Running causal_conv1d benchmark on cuda with 24 workloads.
======================================================================
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 430.557us 2231.44% 430.557us 430.557us 1
torch_eager 10.26% 219.304us 99.68% 2.131ms 2.131ms 0.000us 0.00% 21.630us 21.630us 1
aten::to 0.52% 11.162us 80.31% 1.717ms 286.188us 0.000us 0.00% 14.269us 2.378us 6
aten::_to_copy 1.64% 35.099us 79.78% 1.706ms 284.328us 0.000us 0.00% 14.269us 2.378us 6
aten::copy_ 2.90% 61.950us 75.45% 1.613ms 268.888us 11.934us 61.85% 14.269us 2.378us 6
aten::conv1d 0.34% 7.309us 7.15% 152.793us 50.931us 0.000us 0.00% 7.361us 2.454us 3
aten::convolution 0.74% 15.802us 6.80% 145.484us 48.495us 0.000us 0.00% 7.361us 2.454us 3
aten::_convolution 1.46% 31.300us 6.06% 129.682us 43.227us 0.000us 0.00% 7.361us 2.454us 3
aten::_conv_depthwise2d 1.58% 33.771us 3.83% 81.842us 27.281us 7.361us 38.15% 7.361us 2.454us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.361us 38.15% 7.361us 2.454us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.271us 32.50% 6.271us 2.090us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.663us 29.35% 5.663us 1.888us 3
Activity Buffer Request 69.43% 1.484ms 69.43% 1.484ms 1.484ms 2.335us 12.10% 2.335us 2.335us 1
aten::empty_strided 2.69% 57.542us 2.69% 57.542us 9.590us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 4.29% 91.753us 4.29% 91.753us 10.195us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 1.26% 26.879us 1.60% 34.300us 3.811us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.54% 11.581us 0.54% 11.581us 0.772us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.53% 11.370us 0.53% 11.370us 3.790us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.55% 11.861us 0.55% 11.861us 3.954us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.32% 6.790us 0.37% 8.000us 2.667us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.138ms
Self CUDA time total: 19.295us
======================================================================
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 335.581us 1716.35% 335.581us 335.581us 1
torch_eager 7.79% 148.313us 99.72% 1.898ms 1.898ms 0.000us 0.00% 21.664us 21.664us 1
aten::to 0.37% 7.132us 84.25% 1.604ms 267.308us 0.000us 0.00% 13.760us 2.293us 6
aten::_to_copy 1.28% 24.280us 83.88% 1.597ms 266.119us 0.000us 0.00% 13.760us 2.293us 6
aten::copy_ 2.64% 50.321us 81.08% 1.543ms 257.239us 11.648us 59.57% 13.760us 2.293us 6
aten::conv1d 0.32% 6.130us 6.26% 119.243us 39.748us 0.000us 0.00% 7.904us 2.635us 3
aten::convolution 0.58% 11.131us 5.94% 113.113us 37.704us 0.000us 0.00% 7.904us 2.635us 3
aten::_convolution 1.18% 22.459us 5.36% 101.982us 33.994us 0.000us 0.00% 7.904us 2.635us 3
aten::_conv_depthwise2d 1.08% 20.592us 3.32% 63.132us 21.044us 7.904us 40.43% 7.904us 2.635us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.904us 40.43% 7.904us 2.635us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.080us 31.10% 6.080us 2.027us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.568us 28.48% 5.568us 1.856us 3
Activity Buffer Request 75.89% 1.445ms 75.89% 1.445ms 1.445ms 2.112us 10.80% 2.112us 2.112us 1
aten::empty_strided 1.52% 29.001us 1.52% 29.001us 4.834us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 3.69% 70.220us 3.69% 70.220us 7.802us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.92% 17.542us 1.22% 23.251us 2.583us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.48% 9.139us 0.48% 9.139us 0.609us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.58% 11.060us 0.58% 11.060us 3.687us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.51% 9.700us 0.51% 9.700us 3.233us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.31% 5.860us 0.37% 7.020us 2.340us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 1.904ms
Self CUDA time total: 19.552us
======================================================================
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 343.036us 1844.97% 343.036us 343.036us 1
torch_eager 7.68% 146.161us 99.72% 1.897ms 1.897ms 0.000us 0.00% 20.481us 20.481us 1
aten::to 0.37% 6.953us 83.90% 1.596ms 266.066us 0.000us 0.00% 13.536us 2.256us 6
aten::_to_copy 1.25% 23.842us 83.53% 1.589ms 264.907us 0.000us 0.00% 13.536us 2.256us 6
aten::copy_ 2.67% 50.789us 80.65% 1.535ms 255.762us 11.648us 62.65% 13.536us 2.256us 6
aten::conv1d 0.33% 6.290us 6.66% 126.782us 42.261us 0.000us 0.00% 6.945us 2.315us 3
aten::convolution 0.51% 9.650us 6.33% 120.492us 40.164us 0.000us 0.00% 6.945us 2.315us 3
aten::_convolution 1.22% 23.120us 5.83% 110.842us 36.947us 0.000us 0.00% 6.945us 2.315us 3
aten::_conv_depthwise2d 1.44% 27.490us 3.78% 71.832us 23.944us 6.945us 37.35% 6.945us 2.315us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 6.945us 37.35% 6.945us 2.315us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 5.952us 32.01% 5.952us 1.984us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.696us 30.64% 5.696us 1.899us 3
Activity Buffer Request 75.34% 1.433ms 75.34% 1.433ms 1.433ms 1.888us 10.15% 1.888us 1.888us 1
aten::empty_strided 1.63% 31.030us 1.63% 31.030us 5.172us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 3.99% 75.911us 3.99% 75.911us 8.435us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.90% 17.191us 1.18% 22.431us 2.492us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.46% 8.831us 0.46% 8.831us 0.589us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.55% 10.511us 0.55% 10.511us 3.504us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.43% 8.261us 0.43% 8.261us 2.754us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.31% 5.930us 0.38% 7.310us 2.437us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 1.903ms
Self CUDA time total: 18.593us
======================================================================
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 329.535us 1679.93% 329.535us 329.535us 1
