flow._C.frac #9979

ccssu · 2023-03-13T02:02:01Z

参考文档:

https://github.com/Oneflow-Inc/OneTeam/blob/master/tutorial/howto_add_user_op.md
20220726-工作注意事项和做法.pptx
Tensor 的 api 目前主要用 Cpython 的方式导出，写法可参考 Add addcdiv
OneFlow inplace 算子开发指南

CLAassistant · 2023-03-13T02:02:08Z

All committers have signed the CLA.

BBuf

可以参考这个pr直接用element_wise的模板来实现，效率更高。#9958 （忽略这个pr里面的高阶导部分即可）

ccssu · 2023-03-13T02:29:06Z

可以参考这个pr直接用element_wise的模板来实现，效率更高。#9958 （忽略这个pr里面的高阶导部分即可）

嗯嗯，好的

…functor.cpp

ccssu · 2023-03-16T02:24:41Z

oneflow/core/autograd/gradient_funcs/activation.cpp

oneflow/core/functional/impl/activation_functor.cpp

oneflow/user/kernels/frac_kernel.cpp

oneflow/core/functional/impl/activation_functor.cpp

oneflow/core/autograd/gradient_funcs/activation.cpp

Co-authored-by: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>

github-actions · 2023-03-20T03:40:21Z

Speed stats:

github-actions · 2023-03-20T07:40:21Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 140.9ms (= 14089.8ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.9ms (= 14387.3ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.02 (= 143.9ms / 140.9ms)

OneFlow resnet50 time: 80.5ms (= 8054.9ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 83.7ms (= 8369.8ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.04 (= 83.7ms / 80.5ms)

OneFlow resnet50 time: 49.1ms (= 9816.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 60.6ms (= 12119.7ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.23 (= 60.6ms / 49.1ms)

OneFlow resnet50 time: 33.5ms (= 6697.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 54.0ms (= 10807.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.61 (= 54.0ms / 33.5ms)

OneFlow resnet50 time: 25.0ms (= 5009.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 42.1ms (= 8413.8ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.68 (= 42.1ms / 25.0ms)

OneFlow swin dataloader time: 0.244s (= 48.861s / 200, num_workers=1)
PyTorch swin dataloader time: 0.148s (= 29.657s / 200, num_workers=1)
Relative speed: 0.607 (= 0.148s / 0.244s)

OneFlow swin dataloader time: 0.070s (= 13.933s / 200, num_workers=4)
PyTorch swin dataloader time: 0.046s (= 9.136s / 200, num_workers=4)
Relative speed: 0.656 (= 0.046s / 0.070s)

OneFlow swin dataloader time: 0.043s (= 8.604s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.340s / 200, num_workers=8)
Relative speed: 0.504 (= 0.022s / 0.043s)

❌ OneFlow resnet50 time: 152.4ms (= 15242.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.2ms (= 16123.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.06 (= 161.2ms / 152.4ms)

OneFlow resnet50 time: 90.9ms (= 9091.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 102.6ms (= 10257.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 102.6ms / 90.9ms)

OneFlow resnet50 time: 58.8ms (= 11760.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 80.4ms (= 16075.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.37 (= 80.4ms / 58.8ms)

OneFlow resnet50 time: 41.9ms (= 8376.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 74.7ms (= 14937.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.78 (= 74.7ms / 41.9ms)

OneFlow resnet50 time: 36.5ms (= 7299.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 70.2ms (= 14045.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.92 (= 70.2ms / 36.5ms)

github-actions · 2023-03-20T07:51:20Z

CI failed when running job: cpu-misc. PR label automerge has been removed

github-actions · 2023-03-21T04:18:22Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.1ms (= 14105.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.5ms (= 14351.3ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.02 (= 143.5ms / 141.1ms)

OneFlow resnet50 time: 80.6ms (= 8060.9ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 83.3ms (= 8330.2ms / 100, input_shape=[8, 3, 224, 224])
❌ Relative speed: 1.03 (= 83.3ms / 80.6ms)

