Wheel skip nccl

[jackalcooper]

No description provided.

[github-actions]

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

[github-actions]

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10452/

[github-actions]

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.5ms (= 4348.1ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 58.2ms (= 5817.1ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.34 (= 58.2ms / 43.5ms)

OneFlow resnet50 time: 26.1ms (= 2612.1ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.6ms (= 3756.9ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.44 (= 37.6ms / 26.1ms)

OneFlow resnet50 time: 18.4ms (= 3679.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 34.8ms (= 6964.1ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.89 (= 34.8ms / 18.4ms)

OneFlow resnet50 time: 17.4ms (= 3487.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 33.5ms (= 6709.9ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.92 (= 33.5ms / 17.4ms)

OneFlow resnet50 time: 16.3ms (= 3269.1ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.8ms (= 5961.3ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.82 (= 29.8ms / 16.3ms)

OneFlow swin dataloader time: 0.200s (= 40.050s / 200, num_workers=1)
PyTorch swin dataloader time: 0.128s (= 25.547s / 200, num_workers=1)
Relative speed: 0.638 (= 0.128s / 0.200s)

OneFlow swin dataloader time: 0.055s (= 11.044s / 200, num_workers=4)
PyTorch swin dataloader time: 0.033s (= 6.554s / 200, num_workers=4)
Relative speed: 0.593 (= 0.033s / 0.055s)

OneFlow swin dataloader time: 0.032s (= 6.367s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.355s / 200, num_workers=8)
Relative speed: 0.527 (= 0.017s / 0.032s)

❌ OneFlow resnet50 time: 49.3ms (= 4931.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.1ms (= 6606.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 66.1ms / 49.3ms)

OneFlow resnet50 time: 35.8ms (= 3579.8ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 44.8ms (= 4483.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.25 (= 44.8ms / 35.8ms)

OneFlow resnet50 time: 27.8ms (= 5555.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.9ms (= 7971.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.43 (= 39.9ms / 27.8ms)

OneFlow resnet50 time: 24.9ms (= 4983.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.2ms (= 7847.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.57 (= 39.2ms / 24.9ms)

OneFlow resnet50 time: 24.9ms (= 4972.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 35.9ms (= 7188.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.45 (= 35.9ms / 24.9ms)