torch_eager 6.84% 137.444us 99.71% 2.005ms 2.005ms 0.000us 0.00% 21.760us 21.760us 1
aten::to 0.33% 6.580us 85.73% 1.724ms 287.253us 0.000us 0.00% 14.048us 2.341us 6
aten::_to_copy 1.15% 23.069us 85.41% 1.717ms 286.156us 0.000us 0.00% 14.048us 2.341us 6
aten::copy_ 2.51% 50.362us 82.79% 1.664ms 277.390us 11.904us 60.69% 14.048us 2.341us 6
aten::conv1d 0.28% 5.589us 5.76% 115.862us 38.621us 0.000us 0.00% 7.712us 2.571us 3
aten::convolution 0.52% 10.381us 5.49% 110.273us 36.758us 0.000us 0.00% 7.712us 2.571us 3
aten::_convolution 1.13% 22.651us 4.97% 99.892us 33.297us 0.000us 0.00% 7.712us 2.571us 3
aten::_conv_depthwise2d 1.01% 20.392us 3.08% 61.981us 20.660us 7.712us 39.31% 7.712us 2.571us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.712us 39.31% 7.712us 2.571us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.176us 31.48% 6.176us 2.059us 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 70.54% 1.418ms 70.54% 1.418ms 1.418ms 2.144us 10.93% 2.144us 2.144us 1
aten::empty_strided 1.47% 29.531us 1.47% 29.531us 4.922us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.85% 218.024us 10.85% 218.024us 24.225us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.87% 17.580us 1.13% 22.650us 2.517us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.42% 8.370us 0.42% 8.370us 0.558us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.48% 9.699us 0.48% 9.699us 3.233us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.49% 9.760us 0.49% 9.760us 3.253us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.28% 5.670us 0.34% 6.790us 2.263us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.010ms
Self CUDA time total: 19.616us
======================================================================
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 344.351us 1401.22% 344.351us 344.351us 1
torch_eager 7.42% 151.244us 99.76% 2.034ms 2.034ms 0.000us 0.00% 26.847us 26.847us 1
aten::to 0.33% 6.730us 85.23% 1.738ms 289.583us 0.000us 0.00% 15.264us 2.544us 6
aten::_to_copy 1.15% 23.491us 84.90% 1.731ms 288.462us 0.000us 0.00% 15.264us 2.544us 6
aten::copy_ 2.84% 57.871us 82.24% 1.677ms 279.428us 12.992us 52.87% 15.264us 2.544us 6
aten::conv1d 0.31% 6.410us 5.76% 117.443us 39.148us 0.000us 0.00% 11.583us 3.861us 3
aten::convolution 0.49% 10.031us 5.45% 111.033us 37.011us 0.000us 0.00% 11.583us 3.861us 3
aten::_convolution 1.08% 22.081us 4.95% 101.002us 33.667us 0.000us 0.00% 11.583us 3.861us 3
aten::_conv_depthwise2d 1.04% 21.239us 3.10% 63.201us 21.067us 11.583us 47.13% 11.583us 3.861us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 11.583us 47.13% 11.583us 3.861us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.656us 27.08% 6.656us 2.219us 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 70.08% 1.429ms 70.08% 1.429ms 1.429ms 2.272us 9.25% 2.272us 2.272us 1
aten::empty_strided 1.51% 30.710us 1.51% 30.710us 5.118us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.42% 212.467us 10.42% 212.467us 23.607us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.89% 18.130us 1.15% 23.350us 2.594us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.43% 8.790us 0.43% 8.790us 0.586us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.48% 9.800us 0.48% 9.800us 3.267us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.47% 9.640us 0.47% 9.640us 3.213us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.28% 5.760us 0.35% 7.050us 2.350us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.039ms
Self CUDA time total: 24.575us
======================================================================
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 332.990us 1278.37% 332.990us 332.990us 1
torch_eager 7.09% 142.971us 99.76% 2.011ms 2.011ms 0.000us 0.00% 28.288us 28.288us 1
aten::to 0.35% 7.062us 85.44% 1.723ms 287.120us 0.000us 0.00% 15.232us 2.539us 6
aten::_to_copy 1.18% 23.771us 85.09% 1.716ms 285.943us 0.000us 0.00% 15.232us 2.539us 6
aten::copy_ 2.51% 50.519us 82.47% 1.663ms 277.136us 12.992us 49.88% 15.232us 2.539us 6
aten::conv1d 0.32% 6.541us 5.84% 117.833us 39.278us 0.000us 0.00% 13.056us 4.352us 3
aten::convolution 0.52% 10.410us 5.52% 111.292us 37.097us 0.000us 0.00% 13.056us 4.352us 3
aten::_convolution 1.19% 24.049us 5.00% 100.882us 33.627us 0.000us 0.00% 13.056us 4.352us 3
aten::_conv_depthwise2d 1.01% 20.460us 2.98% 60.052us 20.017us 13.056us 50.12% 13.056us 4.352us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 13.056us 50.12% 13.056us 4.352us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.624us 25.43% 6.624us 2.208us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 6.368us 24.45% 6.368us 2.123us 3
Activity Buffer Request 70.71% 1.426ms 70.71% 1.426ms 1.426ms 2.240us 8.60% 2.240us 2.240us 1
aten::empty_strided 1.44% 29.071us 1.44% 29.071us 4.845us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.31% 207.805us 10.31% 207.805us 23.089us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.90% 18.081us 1.15% 23.201us 2.578us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.43% 8.650us 0.43% 8.650us 0.577us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.49% 9.891us 0.49% 9.891us 3.297us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.42% 8.561us 0.42% 8.561us 2.854us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.31% 6.350us 0.38% 7.610us 2.537us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.016ms
Self CUDA time total: 26.048us
======================================================================
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 332.220us 868.07% 332.220us 332.220us 1
torch_eager 7.10% 144.065us 99.76% 2.024ms 2.024ms 0.000us 0.00% 40.831us 40.831us 1
aten::conv1d 0.30% 6.030us 5.72% 116.102us 38.701us 0.000us 0.00% 22.464us 7.488us 3
aten::convolution 0.49% 9.861us 5.42% 110.072us 36.691us 0.000us 0.00% 22.464us 7.488us 3
aten::_convolution 1.11% 22.459us 4.94% 100.211us 33.404us 0.000us 0.00% 22.464us 7.488us 3
