mv_boxing_folder_to_core

[clackhan]

将oneflow/core/framework/op_interpreter/boxing目录移动到oneflow/core

[lixinqi]

这些地方很好，处理得很细腻。

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

OneFlow resnet50 time: 128.4ms (= 6420.5ms / 50, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 140.4ms (= 7018.9ms / 50, input_shape=[16, 3, 224, 224])
Relative speed: 1.09 (= 140.4ms / 128.4ms)

OneFlow resnet50 time: 74.9ms (= 3743.2ms / 50, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 87.2ms (= 4360.6ms / 50, input_shape=[8, 3, 224, 224])
Relative speed: 1.16 (= 87.2ms / 74.9ms)

OneFlow resnet50 time: 48.3ms (= 2416.1ms / 50, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 59.4ms (= 2971.9ms / 50, input_shape=[4, 3, 224, 224])
Relative speed: 1.23 (= 59.4ms / 48.3ms)

OneFlow resnet50 time: 40.6ms (= 2028.4ms / 50, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 48.7ms (= 2435.9ms / 50, input_shape=[2, 3, 224, 224])
Relative speed: 1.20 (= 48.7ms / 40.6ms)

OneFlow resnet50 time: 38.4ms (= 1918.8ms / 50, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 43.3ms (= 2164.6ms / 50, input_shape=[1, 3, 224, 224])
Relative speed: 1.13 (= 43.3ms / 38.4ms)

OneFlow resnet50 time: 142.5ms (= 7125.7ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.4ms (= 8019.4ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
Relative speed: 1.13 (= 160.4ms / 142.5ms)

OneFlow resnet50 time: 87.8ms (= 4387.6ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 103.9ms (= 5197.3ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
Relative speed: 1.18 (= 103.9ms / 87.8ms)

OneFlow resnet50 time: 72.5ms (= 3624.9ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 75.7ms (= 3782.8ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
Relative speed: 1.04 (= 75.7ms / 72.5ms)

OneFlow resnet50 time: 58.9ms (= 2944.8ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.3ms (= 3214.1ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
Relative speed: 1.09 (= 64.3ms / 58.9ms)

OneFlow resnet50 time: 60.4ms (= 3021.0ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 61.5ms (= 3074.5ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
Relative speed: 1.02 (= 61.5ms / 60.4ms)