.44% 10.580us 0.44% 10.580us 3.527us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.24% 5.730us 0.29% 7.020us 2.340us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 341.050us 488.00% 341.050us 341.050us 1
+ torch_eager 14.90% 121.512us 99.29% 809.864us 809.864us 0.000us 0.00% 73.952us 73.952us 1
+ aten::conv1d 0.74% 6.030us 14.63% 119.332us 39.777us 0.000us 0.00% 47.456us 15.819us 3
+ aten::convolution 1.12% 9.160us 13.89% 113.302us 37.767us 0.000us 0.00% 47.456us 15.819us 3
+ aten::_convolution 2.63% 21.441us 12.77% 104.142us 34.714us 0.000us 0.00% 47.456us 15.819us 3
+ aten::_conv_depthwise2d 2.71% 22.133us 8.21% 66.923us 22.308us 47.456us 67.90% 47.456us 15.819us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 47.456us 67.90% 47.456us 15.819us 3
+ aten::to 0.75% 6.151us 66.55% 542.789us 90.465us 0.000us 0.00% 26.496us 4.416us 6
+ aten::_to_copy 2.90% 23.661us 65.79% 536.638us 89.440us 0.000us 0.00% 26.496us 4.416us 6
+ aten::copy_ 6.09% 49.711us 59.32% 483.818us 80.636us 22.432us 32.10% 26.496us 4.416us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 11.904us 17.03% 11.904us 3.968us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 10.528us 15.06% 10.528us 3.509us 3
+ Activity Buffer Request 30.67% 250.175us 30.67% 250.175us 250.175us 4.064us 5.82% 4.064us 4.064us 1
+ aten::empty_strided 3.58% 29.159us 3.58% 29.159us 4.860us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 25.38% 207.002us 25.38% 207.002us 23.000us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.12% 17.271us 2.72% 22.171us 2.463us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.06% 8.630us 1.06% 8.630us 0.575us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.48% 12.040us 1.48% 12.040us 4.013us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.19% 9.680us 1.19% 9.680us 3.227us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.69% 5.638us 0.87% 7.098us 2.366us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.382ms
-Self CUDA time total: 70.431us
+Self CPU time total: 815.634us
+Self CUDA time total: 69.888us
@@ -4639,29 +4639,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D2048_S512_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 336.351us 179.68% 336.351us 336.351us 1
- torch_eager 5.85% 142.571us 99.79% 2.430ms 2.430ms 0.000us 0.00% 197.311us 197.311us 1
- aten::conv1d 0.28% 6.741us 4.71% 114.731us 38.244us 0.000us 0.00% 134.368us 44.789us 3
- aten::convolution 0.38% 9.350us 4.43% 107.990us 35.997us 0.000us 0.00% 134.368us 44.789us 3
- aten::_convolution 0.88% 21.488us 4.05% 98.640us 32.880us 0.000us 0.00% 134.368us 44.789us 3
- aten::_conv_depthwise2d 0.83% 20.301us 2.51% 61.091us 20.364us 134.368us 71.78% 134.368us 44.789us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 134.368us 71.78% 134.368us 44.789us 3
- aten::to 0.26% 6.379us 88.22% 2.148ms 358.072us 0.000us 0.00% 62.943us 10.491us 6
- aten::_to_copy 0.93% 22.632us 87.96% 2.142ms 357.009us 0.000us 0.00% 62.943us 10.491us 6
- aten::copy_ 2.03% 49.489us 85.76% 2.089ms 348.110us 52.831us 28.22% 62.943us 10.491us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 29.727us 15.88% 29.727us 9.909us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 23.104us 12.34% 23.104us 7.701us 3
- Activity Buffer Request 76.11% 1.853ms 76.11% 1.853ms 1.853ms 10.112us 5.40% 10.112us 10.112us 1
- aten::empty_strided 1.26% 30.760us 1.26% 30.760us 5.127us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 8.55% 208.274us 8.55% 208.274us 23.142us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.71% 17.184us 0.91% 22.223us 2.469us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.34% 8.338us 0.34% 8.338us 0.556us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.38% 9.180us 0.38% 9.180us 3.060us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.37% 9.020us 0.37% 9.020us 3.007us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.22% 5.460us 0.27% 6.690us 2.230us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 333.665us 179.68% 333.665us 333.665us 1
+ torch_eager 6.31% 142.852us 99.78% 2.260ms 2.260ms 0.000us 0.00% 195.711us 195.711us 1
+ aten::conv1d 0.27% 6.061us 5.13% 116.202us 38.734us 0.000us 0.00% 133.407us 44.469us 3
+ aten::convolution 0.39% 8.731us 4.86% 110.141us 36.714us 0.000us 0.00% 133.407us 44.469us 3
+ aten::_convolution 0.93% 21.019us 4.48% 101.410us 33.803us 0.000us 0.00% 133.407us 44.469us 3
+ aten::_conv_depthwise2d 0.95% 21.630us 2.86% 64.671us 21.557us 133.407us 71.84% 133.407us 44.469us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 133.407us 71.84% 133.407us 44.469us 3
+ aten::to 0.26% 5.950us 87.17% 1.974ms 329.070us 0.000us 0.00% 62.304us 10.384us 6
+ aten::_to_copy 0.93% 21.080us 86.91% 1.968ms 328.079us 0.000us 0.00% 62.304us 10.384us 6
+ aten::copy_ 2.06% 46.600us 84.72% 1.919ms 319.815us 52.288us 28.16% 62.304us 10.384us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 29.503us 15.89% 29.503us 9.834us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 22.785us 12.27% 22.785us 7.595us 3
+ Activity Buffer Request 74.89% 1.696ms 74.89% 1.696ms 1.696ms 10.016us 5.39% 10.016us 10.016us 1
+ aten::empty_strided 1.26% 28.502us 1.26% 28.502us 4.750us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 8.73% 197.643us 8.73% 197.643us 21.960us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.77% 17.390us 1.01% 22.821us 2.536us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.41% 9.201us 0.41% 9.201us 0.613us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.51% 11.561us 0.51% 11.561us 3.854us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.43% 9.850us 0.43% 9.850us 3.283us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.25% 5.550us 0.31% 7.130us 2.377us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 2.435ms
-Self CUDA time total: 187.199us
+Self CPU time total: 2.265ms
+Self CUDA time total: 185.695us
@@ -4671,29 +4671,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D2048_S512_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 335.323us 159.21% 335.323us 335.323us 1
- torch_eager 14.44% 115.471us 99.40% 794.842us 794.842us 0.000us 0.00% 223.709us 223.709us 1
- aten::conv1d 0.70% 5.561us 13.80% 110.362us 36.787us 0.000us 0.00% 154.845us 51.615us 3
- aten::convolution 1.15% 9.189us 13.11% 104.801us 34.934us 0.000us 0.00% 154.845us 51.615us 3
- aten::_convolution 2.52% 20.182us 11.96% 95.612us 31.871us 0.000us 0.00% 154.845us 51.615us 3
- aten::_conv_depthwise2d 2.51% 20.101us 7.60% 60.741us 20.247us 154.845us 73.52% 154.845us 51.615us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 154.845us 73.52% 154.845us 51.615us 3
- aten::to 0.72% 5.750us 68.18% 545.179us 90.863us 0.000us 0.00% 68.864us 11.477us 6
- aten::_to_copy 2.77% 22.130us 67.46% 539.429us 89.905us 0.000us 0.00% 68.864us 11.477us 6
- aten::copy_ 5.86% 46.830us 60.79% 486.078us 81.013us 55.776us 26.48% 68.864us 11.477us 6
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 32.416us 15.39% 32.416us 10.805us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 23.360us 11.09% 23.360us 7.787us 3
- Activity Buffer Request 31.66% 253.204us 31.66% 253.204us 253.204us 13.088us 6.21% 13.088us 13.088us 1
- aten::empty_strided 3.90% 31.221us 3.90% 31.221us 5.203us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 26.02% 208.054us 26.02% 208.054us 23.117us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 1.93% 15.399us 2.47% 19.760us 2.196us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.98% 7.800us 0.98% 7.800us 0.520us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 1.23% 9.810us 1.23% 9.810us 3.270us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 1.10% 8.820us 1.10% 8.820us 2.940us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.69% 5.519us 0.86% 6.899us 2.300us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 336.127us 160.20% 336.127us 336.127us 1
+ torch_eager 15.02% 112.950us 99.30% 746.873us 746.873us 0.000us 0.00% 223.453us 223.453us 1
+ aten::conv1d 0.82% 6.200us 15.13% 113.793us 37.931us 0.000us 0.00% 154.078us 51.359us 3
+ aten::convolution 1.21% 9.111us 14.30% 107.593us 35.864us 0.000us 0.00% 154.078us 51.359us 3
+ aten::_convolution 2.76% 20.789us 13.09% 98.482us 32.827us 0.000us 0.00% 154.078us 51.359us 3
+ aten::_conv_depthwise2d 2.85% 21.433us 8.36% 62.872us 20.957us 154.078us 73.43% 154.078us 51.359us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 154.078us 73.43% 154.078us 51.359us 3
+ aten::to 0.77% 5.760us 65.90% 495.650us 82.608us 0.000us 0.00% 69.375us 11.562us 6
+ aten::_to_copy 2.76% 20.790us 65.13% 489.890us 81.648us 0.000us 0.00% 69.375us 11.562us 6
+ aten::copy_ 6.23% 46.852us 58.63% 441.010us 73.502us 55.743us 26.57% 69.375us 11.562us 6
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 32.736us 15.60% 32.736us 10.912us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 23.007us 10.97% 23.007us 7.669us 3
+ Activity Buffer Request 28.60% 215.113us 28.60% 215.113us 215.113us 13.632us 6.50% 13.632us 13.632us 1
+ aten::empty_strided 3.73% 28.090us 3.73% 28.090us 4.682us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 26.86% 202.035us 26.86% 202.035us 22.448us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 2.10% 15.770us 2.68% 20.191us 2.243us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 1.08% 8.131us 1.08% 8.131us 0.542us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 1.19% 8.950us 1.19% 8.950us 2.983us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 1.26% 9.499us 1.26% 9.499us 3.166us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.71% 5.360us 0.95% 7.120us 2.373us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 799.662us
-Self CUDA time total: 210.621us
+Self CPU time total: 752.153us
+Self CUDA time total: 209.821us
@@ -4703,28 +4703,28 @@ PROFILE TRACE: torch_eager | cuda_B4_D2048_S2048_W2
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 6.62% 120.362us 52.56% 956.135us 956.135us 0.000us 0.00% 1.509ms 1.509ms 1
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 1.411ms 100.41% 1.411ms 1.411ms 1
- aten::to 0.34% 6.140us 38.13% 693.750us 115.625us 0.000us 0.00% 815.515us 135.919us 6
- aten::_to_copy 1.53% 27.810us 37.80% 687.610us 114.602us 0.000us 0.00% 815.515us 135.919us 6
- aten::copy_ 2.83% 51.570us 25.68% 467.247us 77.874us 711.740us 50.66% 815.515us 135.919us 6
- aten::conv1d 0.32% 5.781us 6.36% 115.702us 38.567us 0.000us 0.00% 693.278us 231.093us 3
- aten::convolution 0.51% 9.289us 6.04% 109.921us 36.640us 0.000us 0.00% 693.278us 231.093us 3
- aten::_convolution 1.19% 21.630us 5.53% 100.632us 33.544us 0.000us 0.00% 693.278us 231.093us 3