aten::_conv_depthwise2d 1.00% 20.252us 3.07% 62.362us 20.787us 22.464us 58.70% 22.464us 7.488us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 22.464us 58.70% 22.464us 7.488us 3
aten::to 0.31% 6.271us 85.57% 1.737ms 289.428us 0.000us 0.00% 18.367us 3.061us 6
aten::_to_copy 1.14% 23.180us 85.26% 1.730ms 288.383us 0.000us 0.00% 18.367us 3.061us 6
aten::copy_ 2.42% 49.061us 82.56% 1.675ms 279.226us 15.807us 41.30% 18.367us 3.061us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.448us 22.07% 8.448us 2.816us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.359us 19.23% 7.359us 2.453us 3
Activity Buffer Request 70.88% 1.438ms 70.88% 1.438ms 1.438ms 2.560us 6.69% 2.560us 2.560us 1
aten::empty_strided 1.57% 31.760us 1.57% 31.760us 5.293us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.30% 209.084us 10.30% 209.084us 23.232us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.88% 17.889us 1.13% 22.980us 2.553us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.43% 8.691us 0.43% 8.691us 0.579us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.58% 11.680us 0.58% 11.680us 3.893us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.45% 9.220us 0.45% 9.220us 3.073us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.29% 5.850us 0.36% 7.240us 2.413us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.029ms
Self CUDA time total: 38.271us
======================================================================
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 347.901us 847.38% 347.901us 347.901us 1
torch_eager 7.21% 147.111us 99.76% 2.035ms 2.035ms 0.000us 0.00% 43.616us 43.616us 1
aten::conv1d 0.33% 6.680us 5.94% 121.133us 40.378us 0.000us 0.00% 25.376us 8.459us 3
aten::convolution 0.49% 10.011us 5.61% 114.453us 38.151us 0.000us 0.00% 25.376us 8.459us 3
aten::_convolution 1.21% 24.739us 5.12% 104.442us 34.814us 0.000us 0.00% 25.376us 8.459us 3
aten::_conv_depthwise2d 1.05% 21.431us 3.09% 62.981us 20.994us 25.376us 61.81% 25.376us 8.459us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 25.376us 61.81% 25.376us 8.459us 3
aten::to 0.35% 7.210us 85.28% 1.740ms 289.922us 0.000us 0.00% 18.240us 3.040us 6
aten::_to_copy 1.21% 24.639us 84.93% 1.732ms 288.720us 0.000us 0.00% 18.240us 3.040us 6
aten::copy_ 2.56% 52.303us 82.23% 1.677ms 279.542us 15.680us 38.19% 18.240us 3.040us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.288us 20.19% 8.288us 2.763us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.392us 18.00% 7.392us 2.464us 3
Activity Buffer Request 70.32% 1.434ms 70.32% 1.434ms 1.434ms 2.560us 6.24% 2.560us 2.560us 1
aten::empty_strided 1.49% 30.432us 1.49% 30.432us 5.072us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.49% 213.884us 10.49% 213.884us 23.765us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.88% 17.871us 1.14% 23.242us 2.582us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.44% 8.942us 0.44% 8.942us 0.596us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.48% 9.840us 0.48% 9.840us 3.280us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.42% 8.540us 0.42% 8.540us 2.847us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.31% 6.230us 0.37% 7.540us 2.513us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.040ms
Self CUDA time total: 41.056us
======================================================================
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 334.331us 325.39% 334.331us 334.331us 1
torch_eager 7.01% 141.713us 99.76% 2.018ms 2.018ms 0.000us 0.00% 108.764us 108.764us 1
aten::conv1d 0.30% 6.090us 5.77% 116.623us 38.874us 0.000us 0.00% 70.528us 23.509us 3
aten::convolution 0.56% 11.281us 5.46% 110.533us 36.844us 0.000us 0.00% 70.528us 23.509us 3
aten::_convolution 1.11% 22.501us 4.91% 99.252us 33.084us 0.000us 0.00% 70.528us 23.509us 3
aten::_conv_depthwise2d 1.02% 20.538us 3.03% 61.301us 20.434us 70.528us 68.64% 70.528us 23.509us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 70.528us 68.64% 70.528us 23.509us 3
aten::to 0.31% 6.229us 85.56% 1.731ms 288.457us 0.000us 0.00% 38.236us 6.373us 6
aten::_to_copy 1.17% 23.650us 85.25% 1.725ms 287.419us 0.000us 0.00% 38.236us 6.373us 6
aten::copy_ 2.48% 50.230us 82.63% 1.672ms 278.605us 32.221us 31.36% 38.236us 6.373us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 17.598us 17.13% 17.598us 5.866us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 14.623us 14.23% 14.623us 4.874us 3
Activity Buffer Request 70.91% 1.435ms 70.91% 1.435ms 1.435ms 6.015us 5.85% 6.015us 6.015us 1
aten::empty_strided 1.45% 29.232us 1.45% 29.232us 4.872us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.36% 209.517us 10.36% 209.517us 23.280us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.88% 17.770us 1.13% 22.940us 2.549us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.42% 8.560us 0.42% 8.560us 0.571us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.48% 9.771us 0.48% 9.771us 3.257us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.41% 8.351us 0.41% 8.351us 2.784us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.28% 5.590us 0.33% 6.770us 2.257us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.023ms
Self CUDA time total: 102.749us
======================================================================
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 330.969us 293.42% 330.969us 330.969us 1
torch_eager 14.99% 119.634us 99.39% 793.059us 793.059us 0.000us 0.00% 118.814us 118.814us 1
aten::conv1d 0.68% 5.459us 14.66% 116.982us 38.994us 0.000us 0.00% 80.510us 26.837us 3
aten::convolution 1.26% 10.041us 13.98% 111.523us 37.174us 0.000us 0.00% 80.510us 26.837us 3
aten::_convolution 2.84% 22.661us 12.72% 101.482us 33.827us 0.000us 0.00% 80.510us 26.837us 3
aten::_conv_depthwise2d 2.60% 20.719us 7.95% 63.401us 21.134us 80.510us 71.38% 80.510us 26.837us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 80.510us 71.38% 80.510us 26.837us 3
