Skip to content

[MISC] Release TileLens 3.1.2 - #490

Open
mark14wu wants to merge 1 commit into
mainfrom
release-3.1.2
Open

mark14wu wants to merge 1 commit into
mainfrom
release-3.1.2

Conversation

@mark14wu

@mark14wu mark14wu commented Oct 3, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Bump the version to 3.1.2 to release [COMPAT] Require Triton 3.8 #485.
  • tilelens 3.1.1 on PyPI still declares triton>=3.6.0, but import tilelens.core.simulation.gluon fails on Triton 3.6. Version 3.1.2 declares triton>=3.8.0 (from [COMPAT] Require Triton 3.8 #485), so pip will no longer install TileLens next to a Triton it cannot run with.

Test Plan

  • Built the wheel: tilelens-3.1.2-py3-none-any.whl, Version: 3.1.2, Requires-Dist: triton>=3.8.0.

After merge: tag the merge commit v3.1.2 and publish a GitHub release. The repository is now named tilelens, so PyPI's trusted publisher for tilelens must list repository tilelens (workflow publish.yml) before the release; the existing triton-viz entry no longer matches.

Related Issues

Follows #485.

Breaking Changes

Installing tilelens 3.1.2 requires Triton 3.8 or newer.

Checklist

  • I added tests to all new functionality I added/bugs I fixed.
  • I verified that a human has reviewed all code in this PR.
  • I ran npm run build:frontend if the PR modified any TypeScript code.
  • I made sure that my code is well documented (comments explaining strange code, docstrings for functions, website modified if new functionality added).

Release the Triton 3.8 floor from #485. tilelens 3.1.1 on PyPI still
declares triton>=3.6.0, but its Gluon modules fail to import on
Triton 3.6.
@github-actions

github-actions Bot commented Oct 3, 2026

Copy link
Copy Markdown

Performance Benchmark

Benchmark main (min) PR (min) Change Samples
gemm 0.078s 0.078s -0.0% 20 / 20
gemm_oob 0.088s 0.087s -0.6% 20 / 20
indirect_load 0.016s 0.016s -0.8% 20 / 20
nested_loop 0.163s 0.163s -0.4% 20 / 20
block_pointer_loop_advance 0.087s 0.087s -0.2% 20 / 20
liger_jsd 0.111s 0.110s -0.2% 20 / 20
flaggems_layernorm 0.283s 0.281s -0.6% 20 / 20
swiglu 0.131s 0.131s -0.5% 20 / 20
cross_entropy 0.737s 0.735s -0.3% 20 / 20
fused_linear_jsd 0.167s 0.167s +0.1% 20 / 20
Total 1.861s 1.855s -0.3% N/A

Iterations: 1 warmup + 20 measured
Samples are shown as main / PR; long pytest benchmarks may use fewer samples.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant