not fuse request for nccl 2.10.3

[guo-ran]

No description provided.

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

OneFlow resnet50 time: 127.7ms (= 6382.6ms / 50, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 141.3ms (= 7064.6ms / 50, input_shape=[16, 3, 224, 224])
Relative speed: 1.11 (= 141.3ms / 127.7ms)

OneFlow resnet50 time: 74.3ms (= 3713.4ms / 50, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 85.1ms (= 4257.1ms / 50, input_shape=[8, 3, 224, 224])
Relative speed: 1.15 (= 85.1ms / 74.3ms)

OneFlow resnet50 time: 48.2ms (= 2408.6ms / 50, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 55.0ms (= 2750.8ms / 50, input_shape=[4, 3, 224, 224])
Relative speed: 1.14 (= 55.0ms / 48.2ms)

OneFlow resnet50 time: 39.5ms (= 1974.0ms / 50, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 47.2ms (= 2358.9ms / 50, input_shape=[2, 3, 224, 224])
Relative speed: 1.19 (= 47.2ms / 39.5ms)

OneFlow resnet50 time: 40.9ms (= 2045.2ms / 50, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 40.0ms (= 2002.4ms / 50, input_shape=[1, 3, 224, 224])
Relative speed: 0.98 (= 40.0ms / 40.9ms)

OneFlow resnet50 time: 144.5ms (= 7227.0ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 161.7ms (= 8086.1ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
Relative speed: 1.12 (= 161.7ms / 144.5ms)

OneFlow resnet50 time: 90.6ms (= 4529.6ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.7ms (= 5085.4ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
Relative speed: 1.12 (= 101.7ms / 90.6ms)

OneFlow resnet50 time: 67.6ms (= 3381.9ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 78.5ms (= 3927.1ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
Relative speed: 1.16 (= 78.5ms / 67.6ms)

OneFlow resnet50 time: 66.9ms (= 3346.0ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 65.4ms (= 3270.0ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
Relative speed: 0.98 (= 65.4ms / 66.9ms)

OneFlow resnet50 time: 66.8ms (= 3339.2ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 56.2ms (= 2809.8ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
Relative speed: 0.84 (= 56.2ms / 66.8ms)

[github-actions]

CI failed, removing label automerge

[github-actions]

Speed stats:

GPU Name: GeForce GTX 1080 

OneFlow resnet50 time: 125.5ms (= 6273.7ms / 50, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 138.2ms (= 6910.5ms / 50, input_shape=[16, 3, 224, 224])
Relative speed: 1.10 (= 138.2ms / 125.5ms)

OneFlow resnet50 time: 72.8ms (= 3639.7ms / 50, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 82.1ms (= 4107.3ms / 50, input_shape=[8, 3, 224, 224])
Relative speed: 1.13 (= 82.1ms / 72.8ms)

OneFlow resnet50 time: 47.2ms (= 2357.5ms / 50, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 57.6ms (= 2877.6ms / 50, input_shape=[4, 3, 224, 224])
Relative speed: 1.22 (= 57.6ms / 47.2ms)

OneFlow resnet50 time: 40.0ms (= 1998.8ms / 50, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 48.9ms (= 2443.9ms / 50, input_shape=[2, 3, 224, 224])
Relative speed: 1.22 (= 48.9ms / 40.0ms)

OneFlow resnet50 time: 39.3ms (= 1966.9ms / 50, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.8ms (= 1992.0ms / 50, input_shape=[1, 3, 224, 224])
Relative speed: 1.01 (= 39.8ms / 39.3ms)

OneFlow resnet50 time: 139.5ms (= 6974.0ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 153.8ms (= 7687.6ms / 50, input_shape=[16, 3, 224, 224], ddp, world size=2)
Relative speed: 1.10 (= 153.8ms / 139.5ms)

OneFlow resnet50 time: 91.0ms (= 4552.0ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 100.2ms (= 5011.6ms / 50, input_shape=[8, 3, 224, 224], ddp, world size=2)
Relative speed: 1.10 (= 100.2ms / 91.0ms)

OneFlow resnet50 time: 73.1ms (= 3655.1ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 80.6ms (= 4029.5ms / 50, input_shape=[4, 3, 224, 224], ddp, world size=2)
Relative speed: 1.10 (= 80.6ms / 73.1ms)

OneFlow resnet50 time: 66.5ms (= 3327.1ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.8ms (= 3338.4ms / 50, input_shape=[2, 3, 224, 224], ddp, world size=2)
Relative speed: 1.00 (= 66.8ms / 66.5ms)

OneFlow resnet50 time: 62.9ms (= 3145.2ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 63.7ms (= 3186.6ms / 50, input_shape=[1, 3, 224, 224], ddp, world size=2)
Relative speed: 1.01 (= 63.7ms / 62.9ms)