aten::to 0.74% 5.920us 66.51% 530.742us 88.457us 0.000us 0.00% 38.304us 6.384us 6
aten::_to_copy 2.94% 23.422us 65.77% 524.822us 87.470us 0.000us 0.00% 38.304us 6.384us 6
aten::copy_ 6.43% 51.340us 58.99% 470.681us 78.447us 32.288us 28.62% 38.304us 6.384us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 17.696us 15.69% 17.696us 5.899us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 14.592us 12.94% 14.592us 4.864us 3
Activity Buffer Request 29.02% 231.576us 29.02% 231.576us 231.576us 6.016us 5.33% 6.016us 6.016us 1
aten::empty_strided 3.85% 30.719us 3.85% 30.719us 5.120us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 26.56% 211.935us 26.56% 211.935us 23.548us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.12% 16.940us 2.72% 21.720us 2.413us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.02% 8.121us 1.02% 8.121us 0.541us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.20% 9.582us 1.20% 9.582us 3.194us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.12% 8.930us 1.12% 8.930us 2.977us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.72% 5.780us 0.87% 6.970us 2.323us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 797.960us
Self CUDA time total: 112.798us
======================================================================
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 13.79% 117.896us 93.85% 802.069us 802.069us 0.000us 0.00% 432.858us 432.858us 1
torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 419.356us 106.55% 419.356us 419.356us 1
aten::conv1d 0.66% 5.648us 13.24% 113.161us 37.720us 0.000us 0.00% 251.262us 83.754us 3
aten::convolution 1.11% 9.481us 12.58% 107.513us 35.838us 0.000us 0.00% 251.262us 83.754us 3
aten::_convolution 2.53% 21.627us 11.47% 98.032us 32.677us 0.000us 0.00% 251.262us 83.754us 3
aten::_conv_depthwise2d 2.35% 20.121us 7.14% 61.002us 20.334us 251.262us 63.84% 251.262us 83.754us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 251.262us 63.84% 251.262us 83.754us 3
aten::to 0.66% 5.670us 63.66% 544.101us 90.683us 0.000us 0.00% 181.596us 30.266us 6
aten::_to_copy 2.68% 22.880us 63.00% 538.431us 89.739us 0.000us 0.00% 181.596us 30.266us 6
aten::copy_ 6.04% 51.591us 56.88% 486.161us 81.027us 142.333us 36.16% 181.596us 30.266us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 102.271us 25.98% 102.271us 34.090us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 40.062us 10.18% 40.062us 13.354us 3
Activity Buffer Request 28.73% 245.556us 28.73% 245.556us 245.556us 39.263us 9.98% 39.263us 39.263us 1
aten::empty_strided 3.44% 29.390us 3.44% 29.390us 4.898us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 24.77% 211.714us 24.77% 211.714us 23.524us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 1.99% 17.042us 2.56% 21.904us 2.434us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.01% 8.621us 1.01% 8.621us 0.575us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.11% 9.471us 1.11% 9.471us 3.157us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.02% 8.710us 1.02% 8.710us 2.903us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.71% 6.031us 0.84% 7.190us 2.397us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 854.650us
Self CUDA time total: 393.595us
======================================================================
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 12.73% 119.312us 88.90% 833.220us 833.220us 0.000us 0.00% 487.606us 487.606us 1
torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 476.503us 106.43% 476.503us 476.503us 1
aten::conv1d 0.59% 5.550us 12.38% 116.073us 38.691us 0.000us 0.00% 298.682us 99.561us 3
aten::convolution 1.01% 9.430us 11.79% 110.523us 36.841us 0.000us 0.00% 298.682us 99.561us 3
aten::_convolution 2.32% 21.781us 10.79% 101.093us 33.698us 0.000us 0.00% 298.682us 99.561us 3
aten::_conv_depthwise2d 2.19% 20.491us 6.88% 64.493us 21.498us 298.682us 66.71% 298.682us 99.561us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 298.682us 66.71% 298.682us 99.561us 3
aten::to 0.60% 5.580us 60.86% 570.404us 95.067us 0.000us 0.00% 188.924us 31.487us 6
aten::_to_copy 2.42% 22.662us 60.26% 564.824us 94.137us 0.000us 0.00% 188.924us 31.487us 6
aten::copy_ 5.33% 49.982us 54.62% 511.981us 85.330us 149.053us 33.29% 188.924us 31.487us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 108.926us 24.33% 108.926us 36.309us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 40.127us 8.96% 40.127us 13.376us 3
Activity Buffer Request 29.44% 275.977us 29.44% 275.977us 275.977us 39.871us 8.91% 39.871us 39.871us 1
aten::empty_strided 3.22% 30.181us 3.22% 30.181us 5.030us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 22.44% 210.343us 22.44% 210.343us 23.371us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 1.80% 16.910us 2.35% 22.009us 2.445us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.91% 8.519us 0.91% 8.519us 0.568us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.16% 10.891us 1.16% 10.891us 3.630us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.94% 8.790us 0.94% 8.790us 2.930us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.59% 5.500us 0.71% 6.690us 2.230us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 937.282us
Self CUDA time total: 447.735us
======================================================================
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 323.326us 1733.09% 323.326us 323.326us 1
torch_eager 14.06% 116.944us 99.42% 826.859us 826.859us 0.000us 0.00% 20.544us 20.544us 1
aten::to 0.72% 5.971us 68.57% 570.283us 95.047us 0.000us 0.00% 13.344us 2.224us 6
aten::_to_copy 2.68% 22.330us 67.85% 564.312us 94.052us 0.000us 0.00% 13.344us 2.224us 6
aten::copy_ 6.25% 51.969us 61.73% 513.371us 85.562us 11.456us 61.41% 13.344us 2.224us 6
aten::conv1d 0.66% 5.530us 13.54% 112.622us 37.541us 0.000us 0.00% 7.200us 2.400us 3
aten::convolution 1.25% 10.420us 12.88% 107.092us 35.697us 0.000us 0.00% 7.200us 2.400us 3
