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

fix_structseq_repr_error #10527

Merged
merged 2 commits into from
May 21, 2024
Merged

fix_structseq_repr_error #10527

merged 2 commits into from
May 21, 2024

Conversation

clackhan
Copy link
Contributor

@clackhan clackhan commented May 20, 2024 •
edited

@clackhan clackhan requested a review from ShawnXuan May 20, 2024 05:32
@github-actions GitHub Actions
Copy link
Contributor

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10527/

@github-actions GitHub Actions
Copy link
Contributor

Speed stats: 
GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.9ms (= 4389.3ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.9ms (= 5793.9ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.32 (= 57.9ms / 43.9ms)

OneFlow resnet50 time: 26.2ms (= 2621.1ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.9ms (= 3787.0ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.44 (= 37.9ms / 26.2ms)

OneFlow resnet50 time: 18.8ms (= 3763.6ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 34.4ms (= 6877.0ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.83 (= 34.4ms / 18.8ms)

OneFlow resnet50 time: 16.4ms (= 3278.1ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 31.5ms (= 6293.3ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.92 (= 31.5ms / 16.4ms)

OneFlow resnet50 time: 17.1ms (= 3422.6ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.3ms (= 5858.9ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.71 (= 29.3ms / 17.1ms)

OneFlow swin dataloader time: 0.202s (= 40.431s / 200, num_workers=1)
PyTorch swin dataloader time: 0.129s (= 25.718s / 200, num_workers=1)
Relative speed: 0.636 (= 0.129s / 0.202s)

OneFlow swin dataloader time: 0.055s (= 10.969s / 200, num_workers=4)
PyTorch swin dataloader time: 0.032s (= 6.346s / 200, num_workers=4)
Relative speed: 0.579 (= 0.032s / 0.055s)

OneFlow swin dataloader time: 0.030s (= 6.076s / 200, num_workers=8)
PyTorch swin dataloader time: 0.016s (= 3.272s / 200, num_workers=8)
Relative speed: 0.538 (= 0.016s / 0.030s)

❌ OneFlow resnet50 time: 49.1ms (= 4906.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 64.1ms (= 6407.2ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.31 (= 64.1ms / 49.1ms)

OneFlow resnet50 time: 37.0ms (= 3704.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 46.6ms (= 4661.5ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.26 (= 46.6ms / 37.0ms)

OneFlow resnet50 time: 27.7ms (= 5535.8ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.4ms (= 7688.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.39 (= 38.4ms / 27.7ms)

OneFlow resnet50 time: 25.5ms (= 5101.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.5ms (= 7709.3ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.51 (= 38.5ms / 25.5ms)

OneFlow resnet50 time: 24.9ms (= 4973.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.2ms (= 7632.5ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.53 (= 38.2ms / 24.9ms)

@ShawnXuan ShawnXuan requested review from 0x404 and FxxxxU May 20, 2024 10:28
@github-actions GitHub Actions
Copy link
Contributor

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10527/

@github-actions GitHub Actions
Copy link
Contributor

Speed stats: 
GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 44.0ms (= 4397.6ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.8ms (= 5776.4ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.31 (= 57.8ms / 44.0ms)

OneFlow resnet50 time: 26.2ms (= 2617.3ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.4ms (= 3742.2ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.43 (= 37.4ms / 26.2ms)

OneFlow resnet50 time: 18.3ms (= 3651.1ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 34.2ms (= 6846.3ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.88 (= 34.2ms / 18.3ms)

OneFlow resnet50 time: 16.5ms (= 3306.6ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 33.4ms (= 6670.5ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 2.02 (= 33.4ms / 16.5ms)

OneFlow resnet50 time: 17.5ms (= 3506.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 28.2ms (= 5645.1ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.61 (= 28.2ms / 17.5ms)

OneFlow swin dataloader time: 0.199s (= 39.794s / 200, num_workers=1)
PyTorch swin dataloader time: 0.129s (= 25.749s / 200, num_workers=1)
Relative speed: 0.647 (= 0.129s / 0.199s)

