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

Add flow amin #8042

Merged
merged 54 commits into from
Apr 28, 2022
Merged

Add flow amin #8042

merged 54 commits into from
Apr 28, 2022

Conversation

zhongshsh
Copy link
Contributor

@zhongshsh zhongshsh commented Apr 18, 2022 •
edited
Loading

添加 flow.amin 和 flow.Tensor.amin。

image

@BBuf BBuf requested a review from oneflow-ci-bot April 19, 2022 07:14
@github-actions
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8042/

@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.5ms (= 12854.4ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 141.0ms (= 14100.4ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.10 (= 141.0ms / 128.5ms)

OneFlow resnet50 time: 81.3ms (= 8126.8ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.9ms (= 8486.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.04 (= 84.9ms / 81.3ms)

OneFlow resnet50 time: 47.7ms (= 9543.0ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 55.7ms (= 11135.3ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.17 (= 55.7ms / 47.7ms)

OneFlow resnet50 time: 42.9ms (= 8573.7ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 40.6ms (= 8115.1ms / 200, input_shape=[2, 3, 224, 224])
❌ Relative speed: 0.95 (= 40.6ms / 42.9ms)

OneFlow resnet50 time: 34.7ms (= 6947.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 42.8ms (= 8560.5ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.23 (= 42.8ms / 34.7ms)

OneFlow swin dataloader time: 0.264s (= 52.731s / 200, num_workers=1)
PyTorch swin dataloader time: 0.151s (= 30.158s / 200, num_workers=1)
Relative speed: 0.572 (= 0.151s / 0.264s)

OneFlow swin dataloader time: 0.067s (= 13.450s / 200, num_workers=4)
PyTorch swin dataloader time: 0.040s (= 8.037s / 200, num_workers=4)
Relative speed: 0.598 (= 0.040s / 0.067s)

OneFlow swin dataloader time: 0.036s (= 7.202s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.414s / 200, num_workers=8)
Relative speed: 0.613 (= 0.022s / 0.036s)

✔️ OneFlow resnet50 time: 135.8ms (= 13579.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 159.3ms (= 15929.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.17 (= 159.3ms / 135.8ms)

OneFlow resnet50 time: 89.8ms (= 8982.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 101.6ms (= 10156.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 101.6ms / 89.8ms)

OneFlow resnet50 time: 64.2ms (= 12847.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 76.8ms (= 15360.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.20 (= 76.8ms / 64.2ms)

OneFlow resnet50 time: 52.8ms (= 10566.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.8ms (= 13760.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.30 (= 68.8ms / 52.8ms)

OneFlow resnet50 time: 47.7ms (= 9539.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.9ms (= 13375.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.40 (= 66.9ms / 47.7ms)

@github-actions
Copy link
Contributor

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

@zhongshsh zhongshsh requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 27, 2022 12:58
@github-actions
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8042/

@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.5ms (= 12846.6ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.0ms (= 14303.7ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.11 (= 143.0ms / 128.5ms)

OneFlow resnet50 time: 77.0ms (= 7699.7ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 86.0ms (= 8602.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.12 (= 86.0ms / 77.0ms)

OneFlow resnet50 time: 51.6ms (= 10314.1ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 53.0ms (= 10609.0ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.03 (= 53.0ms / 51.6ms)

OneFlow resnet50 time: 40.6ms (= 8114.2ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 44.2ms (= 8839.2ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.09 (= 44.2ms / 40.6ms)

OneFlow resnet50 time: 36.3ms (= 7251.8ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 47.5ms (= 9500.3ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.31 (= 47.5ms / 36.3ms)

OneFlow swin dataloader time: 0.250s (= 50.027s / 200, num_workers=1)
PyTorch swin dataloader time: 0.149s (= 29.751s / 200, num_workers=1)
Relative speed: 0.595 (= 0.149s / 0.250s)

OneFlow swin dataloader time: 0.066s (= 13.236s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.205s / 200, num_workers=4)
Relative speed: 0.620 (= 0.041s / 0.066s)

OneFlow swin dataloader time: 0.036s (= 7.273s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.398s / 200, num_workers=8)
Relative speed: 0.605 (= 0.022s / 0.036s)