aten::_convolution 2.52% 20.950us 11.62% 96.672us 32.224us 0.000us 0.00% 7.200us 2.400us 3
aten::_conv_depthwise2d 2.43% 20.241us 7.30% 60.692us 20.231us 7.200us 38.59% 7.200us 2.400us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.200us 38.59% 7.200us 2.400us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 5.888us 31.56% 5.888us 1.963us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.568us 29.85% 5.568us 1.856us 3
Activity Buffer Request 31.20% 259.516us 31.20% 259.516us 259.516us 1.888us 10.12% 1.888us 1.888us 1
aten::empty_strided 3.44% 28.611us 3.44% 28.611us 4.768us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 26.78% 222.677us 26.78% 222.677us 24.742us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.11% 17.509us 2.72% 22.620us 2.513us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.03% 8.541us 1.03% 8.541us 0.569us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.40% 11.610us 1.40% 11.610us 3.870us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.97% 8.050us 0.97% 8.050us 2.683us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.68% 5.660us 0.82% 6.830us 2.277us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 831.660us
Self CUDA time total: 18.656us
======================================================================
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 351.835us 1817.33% 351.835us 351.835us 1
torch_eager 14.08% 121.071us 99.43% 854.999us 854.999us 0.000us 0.00% 21.248us 21.248us 1
aten::to 0.71% 6.141us 68.62% 590.084us 98.347us 0.000us 0.00% 13.312us 2.219us 6
aten::_to_copy 2.73% 23.503us 67.91% 583.943us 97.324us 0.000us 0.00% 13.312us 2.219us 6
aten::copy_ 6.25% 53.711us 59.45% 511.250us 85.208us 11.424us 59.01% 13.312us 2.219us 6
aten::conv1d 0.65% 5.630us 13.53% 116.322us 38.774us 0.000us 0.00% 7.936us 2.645us 3
aten::convolution 1.12% 9.630us 12.87% 110.692us 36.897us 0.000us 0.00% 7.936us 2.645us 3
aten::_convolution 2.70% 23.181us 11.75% 101.062us 33.687us 0.000us 0.00% 7.936us 2.645us 3
aten::_conv_depthwise2d 2.42% 20.779us 7.31% 62.821us 20.940us 7.936us 40.99% 7.936us 2.645us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.936us 40.99% 7.936us 2.645us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 5.856us 30.25% 5.856us 1.952us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.568us 28.76% 5.568us 1.856us 3
Activity Buffer Request 31.74% 272.946us 31.74% 272.946us 272.946us 1.888us 9.75% 1.888us 1.888us 1
aten::empty_strided 5.72% 49.190us 5.72% 49.190us 8.198us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 24.15% 207.684us 24.15% 207.684us 23.076us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.01% 17.302us 2.65% 22.752us 2.528us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.04% 8.971us 1.04% 8.971us 0.598us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.16% 9.970us 1.16% 9.970us 3.323us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.04% 8.981us 1.04% 8.981us 2.994us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.67% 5.729us 0.81% 6.930us 2.310us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 859.928us
Self CUDA time total: 19.360us
======================================================================
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 336.603us 1730.16% 336.603us 336.603us 1
torch_eager 7.08% 144.052us 99.72% 2.028ms 2.028ms 0.000us 0.00% 21.631us 21.631us 1
aten::to 0.35% 7.181us 85.48% 1.738ms 289.712us 0.000us 0.00% 14.367us 2.394us 6
aten::_to_copy 1.19% 24.130us 85.13% 1.731ms 288.515us 0.000us 0.00% 14.367us 2.394us 6
aten::copy_ 2.47% 50.222us 82.49% 1.678ms 279.593us 12.191us 62.66% 14.367us 2.394us 6
aten::conv1d 0.31% 6.211us 5.80% 117.993us 39.331us 0.000us 0.00% 7.264us 2.421us 3
aten::convolution 0.49% 9.910us 5.50% 111.782us 37.261us 0.000us 0.00% 7.264us 2.421us 3
aten::_convolution 1.11% 22.590us 5.01% 101.872us 33.957us 0.000us 0.00% 7.264us 2.421us 3
aten::_conv_depthwise2d 1.01% 20.531us 3.08% 62.662us 20.887us 7.264us 37.34% 7.264us 2.421us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.264us 37.34% 7.264us 2.421us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.272us 32.24% 6.272us 2.091us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.919us 30.42% 5.919us 1.973us 3
Activity Buffer Request 70.94% 1.443ms 70.94% 1.443ms 1.443ms 2.176us 11.18% 2.176us 2.176us 1
aten::empty_strided 1.45% 29.401us 1.45% 29.401us 4.900us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.12% 205.814us 10.12% 205.814us 22.868us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.93% 19.000us 1.20% 24.310us 2.701us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.43% 8.650us 0.43% 8.650us 0.577us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.52% 10.541us 0.52% 10.541us 3.514us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.51% 10.450us 0.51% 10.450us 3.483us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.30% 6.000us 0.36% 7.230us 2.410us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.034ms
Self CUDA time total: 19.455us
======================================================================
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 330.813us 1648.79% 330.813us 330.813us 1
torch_eager 15.21% 119.271us 99.33% 778.918us 778.918us 0.000us 0.00% 22.208us 22.208us 1
aten::to 0.72% 5.611us 65.71% 515.312us 85.885us 0.000us 0.00% 14.305us 2.384us 6
aten::_to_copy 2.87% 22.510us 65.00% 509.701us 84.950us 0.000us 0.00% 14.305us 2.384us 6
aten::copy_ 6.38% 50.021us 58.03% 455.090us 75.848us 12.161us 60.61% 14.305us 2.384us 6
aten::conv1d 0.69% 5.380us 15.11% 118.473us 39.491us 0.000us 0.00% 7.903us 2.634us 3
aten::convolution 1.31% 10.292us 14.42% 113.093us 37.698us 0.000us 0.00% 7.903us 2.634us 3
aten::_convolution 2.98% 23.360us 13.11% 102.801us 34.267us 0.000us 0.00% 7.903us 2.634us 3
