Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

align torch new tensor #7973

Merged
merged 30 commits into from
Apr 16, 2022
Merged

align torch new tensor #7973

merged 30 commits into from
Apr 16, 2022

Conversation

liufengwei0103
Copy link
Contributor

No description provided.

@liufengwei0103 liufengwei0103 requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 12, 2022 03:56
@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.9ms (= 12891.0ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 147.2ms (= 14725.0ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.14 (= 147.2ms / 128.9ms)

OneFlow resnet50 time: 79.9ms (= 7989.8ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 85.6ms (= 8561.9ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.07 (= 85.6ms / 79.9ms)

OneFlow resnet50 time: 55.6ms (= 11125.9ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 64.3ms (= 12865.3ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.16 (= 64.3ms / 55.6ms)

OneFlow resnet50 time: 46.9ms (= 9386.4ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 44.7ms (= 8935.3ms / 200, input_shape=[2, 3, 224, 224])
❌ Relative speed: 0.95 (= 44.7ms / 46.9ms)

OneFlow resnet50 time: 40.3ms (= 8069.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 42.6ms (= 8517.4ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.06 (= 42.6ms / 40.3ms)

OneFlow swin dataloader time: 0.253s (= 50.542s / 200, num_workers=1)
PyTorch swin dataloader time: 0.254s (= 50.860s / 200, num_workers=1)
✔️ Relative speed: 1.006 (= 0.254s / 0.253s)

OneFlow swin dataloader time: 0.067s (= 13.339s / 200, num_workers=4)
PyTorch swin dataloader time: 0.067s (= 13.329s / 200, num_workers=4)
✔️ Relative speed: 0.999 (= 0.067s / 0.067s)

OneFlow swin dataloader time: 0.036s (= 7.209s / 200, num_workers=8)
PyTorch swin dataloader time: 0.037s (= 7.338s / 200, num_workers=8)
✔️ Relative speed: 1.018 (= 0.037s / 0.036s)

✔️ OneFlow resnet50 time: 135.5ms (= 13548.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 157.3ms (= 15729.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.16 (= 157.3ms / 135.5ms)

OneFlow resnet50 time: 83.7ms (= 8366.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 102.1ms (= 10209.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.22 (= 102.1ms / 83.7ms)

OneFlow resnet50 time: 61.7ms (= 12335.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.5ms (= 15498.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.26 (= 77.5ms / 61.7ms)

OneFlow resnet50 time: 53.1ms (= 10613.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.6ms (= 13725.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.29 (= 68.6ms / 53.1ms)

OneFlow resnet50 time: 47.5ms (= 9507.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 63.2ms (= 12642.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.33 (= 63.2ms / 47.5ms)

@github-actions
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/7973/

@github-actions
Copy link
Contributor

Speed stats:

@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 129.0ms (= 12897.9ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 141.0ms (= 14097.5ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.09 (= 141.0ms / 129.0ms)

OneFlow resnet50 time: 80.1ms (= 8011.2ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 88.2ms (= 8816.4ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.10 (= 88.2ms / 80.1ms)

OneFlow resnet50 time: 53.3ms (= 10660.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 63.1ms (= 12619.3ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.18 (= 63.1ms / 53.3ms)

OneFlow resnet50 time: 44.6ms (= 8911.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 44.7ms (= 8943.5ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.00 (= 44.7ms / 44.6ms)

OneFlow resnet50 time: 40.0ms (= 7997.0ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 43.1ms (= 8616.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.08 (= 43.1ms / 40.0ms)

OneFlow swin dataloader time: 0.251s (= 50.299s / 200, num_workers=1)
PyTorch swin dataloader time: 0.268s (= 53.566s / 200, num_workers=1)
✔️ Relative speed: 1.065 (= 0.268s / 0.251s)

OneFlow swin dataloader time: 0.068s (= 13.641s / 200, num_workers=4)
PyTorch swin dataloader time: 0.067s (= 13.310s / 200, num_workers=4)
✔️ Relative speed: 0.976 (= 0.067s / 0.068s)

OneFlow swin dataloader time: 0.036s (= 7.237s / 200, num_workers=8)
PyTorch swin dataloader time: 0.037s (= 7.438s / 200, num_workers=8)
✔️ Relative speed: 1.028 (= 0.037s / 0.036s)