✔️ OneFlow resnet50 time: 135.7ms (= 13568.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 160.1ms (= 16005.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.18 (= 160.1ms / 135.7ms)

OneFlow resnet50 time: 89.7ms (= 8969.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 100.5ms (= 10049.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.12 (= 100.5ms / 89.7ms)

OneFlow resnet50 time: 63.8ms (= 12767.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 87.7ms (= 17531.4ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.37 (= 87.7ms / 63.8ms)

OneFlow resnet50 time: 53.4ms (= 10677.1ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.8ms (= 13967.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.31 (= 69.8ms / 53.4ms)

OneFlow resnet50 time: 48.8ms (= 9765.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 69.8ms (= 13967.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.43 (= 69.8ms / 48.8ms)

@github-actions
Copy link
Contributor

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

@github-actions
Copy link
Contributor

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
Copy link
Contributor

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/8042/

@github-actions
Copy link
Contributor

Speed stats: 
GPU Name: NVIDIA GeForce GTX 1080 

❌ OneFlow resnet50 time: 129.4ms (= 12942.7ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 143.0ms (= 14297.9ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.10 (= 143.0ms / 129.4ms)

OneFlow resnet50 time: 84.7ms (= 8468.1ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 84.4ms (= 8439.9ms / 100, input_shape=[8, 3, 224, 224])
❌ Relative speed: 1.00 (= 84.4ms / 84.7ms)

OneFlow resnet50 time: 54.0ms (= 10805.6ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.5ms (= 11693.7ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.08 (= 58.5ms / 54.0ms)

OneFlow resnet50 time: 41.8ms (= 8366.2ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 42.3ms (= 8463.2ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.01 (= 42.3ms / 41.8ms)

OneFlow resnet50 time: 34.3ms (= 6868.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 36.0ms (= 7208.0ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.05 (= 36.0ms / 34.3ms)

OneFlow swin dataloader time: 0.259s (= 51.823s / 200, num_workers=1)
PyTorch swin dataloader time: 0.152s (= 30.332s / 200, num_workers=1)
Relative speed: 0.585 (= 0.152s / 0.259s)

OneFlow swin dataloader time: 0.067s (= 13.382s / 200, num_workers=4)
PyTorch swin dataloader time: 0.041s (= 8.253s / 200, num_workers=4)
Relative speed: 0.617 (= 0.041s / 0.067s)

OneFlow swin dataloader time: 0.037s (= 7.460s / 200, num_workers=8)
PyTorch swin dataloader time: 0.022s (= 4.387s / 200, num_workers=8)
Relative speed: 0.588 (= 0.022s / 0.037s)

❌ OneFlow resnet50 time: 144.8ms (= 14482.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 169.4ms (= 16943.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.17 (= 169.4ms / 144.8ms)

OneFlow resnet50 time: 102.1ms (= 10208.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 111.9ms (= 11185.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.10 (= 111.9ms / 102.1ms)

OneFlow resnet50 time: 77.0ms (= 15408.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 98.2ms (= 19638.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.27 (= 98.2ms / 77.0ms)

OneFlow resnet50 time: 63.9ms (= 12772.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 73.7ms (= 14740.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
❌ Relative speed: 1.15 (= 73.7ms / 63.9ms)

OneFlow resnet50 time: 56.8ms (= 11360.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 80.5ms (= 16091.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.42 (= 80.5ms / 56.8ms)

@zhongshsh zhongshsh merged commit bc4d9f6 into master Apr 28, 2022
@zhongshsh zhongshsh deleted the add_flow_amin branch April 28, 2022 18:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@Flowingsun007 Flowingsun007 Flowingsun007 approved these changes

@BBuf BBuf BBuf approved these changes

@doombeaker doombeaker Awaiting requested review from doombeaker doombeaker is a code owner

@jackalcooper jackalcooper Awaiting requested review from jackalcooper

@hjchen2 hjchen2 Awaiting requested review from hjchen2 hjchen2 is a code owner

@daquexian daquexian Awaiting requested review from daquexian

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

Assignees
No one assigned
Labels
api automerge eager feature op
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@zhongshsh @Flowingsun007 @BBuf @oneflow-ci-bot