aten::_conv_depthwise2d 2.80% 21.952us 8.17% 64.041us 21.347us 7.903us 39.39% 7.903us 2.634us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 7.903us 39.39% 7.903us 2.634us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 6.240us 31.10% 6.240us 2.080us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 5.921us 29.51% 5.921us 1.974us 3
Activity Buffer Request 27.43% 215.065us 27.43% 215.065us 215.065us 2.144us 10.69% 2.144us 2.144us 1
aten::empty_strided 4.09% 32.101us 4.09% 32.101us 5.350us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 27.10% 212.544us 27.10% 212.544us 23.616us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.14% 16.752us 2.74% 21.481us 2.387us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.03% 8.081us 1.03% 8.081us 0.539us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.34% 10.539us 1.34% 10.539us 3.513us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.15% 9.010us 1.15% 9.010us 3.003us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.74% 5.829us 0.90% 7.070us 2.357us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 784.179us
Self CUDA time total: 20.064us
======================================================================
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 332.020us 920.74% 332.020us 332.020us 1
torch_eager 14.90% 117.475us 99.36% 783.438us 783.438us 0.000us 0.00% 38.651us 38.651us 1
aten::conv1d 0.68% 5.380us 14.48% 114.172us 38.057us 0.000us 0.00% 20.190us 6.730us 3
aten::convolution 1.18% 9.340us 13.80% 108.792us 36.264us 0.000us 0.00% 20.190us 6.730us 3
aten::_convolution 2.79% 21.980us 12.61% 99.452us 33.151us 0.000us 0.00% 20.190us 6.730us 3
aten::_conv_depthwise2d 2.62% 20.631us 7.92% 62.452us 20.817us 20.190us 55.99% 20.190us 6.730us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 20.190us 55.99% 20.190us 6.730us 3
aten::to 0.79% 6.191us 66.58% 524.962us 87.494us 0.000us 0.00% 18.461us 3.077us 6
aten::_to_copy 2.94% 23.190us 65.79% 518.771us 86.462us 0.000us 0.00% 18.461us 3.077us 6
aten::copy_ 6.46% 50.920us 58.97% 464.950us 77.492us 15.870us 44.01% 18.461us 3.077us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.479us 23.51% 8.479us 2.826us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.391us 20.50% 7.391us 2.464us 3
Activity Buffer Request 28.77% 226.875us 28.77% 226.875us 226.875us 2.591us 7.19% 2.591us 2.591us 1
aten::empty_strided 3.88% 30.631us 3.88% 30.631us 5.105us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 26.51% 209.015us 26.51% 209.015us 23.224us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.16% 17.000us 2.79% 22.010us 2.446us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.07% 8.441us 1.07% 8.441us 0.563us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.46% 11.521us 1.46% 11.521us 3.840us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.07% 8.440us 1.07% 8.440us 2.813us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.73% 5.720us 0.87% 6.850us 2.283us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 788.468us
Self CUDA time total: 36.060us
======================================================================
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 323.480us 850.14% 323.480us 323.480us 1
torch_eager 14.55% 115.293us 99.34% 787.399us 787.399us 0.000us 0.00% 40.643us 40.643us 1
aten::conv1d 0.68% 5.400us 14.40% 114.153us 38.051us 0.000us 0.00% 22.336us 7.445us 3
aten::convolution 1.16% 9.210us 13.72% 108.753us 36.251us 0.000us 0.00% 22.336us 7.445us 3
aten::_convolution 2.80% 22.227us 12.56% 99.543us 33.181us 0.000us 0.00% 22.336us 7.445us 3
aten::_conv_depthwise2d 2.52% 20.003us 7.87% 62.343us 20.781us 22.336us 58.70% 22.336us 7.445us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 22.336us 58.70% 22.336us 7.445us 3
aten::to 0.94% 7.450us 67.15% 532.253us 88.709us 0.000us 0.00% 18.307us 3.051us 6
aten::_to_copy 2.90% 22.999us 66.21% 524.803us 87.467us 0.000us 0.00% 18.307us 3.051us 6
aten::copy_ 6.35% 50.294us 59.67% 472.953us 78.825us 15.714us 41.30% 18.307us 3.051us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 8.321us 21.87% 8.321us 2.774us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 7.393us 19.43% 7.393us 2.464us 3
Activity Buffer Request 30.44% 241.286us 30.44% 241.286us 241.286us 2.593us 6.81% 2.593us 2.593us 1
aten::empty_strided 3.64% 28.851us 3.64% 28.851us 4.808us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 25.80% 204.463us 25.80% 204.463us 22.718us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.08% 16.472us 2.71% 21.512us 2.390us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.07% 8.500us 1.07% 8.500us 0.567us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.34% 10.600us 1.34% 10.600us 3.533us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.09% 8.650us 1.09% 8.650us 2.883us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.71% 5.641us 0.87% 6.891us 2.297us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 792.609us
Self CUDA time total: 38.050us
======================================================================
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 336.347us 525.30% 336.347us 336.347us 1
torch_eager 15.07% 122.512us 99.37% 807.929us 807.929us 0.000us 0.00% 68.125us 68.125us 1
aten::conv1d 0.67% 5.471us 14.06% 114.283us 38.094us 0.000us 0.00% 41.663us 13.888us 3
aten::convolution 1.12% 9.100us 13.38% 108.812us 36.271us 0.000us 0.00% 41.663us 13.888us 3
aten::_convolution 2.67% 21.730us 12.26% 99.712us 33.237us 0.000us 0.00% 41.663us 13.888us 3
aten::_conv_depthwise2d 2.50% 20.351us 7.64% 62.102us 20.701us 41.663us 65.07% 41.663us 13.888us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 41.663us 65.07% 41.663us 13.888us 3
aten::to 0.85% 6.891us 66.79% 543.023us 90.504us 0.000us 0.00% 26.462us 4.410us 6
aten::_to_copy 2.85% 23.179us 65.94% 536.132us 89.355us 0.000us 0.00% 26.462us 4.410us 6
aten::copy_ 6.23% 50.622us 59.33% 482.372us 80.395us 22.367us 34.93% 26.462us 4.410us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 11.935us 18.64% 11.935us 3.978us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 10.432us 16.29% 10.432us 3.477us 3
