Dev add bitwise shift op #9860

marigoold · 2023-02-13T08:44:26Z

…neflow into dev_add_bitwise_op

github-actions · 2023-03-07T09:56:54Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-03-07T09:57:52Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.5ms (= 14153.2ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 145.0ms (= 14501.3ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.02 (= 145.0ms / 141.5ms)

OneFlow resnet50 time: 84.2ms (= 8425.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 87.9ms (= 8788.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.04 (= 87.9ms / 84.2ms)

OneFlow resnet50 time: 51.2ms (= 10237.7ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 59.0ms (= 11803.2ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.15 (= 59.0ms / 51.2ms)

OneFlow resnet50 time: 33.8ms (= 6753.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 45.6ms (= 9127.5ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.35 (= 45.6ms / 33.8ms)

OneFlow resnet50 time: 25.3ms (= 5063.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 37.7ms (= 7544.8ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.49 (= 37.7ms / 25.3ms)

OneFlow swin dataloader time: 0.235s (= 47.085s / 200, num_workers=1)
PyTorch swin dataloader time: 0.150s (= 29.954s / 200, num_workers=1)
Relative speed: 0.636 (= 0.150s / 0.235s)

OneFlow swin dataloader time: 0.068s (= 13.681s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.301s / 200, num_workers=4)
Relative speed: 0.607 (= 0.042s / 0.068s)

OneFlow swin dataloader time: 0.047s (= 9.379s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.425s / 200, num_workers=8)
Relative speed: 0.472 (= 0.022s / 0.047s)

❌ OneFlow resnet50 time: 155.6ms (= 15564.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 166.4ms (= 16644.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.07 (= 166.4ms / 155.6ms)

OneFlow resnet50 time: 95.2ms (= 9519.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 104.6ms (= 10457.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.10 (= 104.6ms / 95.2ms)

OneFlow resnet50 time: 61.9ms (= 12385.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 83.6ms (= 16726.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 83.6ms / 61.9ms)

OneFlow resnet50 time: 44.1ms (= 8826.5ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 70.5ms (= 14107.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.60 (= 70.5ms / 44.1ms)

OneFlow resnet50 time: 37.5ms (= 7501.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.6ms (= 13325.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.78 (= 66.6ms / 37.5ms)

github-actions · 2023-03-07T12:24:56Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-03-07T16:43:04Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.2ms (= 14116.4ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 144.6ms (= 14459.8ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.02 (= 144.6ms / 141.2ms)

OneFlow resnet50 time: 82.8ms (= 8276.3ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 89.0ms (= 8898.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.08 (= 89.0ms / 82.8ms)

OneFlow resnet50 time: 51.1ms (= 10223.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 59.8ms (= 11956.5ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.17 (= 59.8ms / 51.1ms)

OneFlow resnet50 time: 33.7ms (= 6730.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 42.8ms (= 8564.9ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.27 (= 42.8ms / 33.7ms)

OneFlow resnet50 time: 25.8ms (= 5153.2ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 38.1ms (= 7625.7ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.48 (= 38.1ms / 25.8ms)

OneFlow swin dataloader time: 0.236s (= 47.214s / 200, num_workers=1)
PyTorch swin dataloader time: 0.150s (= 30.072s / 200, num_workers=1)
Relative speed: 0.637 (= 0.150s / 0.236s)

OneFlow swin dataloader time: 0.068s (= 13.514s / 200, num_workers=4)
PyTorch swin dataloader time: 0.042s (= 8.322s / 200, num_workers=4)
Relative speed: 0.616 (= 0.042s / 0.068s)

OneFlow swin dataloader time: 0.038s (= 7.503s / 200, num_workers=8)
PyTorch swin dataloader time: 0.023s (= 4.595s / 200, num_workers=8)
Relative speed: 0.612 (= 0.023s / 0.038s)

❌ OneFlow resnet50 time: 153.8ms (= 15380.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 165.5ms (= 16551.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.08 (= 165.5ms / 153.8ms)

