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

The rasterizer example runs < 15 FPS #6950

Closed
strongoier opened this issue Dec 21, 2022 · 2 comments · Fixed by #6951
Closed

The rasterizer example runs < 15 FPS #6950

strongoier opened this issue Dec 21, 2022 · 2 comments · Fixed by #6951
Assignees
@strongoier
Labels
bug We've confirmed that this is an BUG

Comments

@strongoier
Copy link
Contributor

In the past the rasterizer example can run 60 FPS. Now it can only run < 15 FPS.

After bisecting I find that the problem have existed since commit acc2cb0. The change to fill() in #6892 accidentally causes this issue. I'll send a PR to fix it.
@strongoier strongoier self-assigned this Dec 21, 2022
@taichi-gardener taichi-gardener moved this to Untriaged in Taichi Lang Dec 21, 2022
@strongoier strongoier added the bug We've confirmed that this is an BUG label Dec 21, 2022
@strongoier strongoier mentioned this issue Dec 21, 2022
Merged
@turbo0628
Copy link
Member

turbo0628 commented Dec 21, 2022 •
edited
Loading

I have also observed sharp performance regression in filling large ndarrays between v1.0.2->v1.0.3, seems we have some other bottlenecks. I'll post a chart when the benchmark finishes.

Update:
Perf monitor chart
image
Tested size are 16/32/128MB, vulkan and CUDA backends.
The benchmark failed to run after v1.1.2.

@strongoier
Copy link
Contributor Author

Interesting. That sounds like another problem. What about tracking this in another issue?

strongoier added a commit that referenced this issue Dec 22, 2022
@strongoier
[bug] Fix recompilation of filling a matrix field with the same matrix (
be15358 
#6951)

Issue: fix #6950

### Brief Summary

The change to `fill()` in #6892 (skips turning a `ti.Matrix` into a
tuple) accidentally disallows compilation caching when the argument is a
`ti.Matrix`, which causes #6950. This PR puts that behavior back.
Repository owner moved this from Untriaged to Done in Taichi Lang Dec 22, 2022
quadpixels pushed a commit to quadpixels/taichi that referenced this issue May 13, 2023
@strongoier @quadpixels
[bug] Fix recompilation of filling a matrix field with the same matrix (
e18affc 
taichi-dev#6951)

Issue: fix taichi-dev#6950

### Brief Summary

The change to `fill()` in taichi-dev#6892 (skips turning a `ti.Matrix` into a
tuple) accidentally disallows compilation caching when the argument is a
`ti.Matrix`, which causes taichi-dev#6950. This PR puts that behavior back.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@strongoier strongoier

Labels
bug We've confirmed that this is an BUG
Projects
Taichi Lang
Status: Done
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

[bug] Fix recompilation of filling a matrix field with the same matrix strongoier/taichi
2 participants
@strongoier @turbo0628