✔️ OneFlow resnet50 time: 135.8ms (= 13577.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 157.8ms (= 15778.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.16 (= 157.8ms / 135.8ms)

OneFlow resnet50 time: 90.0ms (= 9000.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.9ms (= 10187.8ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 101.9ms / 90.0ms)

OneFlow resnet50 time: 61.0ms (= 12206.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 86.4ms (= 17274.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.42 (= 86.4ms / 61.0ms)

OneFlow resnet50 time: 52.4ms (= 10470.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 67.2ms (= 13446.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.28 (= 67.2ms / 52.4ms)

OneFlow resnet50 time: 47.3ms (= 9462.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 72.5ms (= 14508.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.53 (= 72.5ms / 47.3ms)

@github-actions
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/7973/

@github-actions
Copy link
Contributor

CI failed when running job: cuda-misc. PR label automerge has been removed

@github-actions
Copy link
Contributor

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

@github-actions
Copy link
Contributor

CI failed when running job: cuda-benchmark. PR label automerge has been removed

@liufengwei0103 liufengwei0103 requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 16, 2022 04:12
@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.6ms (= 12862.5ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 142.3ms (= 14227.2ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.11 (= 142.3ms / 128.6ms)

OneFlow resnet50 time: 80.9ms (= 8088.0ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.7ms (= 8468.0ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.05 (= 84.7ms / 80.9ms)

OneFlow resnet50 time: 50.9ms (= 10188.3ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 60.9ms (= 12186.4ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.20 (= 60.9ms / 50.9ms)

OneFlow resnet50 time: 43.5ms (= 8708.5ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 43.4ms (= 8677.0ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.00 (= 43.4ms / 43.5ms)

OneFlow resnet50 time: 38.4ms (= 7675.6ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 39.7ms (= 7939.8ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.03 (= 39.7ms / 38.4ms)

OneFlow swin dataloader time: 0.249s (= 49.821s / 200, num_workers=1)
PyTorch swin dataloader time: 0.250s (= 50.079s / 200, num_workers=1)
✔️ Relative speed: 1.005 (= 0.250s / 0.249s)

OneFlow swin dataloader time: 0.068s (= 13.616s / 200, num_workers=4)
PyTorch swin dataloader time: 0.066s (= 13.173s / 200, num_workers=4)
✔️ Relative speed: 0.967 (= 0.066s / 0.068s)

OneFlow swin dataloader time: 0.037s (= 7.494s / 200, num_workers=8)
PyTorch swin dataloader time: 0.038s (= 7.692s / 200, num_workers=8)
✔️ Relative speed: 1.026 (= 0.038s / 0.037s)

✔️ OneFlow resnet50 time: 135.3ms (= 13530.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 156.7ms (= 15666.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.16 (= 156.7ms / 135.3ms)

OneFlow resnet50 time: 88.5ms (= 8849.6ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 106.7ms (= 10675.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.21 (= 106.7ms / 88.5ms)

OneFlow resnet50 time: 61.9ms (= 12385.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 77.3ms (= 15464.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.25 (= 77.3ms / 61.9ms)

OneFlow resnet50 time: 53.3ms (= 10651.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.8ms (= 13366.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.25 (= 66.8ms / 53.3ms)

OneFlow resnet50 time: 48.3ms (= 9659.4ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 62.3ms (= 12457.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.29 (= 62.3ms / 48.3ms)

@github-actions
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/7973/

@github-actions
Copy link
Contributor

CI failed when running job: cuda-benchmark. PR label automerge has been removed

@liufengwei0103 liufengwei0103 requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 16, 2022 14:04
@mergify mergify bot merged commit 7d91ecc into master Apr 16, 2022
@mergify mergify bot deleted the api_new_tensor branch April 16, 2022 15:17
@BBuf BBuf mentioned this pull request Apr 22, 2022
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@MARD1NO MARD1NO MARD1NO approved these changes

@BBuf BBuf BBuf approved these changes

@oneflow-ci-bot oneflow-ci-bot Awaiting requested review from oneflow-ci-bot

Assignees
No one assigned
Labels
api automerge enhancement python
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@liufengwei0103 @BBuf @MARD1NO