OneFlow resnet50 time: 49.2ms (= 9847.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 55.6ms (= 11118.9ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.13 (= 55.6ms / 49.2ms)

OneFlow resnet50 time: 32.1ms (= 6426.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 45.7ms (= 9142.7ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.42 (= 45.7ms / 32.1ms)

OneFlow resnet50 time: 25.5ms (= 5092.6ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 37.0ms (= 7400.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.45 (= 37.0ms / 25.5ms)

OneFlow swin dataloader time: 0.235s (= 47.096s / 200, num_workers=1)
PyTorch swin dataloader time: 0.153s (= 30.647s / 200, num_workers=1)
Relative speed: 0.651 (= 0.153s / 0.235s)

OneFlow swin dataloader time: 0.066s (= 13.146s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.223s / 200, num_workers=4)
Relative speed: 0.626 (= 0.041s / 0.066s)

OneFlow swin dataloader time: 0.038s (= 7.615s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.350s / 200, num_workers=8)
Relative speed: 0.571 (= 0.022s / 0.038s)

❌ OneFlow resnet50 time: 152.3ms (= 15230.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.5ms (= 16154.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.06 (= 161.5ms / 152.3ms)

OneFlow resnet50 time: 91.0ms (= 9097.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 102.6ms (= 10262.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 102.6ms / 91.0ms)

OneFlow resnet50 time: 59.3ms (= 11850.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.2ms (= 15642.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.32 (= 78.2ms / 59.3ms)

OneFlow resnet50 time: 42.4ms (= 8474.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 70.5ms (= 14095.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.66 (= 70.5ms / 42.4ms)

OneFlow resnet50 time: 36.5ms (= 7307.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.9ms (= 13772.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.88 (= 68.9ms / 36.5ms)

github-actions · 2023-03-24T02:28:26Z

Speed stats:

github-actions · 2023-03-24T12:00:38Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.0ms (= 14101.9ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 145.2ms (= 14516.8ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.03 (= 145.2ms / 141.0ms)

OneFlow resnet50 time: 80.7ms (= 8066.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 83.7ms (= 8368.0ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.04 (= 83.7ms / 80.7ms)

OneFlow resnet50 time: 49.6ms (= 9927.5ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 57.7ms (= 11536.3ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.16 (= 57.7ms / 49.6ms)

OneFlow resnet50 time: 32.8ms (= 6566.2ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 41.9ms (= 8384.7ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.28 (= 41.9ms / 32.8ms)

OneFlow resnet50 time: 24.8ms (= 4961.0ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 35.8ms (= 7167.6ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.44 (= 35.8ms / 24.8ms)

OneFlow swin dataloader time: 0.236s (= 47.187s / 200, num_workers=1)
PyTorch swin dataloader time: 0.149s (= 29.726s / 200, num_workers=1)
Relative speed: 0.630 (= 0.149s / 0.236s)

OneFlow swin dataloader time: 0.069s (= 13.839s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.367s / 200, num_workers=4)
Relative speed: 0.605 (= 0.042s / 0.069s)

OneFlow swin dataloader time: 0.039s (= 7.890s / 200, num_workers=8)
PyTorch swin dataloader time: 0.023s (= 4.580s / 200, num_workers=8)
Relative speed: 0.580 (= 0.023s / 0.039s)

❌ OneFlow resnet50 time: 152.8ms (= 15284.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.2ms (= 16115.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.05 (= 161.2ms / 152.8ms)

OneFlow resnet50 time: 91.2ms (= 9116.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 102.8ms (= 10284.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 102.8ms / 91.2ms)

OneFlow resnet50 time: 59.0ms (= 11798.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.9ms (= 15784.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 78.9ms / 59.0ms)

OneFlow resnet50 time: 42.5ms (= 8506.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 76.8ms (= 15355.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.81 (= 76.8ms / 42.5ms)