OneFlow swin dataloader time: 0.054s (= 10.839s / 200, num_workers=4)
PyTorch swin dataloader time: 0.033s (= 6.536s / 200, num_workers=4)
Relative speed: 0.603 (= 0.033s / 0.054s)

OneFlow swin dataloader time: 0.031s (= 6.200s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.350s / 200, num_workers=8)
Relative speed: 0.540 (= 0.017s / 0.031s)

❌ OneFlow resnet50 time: 49.3ms (= 4934.6ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.1ms (= 6609.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 66.1ms / 49.3ms)

OneFlow resnet50 time: 36.5ms (= 3652.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 49.0ms (= 4903.1ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 49.0ms / 36.5ms)

OneFlow resnet50 time: 27.8ms (= 5566.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.0ms (= 7808.9ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.40 (= 39.0ms / 27.8ms)

OneFlow resnet50 time: 25.4ms (= 5070.0ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.5ms (= 7690.6ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.52 (= 38.5ms / 25.4ms)

OneFlow resnet50 time: 24.5ms (= 4899.7ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 35.9ms (= 7178.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.47 (= 35.9ms / 24.5ms)

@github-actions GitHub Actions
Copy link
Contributor

Speed stats: 
GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.6ms (= 4356.7ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.6ms (= 5757.5ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.32 (= 57.6ms / 43.6ms)

OneFlow resnet50 time: 26.4ms (= 2642.3ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.8ms (= 3781.6ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.43 (= 37.8ms / 26.4ms)

OneFlow resnet50 time: 17.8ms (= 3551.9ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 35.9ms (= 7179.6ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 2.02 (= 35.9ms / 17.8ms)

OneFlow resnet50 time: 17.1ms (= 3428.8ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 32.5ms (= 6504.8ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.90 (= 32.5ms / 17.1ms)

OneFlow resnet50 time: 17.1ms (= 3425.6ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 28.0ms (= 5608.9ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.64 (= 28.0ms / 17.1ms)

OneFlow swin dataloader time: 0.200s (= 39.985s / 200, num_workers=1)
PyTorch swin dataloader time: 0.127s (= 25.379s / 200, num_workers=1)
Relative speed: 0.635 (= 0.127s / 0.200s)

OneFlow swin dataloader time: 0.054s (= 10.714s / 200, num_workers=4)
PyTorch swin dataloader time: 0.033s (= 6.612s / 200, num_workers=4)
Relative speed: 0.617 (= 0.033s / 0.054s)

OneFlow swin dataloader time: 0.030s (= 5.967s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.339s / 200, num_workers=8)
Relative speed: 0.560 (= 0.017s / 0.030s)

❌ OneFlow resnet50 time: 49.3ms (= 4928.0ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 65.4ms (= 6535.9ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.33 (= 65.4ms / 49.3ms)

OneFlow resnet50 time: 37.3ms (= 3727.3ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 45.8ms (= 4576.2ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.23 (= 45.8ms / 37.3ms)

OneFlow resnet50 time: 27.8ms (= 5565.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 39.2ms (= 7841.6ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.41 (= 39.2ms / 27.8ms)

OneFlow resnet50 time: 25.4ms (= 5076.2ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.3ms (= 7653.8ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.51 (= 38.3ms / 25.4ms)

OneFlow resnet50 time: 24.9ms (= 4974.4ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 35.8ms (= 7169.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.44 (= 35.8ms / 24.9ms)

@ShawnXuan ShawnXuan merged commit 2afc090 into master May 21, 2024
20 checks passed
@ShawnXuan ShawnXuan deleted the fix_structseq_repr_error branch May 21, 2024 04:14
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@ShawnXuan ShawnXuan ShawnXuan approved these changes

@0x404 0x404 Awaiting requested review from 0x404

@FxxxxU FxxxxU Awaiting requested review from FxxxxU

Assignees
No one assigned
Labels
bug eager system
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
3 participants
@clackhan @ShawnXuan @FxxxxU