OneFlow resnet50 time: 93.9ms (= 9394.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 104.2ms (= 10416.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.11 (= 104.2ms / 93.9ms)

OneFlow resnet50 time: 61.2ms (= 12249.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 83.9ms (= 16774.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.37 (= 83.9ms / 61.2ms)

OneFlow resnet50 time: 43.1ms (= 8614.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 76.5ms (= 15308.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.78 (= 76.5ms / 43.1ms)

OneFlow resnet50 time: 35.8ms (= 7162.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.4ms (= 13687.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.91 (= 68.4ms / 35.8ms)

github-actions · 2023-03-08T01:39:11Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 141.1ms (= 14108.5ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 145.1ms (= 14505.2ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.03 (= 145.1ms / 141.1ms)

OneFlow resnet50 time: 81.4ms (= 8138.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 85.5ms (= 8545.8ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.05 (= 85.5ms / 81.4ms)

OneFlow resnet50 time: 50.4ms (= 10085.5ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.1ms (= 11610.9ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.15 (= 58.1ms / 50.4ms)

OneFlow resnet50 time: 34.1ms (= 6828.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 46.3ms (= 9268.7ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.36 (= 46.3ms / 34.1ms)

OneFlow resnet50 time: 25.7ms (= 5137.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 35.7ms (= 7145.1ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.39 (= 35.7ms / 25.7ms)

OneFlow swin dataloader time: 0.241s (= 48.146s / 200, num_workers=1)
PyTorch swin dataloader time: 0.153s (= 30.580s / 200, num_workers=1)
Relative speed: 0.635 (= 0.153s / 0.241s)

OneFlow swin dataloader time: 0.065s (= 12.911s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.213s / 200, num_workers=4)
Relative speed: 0.636 (= 0.041s / 0.065s)

OneFlow swin dataloader time: 0.046s (= 9.140s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.409s / 200, num_workers=8)
Relative speed: 0.482 (= 0.022s / 0.046s)

❌ OneFlow resnet50 time: 152.9ms (= 15287.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 167.6ms (= 16763.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.10 (= 167.6ms / 152.9ms)

OneFlow resnet50 time: 92.6ms (= 9258.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 108.3ms (= 10832.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.17 (= 108.3ms / 92.6ms)

OneFlow resnet50 time: 60.3ms (= 12059.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 85.3ms (= 17059.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.41 (= 85.3ms / 60.3ms)

OneFlow resnet50 time: 42.5ms (= 8507.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 73.9ms (= 14789.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.74 (= 73.9ms / 42.5ms)

OneFlow resnet50 time: 36.3ms (= 7266.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 72.0ms (= 14395.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.98 (= 72.0ms / 36.3ms)

github-actions · 2023-03-08T01:53:01Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-03-28T02:52:21Z

Speed stats:

GPU Name: GeForce GTX 1080 

❌ OneFlow resnet50 time: 140.9ms (= 14089.4ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.1ms (= 14308.0ms / 100, input_shape=[16, 3, 224, 224])
❌ Relative speed: 1.02 (= 143.1ms / 140.9ms)

OneFlow resnet50 time: 80.9ms (= 8085.2ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 85.6ms (= 8559.4ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.06 (= 85.6ms / 80.9ms)

OneFlow resnet50 time: 50.1ms (= 10025.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 61.0ms (= 12209.6ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.22 (= 61.0ms / 50.1ms)

OneFlow resnet50 time: 32.9ms (= 6588.0ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 40.5ms (= 8103.9ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.23 (= 40.5ms / 32.9ms)

OneFlow resnet50 time: 25.3ms (= 5063.8ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 37.2ms (= 7433.1ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.47 (= 37.2ms / 25.3ms)

OneFlow swin dataloader time: 0.240s (= 47.917s / 200, num_workers=1)
PyTorch swin dataloader time: 0.150s (= 29.910s / 200, num_workers=1)
Relative speed: 0.624 (= 0.150s / 0.240s)