Activity Buffer Request 30.75% 250.036us 30.75% 250.036us 250.036us 4.095us 6.40% 4.095us 4.095us 1
aten::empty_strided 3.76% 30.581us 3.76% 30.581us 5.097us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 25.31% 205.766us 25.31% 205.766us 22.863us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.22% 18.070us 2.95% 23.970us 2.663us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.13% 9.220us 1.13% 9.220us 0.615us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.19% 9.690us 1.19% 9.690us 3.230us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.99% 8.009us 0.99% 8.009us 2.670us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.69% 5.650us 0.84% 6.800us 2.267us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 813.049us
Self CUDA time total: 64.030us
======================================================================
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 325.852us 467.34% 325.852us 325.852us 1
torch_eager 14.81% 118.946us 99.39% 798.399us 798.399us 0.000us 0.00% 73.789us 73.789us 1
aten::conv1d 0.70% 5.610us 14.26% 114.513us 38.171us 0.000us 0.00% 47.294us 15.765us 3
aten::convolution 1.17% 9.382us 13.56% 108.903us 36.301us 0.000us 0.00% 47.294us 15.765us 3
aten::_convolution 2.75% 22.119us 12.39% 99.521us 33.174us 0.000us 0.00% 47.294us 15.765us 3
aten::_conv_depthwise2d 2.53% 20.361us 7.64% 61.351us 20.450us 47.294us 67.83% 47.294us 15.765us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 47.294us 67.83% 47.294us 15.765us 3
aten::to 0.79% 6.379us 67.07% 538.781us 89.797us 0.000us 0.00% 26.495us 4.416us 6
aten::_to_copy 2.79% 22.401us 66.28% 532.402us 88.734us 0.000us 0.00% 26.495us 4.416us 6
aten::copy_ 6.32% 50.749us 59.88% 480.970us 80.162us 22.431us 32.17% 26.495us 4.416us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 12.031us 17.25% 12.031us 4.010us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 10.400us 14.92% 10.400us 3.467us 3
Activity Buffer Request 31.04% 249.326us 31.04% 249.326us 249.326us 4.064us 5.83% 4.064us 4.064us 1
aten::empty_strided 3.61% 29.031us 3.61% 29.031us 4.839us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 25.24% 202.725us 25.24% 202.725us 22.525us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.08% 16.713us 2.70% 21.680us 2.409us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.07% 8.568us 1.07% 8.568us 0.571us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.32% 10.569us 1.32% 10.569us 3.523us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.07% 8.591us 1.07% 8.591us 2.864us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.82% 6.609us 1.00% 8.010us 2.670us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 803.260us
Self CUDA time total: 69.725us
======================================================================
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 346.241us 186.20% 346.241us 346.241us 1
torch_eager 7.14% 146.093us 99.75% 2.041ms 2.041ms 0.000us 0.00% 196.000us 196.000us 1
aten::conv1d 0.31% 6.251us 5.79% 118.533us 39.511us 0.000us 0.00% 133.248us 44.416us 3
aten::convolution 0.51% 10.359us 5.49% 112.282us 37.427us 0.000us 0.00% 133.248us 44.416us 3
aten::_convolution 1.19% 24.280us 4.98% 101.923us 33.974us 0.000us 0.00% 133.248us 44.416us 3
aten::_conv_depthwise2d 1.04% 21.191us 3.02% 61.762us 20.587us 133.248us 71.66% 133.248us 44.416us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 133.248us 71.66% 133.248us 44.416us 3
aten::to 0.32% 6.489us 85.48% 1.749ms 291.443us 0.000us 0.00% 62.752us 10.459us 6
aten::_to_copy 1.16% 23.832us 85.16% 1.742ms 290.362us 0.000us 0.00% 62.752us 10.459us 6
aten::copy_ 2.53% 51.751us 82.59% 1.689ms 281.575us 52.704us 28.34% 62.752us 10.459us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 29.760us 16.00% 29.760us 9.920us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 22.944us 12.34% 22.944us 7.648us 3
Activity Buffer Request 71.04% 1.453ms 71.04% 1.453ms 1.453ms 10.048us 5.40% 10.048us 10.048us 1
aten::empty_strided 1.41% 28.891us 1.41% 28.891us 4.815us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 10.04% 205.324us 10.04% 205.324us 22.814us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.86% 17.550us 1.11% 22.661us 2.518us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.42% 8.641us 0.42% 8.641us 0.576us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.53% 10.750us 0.53% 10.750us 3.583us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.44% 9.021us 0.44% 9.021us 3.007us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.30% 6.220us 0.37% 7.470us 2.490us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 2.046ms
Self CUDA time total: 185.952us
======================================================================
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 341.281us 162.73% 341.281us 341.281us 1
torch_eager 15.25% 117.693us 99.36% 766.878us 766.878us 0.000us 0.00% 223.168us 223.168us 1
aten::conv1d 0.68% 5.279us 14.60% 112.702us 37.567us 0.000us 0.00% 154.016us 51.339us 3
aten::convolution 1.24% 9.560us 13.92% 107.423us 35.808us 0.000us 0.00% 154.016us 51.339us 3
aten::_convolution 2.67% 20.611us 12.68% 97.863us 32.621us 0.000us 0.00% 154.016us 51.339us 3
aten::_conv_depthwise2d 2.74% 21.170us 8.14% 62.852us 20.951us 154.016us 73.44% 154.016us 51.339us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 154.016us 73.44% 154.016us 51.339us 3
aten::to 0.75% 5.750us 66.22% 511.121us 85.187us 0.000us 0.00% 69.152us 11.525us 6
aten::_to_copy 2.86% 22.060us 65.48% 505.371us 84.228us 0.000us 0.00% 69.152us 11.525us 6
aten::copy_ 6.81% 52.581us 58.74% 453.391us 75.565us 55.712us 26.56% 69.152us 11.525us 6
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 32.705us 15.59% 32.705us 10.902us 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 28.38% 219.045us 28.38% 219.045us 219.045us 13.440us 6.41% 13.440us 13.440us 1