OneFlow resnet50 time: 35.8ms (= 7167.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 72.6ms (= 14525.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 2.03 (= 72.6ms / 35.8ms)

github-actions · 2023-03-25T03:56:11Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.0ms (= 14096.6ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.8ms (= 14378.2ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.02 (= 143.8ms / 141.0ms)

OneFlow resnet50 time: 80.4ms (= 8043.9ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.4ms (= 8440.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.05 (= 84.4ms / 80.4ms)

OneFlow resnet50 time: 48.8ms (= 9766.8ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 56.4ms (= 11273.7ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.15 (= 56.4ms / 48.8ms)

OneFlow resnet50 time: 32.2ms (= 6436.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 47.6ms (= 9520.1ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.48 (= 47.6ms / 32.2ms)

OneFlow resnet50 time: 24.5ms (= 4906.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 40.6ms (= 8110.7ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.65 (= 40.6ms / 24.5ms)

OneFlow swin dataloader time: 0.257s (= 51.313s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.160s / 200, num_workers=1)
Relative speed: 0.588 (= 0.151s / 0.257s)

OneFlow swin dataloader time: 0.072s (= 14.334s / 200, num_workers=4)
PyTorch swin dataloader time: 0.044s (= 8.899s / 200, num_workers=4)
Relative speed: 0.621 (= 0.044s / 0.072s)

OneFlow swin dataloader time: 0.040s (= 8.056s / 200, num_workers=8)
PyTorch swin dataloader time: 0.023s (= 4.547s / 200, num_workers=8)
Relative speed: 0.564 (= 0.023s / 0.040s)

❌ OneFlow resnet50 time: 152.8ms (= 15280.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.0ms (= 16099.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.05 (= 161.0ms / 152.8ms)

OneFlow resnet50 time: 91.0ms (= 9102.8ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 103.3ms (= 10329.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 103.3ms / 91.0ms)

OneFlow resnet50 time: 59.3ms (= 11857.3ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.1ms (= 15613.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.32 (= 78.1ms / 59.3ms)

OneFlow resnet50 time: 41.9ms (= 8381.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 73.6ms (= 14712.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.76 (= 73.6ms / 41.9ms)

OneFlow resnet50 time: 36.6ms (= 7326.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.9ms (= 13770.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.88 (= 68.9ms / 36.6ms)

github-actions · 2023-03-28T05:24:44Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.2ms (= 14124.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 142.9ms (= 14293.2ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.01 (= 142.9ms / 141.2ms)

OneFlow resnet50 time: 81.8ms (= 8176.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 86.0ms (= 8600.9ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.05 (= 86.0ms / 81.8ms)

OneFlow resnet50 time: 50.8ms (= 10164.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.0ms (= 11594.6ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.14 (= 58.0ms / 50.8ms)

OneFlow resnet50 time: 33.8ms (= 6767.9ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 42.5ms (= 8500.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.26 (= 42.5ms / 33.8ms)

OneFlow resnet50 time: 27.4ms (= 5483.9ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.9ms (= 7973.7ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.45 (= 39.9ms / 27.4ms)

OneFlow swin dataloader time: 0.239s (= 47.883s / 200, num_workers=1)
PyTorch swin dataloader time: 0.147s (= 29.334s / 200, num_workers=1)
Relative speed: 0.613 (= 0.147s / 0.239s)

OneFlow swin dataloader time: 0.071s (= 14.218s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.149s / 200, num_workers=4)
Relative speed: 0.573 (= 0.041s / 0.071s)

OneFlow swin dataloader time: 0.043s (= 8.694s / 200, num_workers=8)
PyTorch swin dataloader time: 0.023s (= 4.522s / 200, num_workers=8)
Relative speed: 0.520 (= 0.023s / 0.043s)

❌ OneFlow resnet50 time: 153.1ms (= 15313.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 163.7ms (= 16373.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.07 (= 163.7ms / 153.1ms)