- aten::_conv_depthwise2d 1.16% 21.108us 3.52% 63.951us 21.317us 693.278us 49.34% 693.278us 231.093us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 693.278us 49.34% 693.278us 231.093us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 405.439us 28.86% 405.439us 135.146us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 306.301us 21.80% 306.301us 102.100us 3
- Activity Buffer Request 12.14% 220.924us 12.14% 220.924us 220.924us 103.775us 7.39% 103.775us 103.775us 1
- aten::empty_strided 1.98% 36.051us 10.58% 192.553us 32.092us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 12.05% 219.204us 12.05% 219.204us 24.356us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.93% 16.940us 1.22% 22.200us 2.467us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.48% 8.651us 0.48% 8.651us 0.577us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.51% 9.201us 0.51% 9.201us 3.067us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.51% 9.191us 0.51% 9.191us 3.064us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.31% 5.621us 0.38% 6.871us 2.290us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 6.57% 117.922us 51.18% 918.465us 918.465us 0.000us 0.00% 1.509ms 1.509ms 1
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 1.411ms 100.40% 1.411ms 1.411ms 1
+ aten::to 0.34% 6.101us 36.65% 657.591us 109.598us 0.000us 0.00% 815.747us 135.958us 6
+ aten::_to_copy 1.52% 27.229us 36.31% 651.490us 108.582us 0.000us 0.00% 815.747us 135.958us 6
+ aten::copy_ 2.73% 48.941us 24.59% 441.187us 73.531us 712.610us 50.71% 815.747us 135.958us 6
+ aten::conv1d 0.34% 6.120us 6.48% 116.312us 38.771us 0.000us 0.00% 692.768us 230.923us 3
+ aten::convolution 0.51% 9.129us 6.14% 110.192us 36.731us 0.000us 0.00% 692.768us 230.923us 3
+ aten::_convolution 1.19% 21.333us 5.63% 101.063us 33.688us 0.000us 0.00% 692.768us 230.923us 3
+ aten::_conv_depthwise2d 1.24% 22.308us 3.59% 64.481us 21.494us 692.768us 49.29% 692.768us 230.923us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 692.768us 49.29% 692.768us 230.923us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 407.041us 28.96% 407.041us 135.680us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 305.569us 21.74% 305.569us 101.856us 3
+ Activity Buffer Request 11.66% 209.253us 11.66% 209.253us 209.253us 103.137us 7.34% 103.137us 103.137us 1
+ aten::empty_strided 1.92% 34.411us 10.20% 183.074us 30.512us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 11.50% 206.424us 11.50% 206.424us 22.936us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.94% 16.930us 1.24% 22.220us 2.469us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.50% 8.941us 0.50% 8.941us 0.596us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.52% 9.381us 0.52% 9.381us 3.127us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.52% 9.361us 0.52% 9.361us 3.120us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.31% 5.508us 0.38% 6.869us 2.290us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 1.819ms
+Self CPU time total: 1.794ms
Self CUDA time total: 1.405ms
@@ -4735,29 +4735,29 @@ PROFILE TRACE: torch_eager | cuda_B4_D2048_S2048_W4
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Name Self CPU % Self CPU CPU total % CPU total CPU time avg Self CUDA Self CUDA % CUDA total CUDA time avg # of Calls
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
- torch_eager 6.07% 112.213us 42.26% 781.792us 781.792us 0.000us 0.00% 1.498ms 1.498ms 1
- torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 1.428ms 100.39% 1.428ms 1.428ms 1
- aten::to 0.33% 6.130us 28.74% 531.749us 88.625us 0.000us 0.00% 757.569us 126.261us 6