OneFlow swin dataloader time: 0.069s (= 13.881s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.188s / 200, num_workers=4)
Relative speed: 0.590 (= 0.041s / 0.069s)

OneFlow swin dataloader time: 0.045s (= 9.015s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.494s / 200, num_workers=8)
Relative speed: 0.499 (= 0.022s / 0.045s)

❌ OneFlow resnet50 time: 152.5ms (= 15252.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 164.9ms (= 16489.3ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.08 (= 164.9ms / 152.5ms)

OneFlow resnet50 time: 91.5ms (= 9154.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.5ms (= 10145.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.11 (= 101.5ms / 91.5ms)

OneFlow resnet50 time: 60.3ms (= 12051.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 79.5ms (= 15892.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.32 (= 79.5ms / 60.3ms)

OneFlow resnet50 time: 41.9ms (= 8388.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 82.3ms (= 16454.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.96 (= 82.3ms / 41.9ms)

OneFlow resnet50 time: 36.8ms (= 7362.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.8ms (= 15555.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 2.11 (= 77.8ms / 36.8ms)

github-actions · 2023-03-28T03:04:00Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-08-07T13:41:07Z

Speed stats:

github-actions · 2023-08-09T07:11:44Z

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

github-actions · 2023-08-09T07:12:15Z

Speed stats:

…-Inc/oneflow into dev_add_bitwise_shift_op

github-actions · 2023-08-09T07:57:08Z

Speed stats:

github-actions · 2023-08-09T08:59:10Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-08-09T09:04:33Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.9ms (= 4393.3ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 61.1ms (= 6111.9ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.39 (= 61.1ms / 43.9ms)

OneFlow resnet50 time: 26.1ms (= 2605.4ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.6ms (= 3763.0ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.44 (= 37.6ms / 26.1ms)

OneFlow resnet50 time: 18.6ms (= 3724.5ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 35.4ms (= 7074.1ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.90 (= 35.4ms / 18.6ms)

OneFlow resnet50 time: 19.1ms (= 3828.9ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 31.4ms (= 6274.3ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.64 (= 31.4ms / 19.1ms)

OneFlow resnet50 time: 18.2ms (= 3649.8ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 30.0ms (= 5991.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.64 (= 30.0ms / 18.2ms)

OneFlow swin dataloader time: 0.201s (= 40.166s / 200, num_workers=1)
PyTorch swin dataloader time: 0.129s (= 25.861s / 200, num_workers=1)
Relative speed: 0.644 (= 0.129s / 0.201s)

OneFlow swin dataloader time: 0.054s (= 10.850s / 200, num_workers=4)
PyTorch swin dataloader time: 0.033s (= 6.527s / 200, num_workers=4)
Relative speed: 0.602 (= 0.033s / 0.054s)

OneFlow swin dataloader time: 0.030s (= 6.050s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.365s / 200, num_workers=8)
Relative speed: 0.556 (= 0.017s / 0.030s)

❌ OneFlow resnet50 time: 47.5ms (= 4748.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.1ms (= 6407.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 64.1ms / 47.5ms)

OneFlow resnet50 time: 31.2ms (= 3115.8ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 43.8ms (= 4380.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.41 (= 43.8ms / 31.2ms)

OneFlow resnet50 time: 24.2ms (= 4846.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 42.7ms (= 8549.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.76 (= 42.7ms / 24.2ms)

OneFlow resnet50 time: 23.0ms (= 4601.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 36.2ms (= 7232.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.57 (= 36.2ms / 23.0ms)

OneFlow resnet50 time: 19.0ms (= 3798.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 34.0ms (= 6796.4ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.79 (= 34.0ms / 19.0ms)

github-actions · 2023-08-10T03:13:59Z

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-08-10T03:23:25Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.5ms (= 4348.2ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 61.3ms (= 6128.1ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.41 (= 61.3ms / 43.5ms)