OneFlow resnet50 time: 92.6ms (= 9263.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 103.8ms (= 10382.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.12 (= 103.8ms / 92.6ms)

OneFlow resnet50 time: 61.1ms (= 12227.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.2ms (= 15646.3ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.28 (= 78.2ms / 61.1ms)

OneFlow resnet50 time: 43.0ms (= 8595.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 70.2ms (= 14038.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.63 (= 70.2ms / 43.0ms)

OneFlow resnet50 time: 36.6ms (= 7323.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 67.7ms (= 13549.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.85 (= 67.7ms / 36.6ms)

github-actions · 2023-03-30T10:46:08Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.2ms (= 14124.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 145.1ms (= 14508.4ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.03 (= 145.1ms / 141.2ms)

OneFlow resnet50 time: 83.3ms (= 8325.5ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 86.4ms (= 8644.0ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.04 (= 86.4ms / 83.3ms)

OneFlow resnet50 time: 51.6ms (= 10320.9ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.0ms (= 11608.7ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.12 (= 58.0ms / 51.6ms)

OneFlow resnet50 time: 33.9ms (= 6778.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 44.6ms (= 8914.1ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.32 (= 44.6ms / 33.9ms)

OneFlow resnet50 time: 26.3ms (= 5250.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 44.5ms (= 8893.8ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.69 (= 44.5ms / 26.3ms)

OneFlow swin dataloader time: 0.239s (= 47.717s / 200, num_workers=1)
PyTorch swin dataloader time: 0.148s (= 29.661s / 200, num_workers=1)
Relative speed: 0.622 (= 0.148s / 0.239s)

OneFlow swin dataloader time: 0.069s (= 13.725s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.235s / 200, num_workers=4)
Relative speed: 0.600 (= 0.041s / 0.069s)

OneFlow swin dataloader time: 0.038s (= 7.521s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.488s / 200, num_workers=8)
Relative speed: 0.597 (= 0.022s / 0.038s)

❌ OneFlow resnet50 time: 154.4ms (= 15435.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 164.9ms (= 16486.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.07 (= 164.9ms / 154.4ms)

OneFlow resnet50 time: 93.8ms (= 9376.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 104.6ms (= 10458.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.12 (= 104.6ms / 93.8ms)

OneFlow resnet50 time: 61.3ms (= 12260.7ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 79.1ms (= 15818.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.29 (= 79.1ms / 61.3ms)

OneFlow resnet50 time: 43.3ms (= 8668.3ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 72.8ms (= 14559.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.68 (= 72.8ms / 43.3ms)

OneFlow resnet50 time: 35.9ms (= 7178.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.8ms (= 13369.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.86 (= 66.8ms / 35.9ms)

github-actions · 2023-03-30T11:12:23Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9979/

flow._C.frac

5f2e23c

ccssu requested review from hjchen2, BBuf, jackalcooper and liujuncheng as code owners March 13, 2023 02:02

BBuf reviewed Mar 13, 2023

View reviewed changes

fix oneflow.frac

e767801

ccssu requested a review from daquexian as a code owner March 14, 2023 01:22

add_docstr oneflow.frac

3f7a96b

ccssu requested a review from doombeaker as a code owner March 14, 2023 01:42

ccssu added 5 commits March 15, 2023 00:07

support oneflow.frac & oneflow.frac_

2822ee0

support tensor.frac

23ef5f9

fix impl::FracGradFunctor in oneflow/core/functional/impl/activation_…

3c76313

…functor.cpp

[add] test frac_op

59bce36

[add] test frac_op

ace4150

ccssu and others added 2 commits March 16, 2023 10:25

Merge branch 'master' into dev_add_frac_op

00ac7d5

[add] docs frac frac_

1f3c507

hhhfccz reviewed Mar 16, 2023

View reviewed changes

oneflow/core/autograd/gradient_funcs/activation.cpp Outdated Show resolved Hide resolved