- aten::_to_copy 1.23% 22.780us 28.41% 525.619us 87.603us 0.000us 0.00% 757.569us 126.261us 6
- aten::copy_ 2.64% 48.852us 25.56% 472.969us 78.828us 682.049us 47.95% 757.569us 126.261us 6
- aten::conv1d 0.33% 6.130us 6.13% 113.361us 37.787us 0.000us 0.00% 740.449us 246.816us 3
- aten::convolution 0.48% 8.889us 5.80% 107.231us 35.744us 0.000us 0.00% 740.449us 246.816us 3
- aten::_convolution 1.13% 20.931us 5.32% 98.342us 32.781us 0.000us 0.00% 740.449us 246.816us 3
- aten::_conv_depthwise2d 1.15% 21.330us 3.38% 62.491us 20.830us 740.449us 52.05% 740.449us 246.816us 3
-void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 740.449us 52.05% 740.449us 246.816us 3
-void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 397.857us 27.97% 397.857us 132.619us 3
-void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 284.192us 19.98% 284.192us 94.731us 3
- Activity Buffer Request 12.95% 239.644us 12.95% 239.644us 239.644us 75.520us 5.31% 75.520us 75.520us 1
- aten::empty_strided 1.61% 29.870us 1.61% 29.870us 4.978us 0.000us 0.00% 0.000us 0.000us 6
- cudaLaunchKernel 11.17% 206.574us 11.17% 206.574us 22.953us 0.000us 0.00% 0.000us 0.000us 9
- aten::unsqueeze 0.85% 15.779us 1.12% 20.809us 2.312us 0.000us 0.00% 0.000us 0.000us 9
- aten::as_strided 0.45% 8.409us 0.45% 8.409us 0.561us 0.000us 0.00% 0.000us 0.000us 15
- aten::empty 0.49% 9.120us 0.49% 9.120us 3.040us 0.000us 0.00% 0.000us 0.000us 3
- aten::resize_ 0.54% 9.940us 0.54% 9.940us 3.313us 0.000us 0.00% 0.000us 0.000us 3
- aten::squeeze 0.29% 5.381us 0.36% 6.700us 2.233us 0.000us 0.00% 0.000us 0.000us 3
+ torch_eager 6.19% 116.332us 43.22% 812.174us 812.174us 0.000us 0.00% 1.498ms 1.498ms 1
+ torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 1.429ms 100.40% 1.429ms 1.429ms 1
+ aten::to 0.31% 5.839us 29.49% 554.079us 92.346us 0.000us 0.00% 756.667us 126.111us 6
+ aten::_to_copy 1.13% 21.183us 29.18% 548.240us 91.373us 0.000us 0.00% 756.667us 126.111us 6
+ aten::copy_ 2.52% 47.321us 26.61% 500.037us 83.340us 681.756us 47.91% 756.667us 126.111us 6
+ aten::conv1d 0.30% 5.691us 6.12% 114.963us 38.321us 0.000us 0.00% 741.276us 247.092us 3
+ aten::convolution 0.46% 8.681us 5.82% 109.272us 36.424us 0.000us 0.00% 741.276us 247.092us 3
+ aten::_convolution 1.12% 21.121us 5.35% 100.591us 33.530us 0.000us 0.00% 741.276us 247.092us 3
+ aten::_conv_depthwise2d 1.12% 21.039us 3.43% 64.490us 21.497us 741.276us 52.09% 741.276us 247.092us 3
+void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 741.276us 52.09% 741.276us 247.092us 3
+void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 397.598us 27.94% 397.598us 132.533us 3
+void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 284.158us 19.97% 284.158us 94.719us 3
+ Activity Buffer Request 11.11% 208.773us 11.11% 208.773us 208.773us 74.911us 5.26% 74.911us 74.911us 1
+ aten::empty_strided 1.44% 27.020us 1.44% 27.020us 4.503us 0.000us 0.00% 0.000us 0.000us 6
+ cudaLaunchKernel 14.26% 268.023us 14.26% 268.023us 29.780us 0.000us 0.00% 0.000us 0.000us 9
+ aten::unsqueeze 0.93% 17.452us 1.22% 23.001us 2.556us 0.000us 0.00% 0.000us 0.000us 9
+ aten::as_strided 0.48% 8.988us 0.48% 8.988us 0.599us 0.000us 0.00% 0.000us 0.000us 15
+ aten::empty 0.52% 9.701us 0.52% 9.701us 3.234us 0.000us 0.00% 0.000us 0.000us 3
+ aten::resize_ 0.51% 9.670us 0.51% 9.670us 3.223us 0.000us 0.00% 0.000us 0.000us 3
+ aten::squeeze 0.29% 5.400us 0.35% 6.640us 2.213us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
-Self CPU time total: 1.850ms
-Self CUDA time total: 1.422ms
+Self CPU time total: 1.879ms
+Self CUDA time total: 1.423ms
impl wl p50(ms) ok