OneFlow resnet50 time: 26.0ms (= 2600.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 38.7ms (= 3873.1ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.49 (= 38.7ms / 26.0ms)

OneFlow resnet50 time: 19.9ms (= 3973.4ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 35.1ms (= 7026.8ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.77 (= 35.1ms / 19.9ms)

OneFlow resnet50 time: 17.4ms (= 3473.2ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 32.3ms (= 6462.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.86 (= 32.3ms / 17.4ms)

OneFlow resnet50 time: 17.4ms (= 3484.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.6ms (= 5914.2ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.70 (= 29.6ms / 17.4ms)

OneFlow swin dataloader time: 0.200s (= 40.037s / 200, num_workers=1)
PyTorch swin dataloader time: 0.130s (= 25.905s / 200, num_workers=1)
Relative speed: 0.647 (= 0.130s / 0.200s)

OneFlow swin dataloader time: 0.056s (= 11.282s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.492s / 200, num_workers=4)
Relative speed: 0.575 (= 0.032s / 0.056s)

OneFlow swin dataloader time: 0.032s (= 6.339s / 200, num_workers=8)
PyTorch swin dataloader time: 0.016s (= 3.293s / 200, num_workers=8)
Relative speed: 0.519 (= 0.016s / 0.032s)

❌ OneFlow resnet50 time: 47.5ms (= 4751.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 65.8ms (= 6576.3ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.38 (= 65.8ms / 47.5ms)

OneFlow resnet50 time: 30.6ms (= 3061.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 43.5ms (= 4351.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.42 (= 43.5ms / 30.6ms)

OneFlow resnet50 time: 24.0ms (= 4801.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 40.5ms (= 8094.3ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.69 (= 40.5ms / 24.0ms)

OneFlow resnet50 time: 22.1ms (= 4426.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 36.7ms (= 7340.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.66 (= 36.7ms / 22.1ms)

OneFlow resnet50 time: 20.7ms (= 4132.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 34.1ms (= 6813.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.65 (= 34.1ms / 20.7ms)

github-actions · 2023-09-05T03:17:10Z

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/9860/

github-actions · 2023-09-05T03:21:40Z

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.7ms (= 4366.7ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 61.8ms (= 6176.7ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.41 (= 61.8ms / 43.7ms)

OneFlow resnet50 time: 26.0ms (= 2603.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.9ms (= 3791.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.46 (= 37.9ms / 26.0ms)

OneFlow resnet50 time: 20.5ms (= 4109.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 36.1ms (= 7219.4ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.76 (= 36.1ms / 20.5ms)

OneFlow resnet50 time: 18.5ms (= 3709.2ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 32.5ms (= 6501.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.75 (= 32.5ms / 18.5ms)

OneFlow resnet50 time: 17.9ms (= 3584.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.7ms (= 5930.1ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.65 (= 29.7ms / 17.9ms)

OneFlow swin dataloader time: 0.201s (= 40.263s / 200, num_workers=1)
PyTorch swin dataloader time: 0.130s (= 25.972s / 200, num_workers=1)
Relative speed: 0.645 (= 0.130s / 0.201s)

OneFlow swin dataloader time: 0.055s (= 11.065s / 200, num_workers=4)
PyTorch swin dataloader time: 0.033s (= 6.528s / 200, num_workers=4)
Relative speed: 0.590 (= 0.033s / 0.055s)

OneFlow swin dataloader time: 0.030s (= 5.941s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.310s / 200, num_workers=8)
Relative speed: 0.557 (= 0.017s / 0.030s)

❌ OneFlow resnet50 time: 47.5ms (= 4747.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 63.8ms (= 6383.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 63.8ms / 47.5ms)

OneFlow resnet50 time: 30.9ms (= 3092.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 43.3ms (= 4328.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.40 (= 43.3ms / 30.9ms)

OneFlow resnet50 time: 24.0ms (= 4808.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 42.2ms (= 8432.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.75 (= 42.2ms / 24.0ms)

