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

Support as tensor op #7855

Merged
merged 54 commits into from
Apr 5, 2022
Merged

Support as tensor op #7855

merged 54 commits into from
Apr 5, 2022

Conversation

BBuf
Copy link
Contributor

@BBuf BBuf commented Mar 22, 2022

  • support as_tensor op，align with pytorch。

docs：

image

MARD1NO
MARD1NO approved these changes Mar 22, 2022
View reviewed changes
@BBuf BBuf requested a review from oneflow-ci-bot March 22, 2022 09:30
@oneflow-ci-bot oneflow-ci-bot requested review from oneflow-ci-bot and removed request for oneflow-ci-bot March 22, 2022 11:22
@BBuf BBuf requested a review from oneflow-ci-bot April 4, 2022 08:46
@github-actions
Copy link
Contributor

github-actions bot commented Apr 4, 2022

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.0ms (= 12803.6ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 139.7ms (= 13972.6ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.09 (= 139.7ms / 128.0ms)

❌ OneFlow resnet50 time: 80.7ms (= 8073.5ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 82.3ms (= 8226.2ms / 100, input_shape=[8, 3, 224, 224])
❌ Relative speed: 1.02 (= 82.3ms / 80.7ms)

OneFlow resnet50 time: 51.9ms (= 10371.2ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 58.2ms (= 11637.0ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.12 (= 58.2ms / 51.9ms)

OneFlow resnet50 time: 43.5ms (= 8695.4ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 48.9ms (= 9781.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.12 (= 48.9ms / 43.5ms)

OneFlow resnet50 time: 38.9ms (= 7770.3ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 36.7ms (= 7333.9ms / 200, input_shape=[1, 3, 224, 224])
❌ Relative speed: 0.94 (= 36.7ms / 38.9ms)

OneFlow swin dataloader time: 0.255s (= 50.975s / 200, num_workers=1)
PyTorch swin dataloader time: 0.252s (= 50.314s / 200, num_workers=1)
✔️ Relative speed: 0.987 (= 0.252s / 0.255s)

OneFlow swin dataloader time: 0.067s (= 13.454s / 200, num_workers=4)
PyTorch swin dataloader time: 0.066s (= 13.196s / 200, num_workers=4)
✔️ Relative speed: 0.981 (= 0.066s / 0.067s)

OneFlow swin dataloader time: 0.039s (= 7.828s / 200, num_workers=8)
PyTorch swin dataloader time: 0.038s (= 7.552s / 200, num_workers=8)
✔️ Relative speed: 0.965 (= 0.038s / 0.039s)

✔️ OneFlow resnet50 time: 135.3ms (= 13528.5ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 155.9ms (= 15585.1ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 155.9ms / 135.3ms)

OneFlow resnet50 time: 87.2ms (= 8723.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 99.3ms (= 9932.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.14 (= 99.3ms / 87.2ms)

OneFlow resnet50 time: 59.6ms (= 11926.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 75.9ms (= 15172.2ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.27 (= 75.9ms / 59.6ms)

OneFlow resnet50 time: 52.6ms (= 10525.7ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 70.9ms (= 14170.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.35 (= 70.9ms / 52.6ms)

OneFlow resnet50 time: 47.7ms (= 9532.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 61.5ms (= 12298.8ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.29 (= 61.5ms / 47.7ms)

@github-actions
Copy link
Contributor

github-actions bot commented Apr 4, 2022

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

@github-actions github-actions bot removed the automerge label Apr 4, 2022
@BBuf BBuf requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 4, 2022 13:15
@BBuf BBuf added the automerge label Apr 4, 2022
@oneflow-ci-bot oneflow-ci-bot requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 5, 2022 01:30
@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

CI failed when running job: cuda-module-distributed-rank-0. PR label automerge has been removed

@github-actions github-actions bot removed the automerge label Apr 5, 2022
@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

CI failed when running job: cuda-module-distributed-rank-1. PR label automerge has been removed

@BBuf BBuf added the automerge label Apr 5, 2022
@BBuf BBuf requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 5, 2022 04:03
@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/7855/

@oneflow-ci-bot oneflow-ci-bot requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 5, 2022 05:21
@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

Speed stats:

@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/7855/

@oneflow-ci-bot oneflow-ci-bot requested review from oneflow-ci-bot and removed request for oneflow-ci-bot April 5, 2022 05:32
@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

Speed stats:

@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

Speed stats: 
GPU Name: GeForce GTX 1080 

✔️ OneFlow resnet50 time: 128.4ms (= 12838.1ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 139.9ms (= 13988.9ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.09 (= 139.9ms / 128.4ms)

✔️ OneFlow resnet50 time: 77.3ms (= 7725.6ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 81.5ms (= 8146.3ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.05 (= 81.5ms / 77.3ms)

OneFlow resnet50 time: 52.1ms (= 10416.1ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 59.6ms (= 11910.8ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.14 (= 59.6ms / 52.1ms)

OneFlow resnet50 time: 43.1ms (= 8612.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 43.1ms (= 8617.4ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.00 (= 43.1ms / 43.1ms)

OneFlow resnet50 time: 39.4ms (= 7878.7ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 37.5ms (= 7509.4ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 0.95 (= 37.5ms / 39.4ms)

OneFlow swin dataloader time: 0.248s (= 49.658s / 200, num_workers=1)
PyTorch swin dataloader time: 0.254s (= 50.704s / 200, num_workers=1)
✔️ Relative speed: 1.021 (= 0.254s / 0.248s)

OneFlow swin dataloader time: 0.064s (= 12.756s / 200, num_workers=4)
PyTorch swin dataloader time: 0.064s (= 12.815s / 200, num_workers=4)
✔️ Relative speed: 1.005 (= 0.064s / 0.064s)

OneFlow swin dataloader time: 0.036s (= 7.165s / 200, num_workers=8)
PyTorch swin dataloader time: 0.037s (= 7.447s / 200, num_workers=8)
✔️ Relative speed: 1.039 (= 0.037s / 0.036s)

✔️ OneFlow resnet50 time: 135.8ms (= 13576.3ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 155.9ms (= 15587.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.15 (= 155.9ms / 135.8ms)

OneFlow resnet50 time: 87.6ms (= 8757.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 99.4ms (= 9935.9ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.13 (= 99.4ms / 87.6ms)

OneFlow resnet50 time: 61.0ms (= 12200.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 75.0ms (= 14993.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.23 (= 75.0ms / 61.0ms)

OneFlow resnet50 time: 52.8ms (= 10560.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 76.6ms (= 15311.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.45 (= 76.6ms / 52.8ms)

OneFlow resnet50 time: 48.6ms (= 9728.1ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 61.7ms (= 12336.9ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.27 (= 61.7ms / 48.6ms)

@github-actions
Copy link
Contributor

github-actions bot commented Apr 5, 2022

View latest API docs preview at: https://staging.oneflow.info/docs/Oneflow-Inc/oneflow/pr/7855/

@mergify mergify bot merged commit e00ba51 into master Apr 5, 2022
@mergify mergify bot deleted the support_as_tensor_op branch April 5, 2022 06:09
@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

@Flowingsun007 Flowingsun007 Flowingsun007 approved these changes

@MARD1NO MARD1NO MARD1NO approved these changes

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

@jackalcooper jackalcooper Awaiting requested review from jackalcooper

@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 bug eager enhancement feature
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@BBuf @Flowingsun007 @MARD1NO @oneflow-ci-bot