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 comfyui sd3 speedup #946

Merged
merged 3 commits into from
Jun 13, 2024
Merged

Support comfyui sd3 speedup #946

merged 3 commits into from
Jun 13, 2024

Conversation

ccssu
Copy link
Contributor

@ccssu ccssu commented Jun 12, 2024 •
edited
Loading

安装

Run WorkFlow

  • 28 step image 1024x1024 mode: max-optimize:max-autotune:low-precision
Accelerator Baseline (non-optimized) OneDiff (optimized) Percentage improvement
NVIDIA A800-SXM4-80GB ~4.03 sec ~2.93 sec ~27.29 % 
动态 shape
 print("Test run with multiple resolutions...")
        sizes = [1024, 512, 768, 256]
        for h in sizes:
            for w in sizes:

workflow (17)

@strint
Copy link
Collaborator

strint commented Jun 13, 2024

需要补充一个 readme

@lixiang007666
Copy link
Contributor

@ccssu 这些是工作流里可能用到的模型。

@strint strint changed the title Support sd3 Speedup Support comfyui sd3 speedup Jun 13, 2024
@ccssu ccssu requested a review from strint June 13, 2024 07:26
@strint strint merged commit 5677af5 into main Jun 13, 2024
@strint strint deleted the dev_support_sd3_comfy branch June 13, 2024 07:46
strint
strint reviewed Jun 13, 2024
View reviewed changes
onediff_comfy_nodes/extras_nodes/nodes_nexfort_booster.py
@@ -7,7 +7,8 @@
{
"jit:disable-runtime-fusion:low-precision": "This compiles super quickly, but the performance might not be optimized very noticeably.",
"jit:benchmark:low-precision:freezing:cudagraphs": "This compiles the model very quickly, but the performance might be not as good as `TorchInductor` optimized models.",
"max-autotune:low-precision": "This will deliver a good performance and adapt quickly to shape changes.",
"max-optimize:max-autotune:low-precision": "This will deliver a good performance and adapt quickly to shape changes.",
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

可以在 readme 里面介绍推荐使用这个模式

strint
strint reviewed Jun 13, 2024
View reviewed changes
onediff_comfy_nodes/sd3_demo/README.md
@@ -0,0 +1,97 @@
## Accelerate SD3 by using onediff
huggingface: https://huggingface.co/stabilityai/stable-diffusion-3-medium
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

可以加个 setup 的说明：https://github.com/siliconflow/onediff/tree/main/onediff_diffusers_extensions/examples/sd3#set-up-sd3

说清楚下 nexfort 安装、comfyui、模型 id

strint
strint reviewed Jun 13, 2024
View reviewed changes
onediff_comfy_nodes/sd3_demo/README.md
### Feature
- ✅ Multiple resolutions

### Performance
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这个可以参考： https://github.com/siliconflow/onediff/tree/main/onediff_diffusers_extensions/examples/sd3#performance-comparation

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@strint strint strint approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@ccssu @strint @lixiang007666