OneFlow resnet50 time: 22.1ms (= 4428.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 36.6ms (= 7323.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.65 (= 36.6ms / 22.1ms)

OneFlow resnet50 time: 21.2ms (= 4242.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 34.0ms (= 6790.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.60 (= 34.0ms / 21.2ms)

marigoold and others added 23 commits February 7, 2023 18:45

add bitwise add kernel

c91039f

add bitwise or xor kernel

6b9f567

add scalar ops

8cc122c

support cuda

1c3ad4d

add unittest

c968f12

add doc

c9d3140

add tensor functions and docs

283f524

add global test

ce8766a

Update math_functor.cpp

368ee38

Merge branch 'master' into dev_add_bitwise_op

4279303

auto format by CI

9e1dc2e

Merge branch 'master' into dev_add_bitwise_op

2075f60

refine

3390f8c

Merge branch 'dev_add_bitwise_op' of https://github.com/Oneflow-Inc/o…

f4d06ba

…neflow into dev_add_bitwise_op

fix bug in global unittest

ac33b58

Merge branch 'master' into dev_add_bitwise_op

45d31fb

Update test_bitwise.py

3092ac5

Update test_bitwise.py

ac40a62

Update test_bitwise.py

1cdfc62

auto format by CI

ef8a0ba

Merge branch 'master' into dev_add_bitwise_op

ef88a86

refine bug in unittest

0d40e7d

add shift ops

66d6fb1

marigoold changed the base branch from master to dev_add_bitwise_op February 13, 2023 08:44

Base automatically changed from dev_add_bitwise_op to master February 13, 2023 09:54

marigoold and others added 5 commits February 13, 2023 22:51

add test, add doc, fix bug of registering kernels

3593bf6

Merge branch 'master' into dev_add_bitwise_shift_op

2d9150b

merge master

f542c57

refine oneflowuserops.td

11158f2

refine docs

65802fd

Update test_global_bitwise.py

f040f08

merge master

319c2de

marigoold added 2 commits August 7, 2023 19:40

merge master

a34ed87

format code and fix bug

70b528c

marigoold added the automerge label Aug 7, 2023

Merge branch 'master' into dev_add_bitwise_shift_op

4a23ccd

github-actions bot removed the automerge label Aug 9, 2023

marigoold added 2 commits August 9, 2023 15:22

refine var name in ir

8353247

Merge branch 'dev_add_bitwise_shift_op' of https://github.com/Oneflow…

88fbbc8

…-Inc/oneflow into dev_add_bitwise_shift_op

undo modify of global bitwise op test

6f93193

marigoold disabled auto-merge August 10, 2023 03:04

Merge branch 'master' into dev_add_bitwise_shift_op

befbb53

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Dev add bitwise shift op #9860

Dev add bitwise shift op #9860

marigoold commented Feb 13, 2023 •

edited

Loading

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 8, 2023

github-actions bot commented Mar 8, 2023

github-actions bot commented Mar 28, 2023

github-actions bot commented Mar 28, 2023

github-actions bot commented Aug 7, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 10, 2023

github-actions bot commented Aug 10, 2023

github-actions bot commented Sep 5, 2023

github-actions bot commented Sep 5, 2023

Dev add bitwise shift op #9860

Are you sure you want to change the base?

Dev add bitwise shift op #9860

Conversation

marigoold commented Feb 13, 2023 • edited Loading

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 7, 2023

github-actions bot commented Mar 8, 2023

github-actions bot commented Mar 8, 2023

github-actions bot commented Mar 28, 2023

github-actions bot commented Mar 28, 2023

github-actions bot commented Aug 7, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 9, 2023

github-actions bot commented Aug 10, 2023

github-actions bot commented Aug 10, 2023

github-actions bot commented Sep 5, 2023

github-actions bot commented Sep 5, 2023

marigoold commented Feb 13, 2023 •

edited

Loading