aten::empty_strided 3.88% 29.920us 3.88% 29.920us 4.987us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 26.50% 204.546us 26.50% 204.546us 22.727us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 2.10% 16.212us 2.69% 20.781us 2.309us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 1.01% 7.798us 1.01% 7.798us 0.520us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 1.33% 10.250us 1.33% 10.250us 3.417us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 1.12% 8.651us 1.12% 8.651us 2.884us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.71% 5.470us 0.87% 6.730us 2.243us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 771.798us
Self CUDA time total: 209.728us
======================================================================
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 6.80% 123.239us 52.01% 942.341us 942.341us 0.000us 0.00% 1.520ms 1.520ms 1
torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 1.422ms 100.42% 1.422ms 1.422ms 1
aten::to 0.34% 6.231us 36.98% 669.957us 111.660us 0.000us 0.00% 824.603us 137.434us 6
aten::_to_copy 1.55% 28.122us 36.63% 663.726us 110.621us 0.000us 0.00% 824.603us 137.434us 6
aten::copy_ 2.95% 53.430us 24.71% 447.748us 74.625us 720.572us 50.89% 824.603us 137.434us 6
aten::conv1d 0.34% 6.111us 6.66% 120.744us 40.248us 0.000us 0.00% 695.357us 231.786us 3
aten::convolution 0.56% 10.201us 6.33% 114.633us 38.211us 0.000us 0.00% 695.357us 231.786us 3
aten::_convolution 1.36% 24.689us 5.76% 104.432us 34.811us 0.000us 0.00% 695.357us 231.786us 3
aten::_conv_depthwise2d 1.22% 22.151us 3.50% 63.431us 21.144us 695.357us 49.11% 695.357us 231.786us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 695.357us 49.11% 695.357us 231.786us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 407.263us 28.76% 407.263us 135.754us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 313.309us 22.13% 313.309us 104.436us 3
Activity Buffer Request 11.46% 207.684us 11.46% 207.684us 207.684us 104.031us 7.35% 104.031us 104.031us 1
aten::empty_strided 2.02% 36.603us 10.37% 187.856us 31.309us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 11.47% 207.874us 11.47% 207.874us 23.097us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.99% 18.011us 1.30% 23.581us 2.620us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.51% 9.270us 0.51% 9.270us 0.618us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.60% 10.830us 0.60% 10.830us 3.610us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.51% 9.210us 0.51% 9.210us 3.070us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.35% 6.401us 0.43% 7.711us 2.570us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 1.812ms
Self CUDA time total: 1.416ms
======================================================================
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 6.34% 116.852us 42.22% 778.698us 778.698us 0.000us 0.00% 1.501ms 1.501ms 1
torch_eager 0.00% 0.000us 0.00% 0.000us 0.000us 1.431ms 100.41% 1.431ms 1.431ms 1
aten::to 0.32% 5.860us 28.27% 521.352us 86.892us 0.000us 0.00% 762.491us 127.082us 6
aten::_to_copy 1.22% 22.580us 27.95% 515.492us 85.915us 0.000us 0.00% 762.491us 127.082us 6
aten::copy_ 2.82% 52.081us 25.06% 462.219us 77.037us 686.779us 48.19% 762.491us 127.082us 6
aten::conv1d 0.30% 5.601us 6.22% 114.743us 38.248us 0.000us 0.00% 738.362us 246.121us 3
aten::convolution 0.51% 9.380us 5.92% 109.142us 36.381us 0.000us 0.00% 738.362us 246.121us 3
aten::_convolution 1.20% 22.202us 5.41% 99.762us 33.254us 0.000us 0.00% 738.362us 246.121us 3
aten::_conv_depthwise2d 1.10% 20.251us 3.34% 61.660us 20.553us 738.362us 51.81% 738.362us 246.121us 3
void at::native::(anonymous namespace)::conv_depthwi... 0.00% 0.000us 0.00% 0.000us 0.000us 738.362us 51.81% 738.362us 246.121us 3
void at::native::elementwise_kernel<128, 4, at::nati... 0.00% 0.000us 0.00% 0.000us 0.000us 398.333us 27.95% 398.333us 132.778us 3
void at::native::unrolled_elementwise_kernel<at::nat... 0.00% 0.000us 0.00% 0.000us 0.000us 288.446us 20.24% 288.446us 96.149us 3
Activity Buffer Request 12.41% 228.855us 12.41% 228.855us 228.855us 75.712us 5.31% 75.712us 75.712us 1
aten::empty_strided 1.66% 30.693us 1.66% 30.693us 5.115us 0.000us 0.00% 0.000us 0.000us 6
cudaLaunchKernel 11.01% 202.993us 11.01% 202.993us 22.555us 0.000us 0.00% 0.000us 0.000us 9
aten::unsqueeze 0.90% 16.659us 1.18% 21.780us 2.420us 0.000us 0.00% 0.000us 0.000us 9
aten::as_strided 0.46% 8.512us 0.46% 8.512us 0.567us 0.000us 0.00% 0.000us 0.000us 15
aten::empty 0.53% 9.739us 0.53% 9.739us 3.246us 0.000us 0.00% 0.000us 0.000us 3
aten::resize_ 0.54% 9.960us 0.54% 9.960us 3.320us 0.000us 0.00% 0.000us 0.000us 3
aten::squeeze 0.33% 6.150us 0.40% 7.440us 2.480us 0.000us 0.00% 0.000us 0.000us 3
------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
Self CPU time total: 1.844ms
Self CUDA time total: 1.425ms
impl wl p50(ms) ok
torch_eager cuda_B2_D2048_S128_W2 0.08 True
torch_eager cuda_B2_D2048_S128_W4 0.08 True
torch_eager cuda_B2_D2048_S2048_W2 0.15 True
torch_eager cuda_B2_D2048_S2048_W4 0.16 True
torch_eager cuda_B2_D2048_S512_W2 0.08 True
torch_eager cuda_B2_D2048_S512_W4 0.08 True
torch_eager cuda_B2_D64_S128_W2 0.07 True
torch_eager cuda_B2_D64_S128_W4 0.09 True
torch_eager cuda_B2_D64_S2048_W2 0.08 True
torch_eager cuda_B2_D64_S2048_W4 0.08 True
torch_eager cuda_B2_D64_S512_W2 0.09 True
torch_eager cuda_B2_D64_S512_W4 0.08 True
torch_eager cuda_B4_D2048_S128_W2 0.08 True
torch_eager cuda_B4_D2048_S128_W4 0.08 True
torch_eager cuda_B4_D2048_S2048_W2 0.49 True
torch_eager cuda_B4_D2048_S2048_W4 0.50 True
torch_eager cuda_B4_D2048_S512_W2 0.09 True
torch_eager cuda_B4_D2048_S512_W4 0.10 True
torch_eager cuda_B4_D64_S128_W2 0.08 True
torch_eager cuda_B4_D64_S128_W4 0.08 True
torch_eager cuda_B4_D64_S2048_W2 0.08 True
torch_eager cuda_B4_D64_S2048_W4 0.08 True
torch_eager cuda_B4_D64_S512_W2 0.08 True
torch_eager cuda_B4_D64_S512_W4 0.08 True