hhhfccz reviewed Mar 16, 2023

View reviewed changes

oneflow/core/functional/impl/activation_functor.cpp Outdated Show resolved Hide resolved

hhhfccz reviewed Mar 16, 2023

View reviewed changes

oneflow/user/kernels/frac_kernel.cpp Outdated Show resolved Hide resolved

hhhfccz reviewed Mar 16, 2023

View reviewed changes

oneflow/core/functional/impl/activation_functor.cpp Outdated Show resolved Hide resolved

ccssu added 3 commits March 16, 2023 03:18

ninja of_format for frac_op

d029c9c

fix add_frac_op

ea38c12

fix add_frac_op

a85c347

BBuf reviewed Mar 17, 2023

View reviewed changes

oneflow/core/autograd/gradient_funcs/activation.cpp Outdated Show resolved Hide resolved

BBuf approved these changes Mar 17, 2023

View reviewed changes

Update oneflow/core/autograd/gradient_funcs/activation.cpp

9362549

Co-authored-by: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>

BBuf removed the request for review from oneflow-ci-bot March 20, 2023 02:33

Merge branch 'master' into dev_add_frac_op

7262db2

BBuf requested a review from oneflow-ci-bot March 20, 2023 02:34

BBuf added enhancement automerge eager api labels Mar 20, 2023

Merge branch 'master' into dev_add_frac_op

00326c9

github-actions bot removed the automerge label Mar 20, 2023

Merge branch 'master' into dev_add_frac_op

bf889c9

ccssu requested review from oneflow-ci-bot and removed request for oneflow-ci-bot March 23, 2023 09:16

Merge branch 'master' into dev_add_frac_op

c20016a

Merge branch 'master' into dev_add_frac_op

9ac4f6c

Merge branch 'master' into dev_add_frac_op

042a966

ccssu and others added 2 commits March 30, 2023 07:35

[fix] frac document

12133fa

Merge branch 'master' into dev_add_frac_op

064a896

ccssu enabled auto-merge (squash) March 30, 2023 07:36

ccssu merged commit a0f0c1c into master Mar 30, 2023

ccssu deleted the dev_add_frac_op branch March 30, 2023 11:30

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

flow._C.frac #9979

flow._C.frac #9979

ccssu commented Mar 13, 2023 •

edited

Loading

CLAassistant commented Mar 13, 2023 •

edited

Loading

BBuf left a comment

ccssu commented Mar 13, 2023

ccssu commented Mar 16, 2023

github-actions bot commented Mar 20, 2023

github-actions bot commented Mar 20, 2023

github-actions bot commented Mar 20, 2023

github-actions bot commented Mar 21, 2023

github-actions bot commented Mar 24, 2023

github-actions bot commented Mar 24, 2023

github-actions bot commented Mar 25, 2023

github-actions bot commented Mar 28, 2023

github-actions bot commented Mar 30, 2023

github-actions bot commented Mar 30, 2023

flow._C.frac #9979

flow._C.frac #9979

Conversation

ccssu commented Mar 13, 2023 • edited Loading

CLAassistant commented Mar 13, 2023 • edited Loading

BBuf left a comment

Choose a reason for hiding this comment

ccssu commented Mar 13, 2023

ccssu commented Mar 16, 2023

github-actions bot commented Mar 20, 2023

github-actions bot commented Mar 20, 2023

github-actions bot commented Mar 20, 2023

github-actions bot commented Mar 21, 2023

github-actions bot commented Mar 24, 2023

github-actions bot commented Mar 24, 2023

github-actions bot commented Mar 25, 2023

github-actions bot commented Mar 28, 2023

github-actions bot commented Mar 30, 2023

github-actions bot commented Mar 30, 2023

ccssu commented Mar 13, 2023 •

edited

Loading

CLAassistant commented Mar 13, 2023 •

edited

Loading