patch

[Flowingsun007]

• patch Optimize slice and tensor getitem #8148 中因为手滑，注释掉的代码段

[github-actions]

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8211/

[github-actions]

Speed stats:

GPU Name: NVIDIA GeForce GTX 1080 

❌ OneFlow resnet50 time: 129.2ms (= 12915.1ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 142.2ms (= 14224.7ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.10 (= 142.2ms / 129.2ms)

OneFlow resnet50 time: 80.6ms (= 8055.1ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 86.2ms (= 8616.8ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.07 (= 86.2ms / 80.6ms)

OneFlow resnet50 time: 53.9ms (= 10784.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 62.4ms (= 12485.6ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.16 (= 62.4ms / 53.9ms)

OneFlow resnet50 time: 44.3ms (= 8852.0ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 45.4ms (= 9080.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.03 (= 45.4ms / 44.3ms)

OneFlow resnet50 time: 39.5ms (= 7900.8ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 43.2ms (= 8645.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.09 (= 43.2ms / 39.5ms)

OneFlow swin dataloader time: 0.257s (= 51.353s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.190s / 200, num_workers=1)
Relative speed: 0.588 (= 0.151s / 0.257s)

OneFlow swin dataloader time: 0.069s (= 13.866s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.475s / 200, num_workers=4)
Relative speed: 0.611 (= 0.042s / 0.069s)

OneFlow swin dataloader time: 0.037s (= 7.480s / 200, num_workers=8)
PyTorch swin dataloader time: 0.021s (= 4.281s / 200, num_workers=8)
Relative speed: 0.572 (= 0.021s / 0.037s)

❌ OneFlow resnet50 time: 144.4ms (= 14437.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 168.8ms (= 16876.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.17 (= 168.8ms / 144.4ms)

OneFlow resnet50 time: 97.4ms (= 9742.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 111.1ms (= 11112.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.14 (= 111.1ms / 97.4ms)

OneFlow resnet50 time: 74.2ms (= 14841.3ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 91.1ms (= 18220.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.23 (= 91.1ms / 74.2ms)

OneFlow resnet50 time: 64.1ms (= 12827.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 85.9ms (= 17180.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 85.9ms / 64.1ms)

OneFlow resnet50 time: 57.9ms (= 11588.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.8ms (= 13962.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.20 (= 69.8ms / 57.9ms)

[github-actions]

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8211/