CPUFJ for local search + simple rounding#405
Merged
rapids-bot[bot] merged 290 commits intobranch-25.10from Sep 29, 2025
Merged
Conversation
This reverts commit 4c1c75e.
Contributor
Author
|
/ok to test 6e4e9b4 |
Contributor
Author
|
/ok to test 323de70 |
rg20
reviewed
Sep 25, 2025
Co-authored-by: Rajesh Gandham <rgandham@nvidia.com>
Co-authored-by: Rajesh Gandham <rgandham@nvidia.com>
Co-authored-by: Rajesh Gandham <rgandham@nvidia.com>
Co-authored-by: Rajesh Gandham <rgandham@nvidia.com>
Fixes two small packaging issues with `cuopt-server`: * removes unnecessary dependency on `httpx` (`cuopt-server` does not use this) * declares dependency on `psutil` (this is declared in wheels but was missing in conda packages) ## Notes for Reviewers ### How I found these Was looking at the conda recipes and checked these things, similar to this: ```shell git grep httpx ``` # Authors: - James Lamb (https://github.com/jameslamb) Approvers: - Ramakrishnap (https://github.com/rgsl888prabhu) URL: #413
This dataset was disabled in PR #335 since the website was down, the services have been restored as of now, so enabling this back. And also adding details to README to benchmark mittleman and miplib. ## Issue #335 Authors: - Ramakrishnap (https://github.com/rgsl888prabhu) Approvers: - Rajesh Gandham (https://github.com/rg20) URL: #414
Adds 2 more video links to docs Authors: - Ramakrishnap (https://github.com/rgsl888prabhu) Approvers: - Trevor McKay (https://github.com/tmckayus) URL: #416
nguidotti
reviewed
Sep 26, 2025
Contributor
Author
|
/ok to test acaa643 |
Contributor
Author
|
/ok to test bb86c49 |
rg20
approved these changes
Sep 29, 2025
Contributor
|
/merge |
jieyibi
pushed a commit
to yining043/cuopt
that referenced
this pull request
Mar 26, 2026
This PR introduces a host-side implementation of Feasibility Jump 2.0 (Efficient Local Search), with CPU-specific tweaks and optimizations. This heuristic is run in parallel with GPU FJ during local search, and the best resulting solution is picked. Also included in this PR: - Implementation of the simple rounding algorithm (up/down-lock based rounding, cf. "Rounding and Propagation Heuristics for Mixed Integer Programming") - Some crash/assert trigger fixes - Minor hyperparameter tuning - `build.sh` option to emit lineinfo for host source files **Benchmark results:** Main branch [53d6e74](NVIDIA@53d6e74): ``` Regression Baseline Summary (Time Limit: 600s): Feasible instances: 221/240 Average gap: 0.207478 Average integral: 0.261499 ``` [Excel results](https://docs.google.com/spreadsheets/d/1caQCORK0n-1ynOpxkIrt00rq6mKiAhg2/edit?usp=drive_link&ouid=101768445160958513821&rtpof=true&sd=true) PR (cad9773): ``` cuopt 600s: feasible 222/240 gap 0.176733 integral 0.241953 ``` [Excel results](https://docs.google.com/spreadsheets/d/1qt1dIjp-FNBaNQwFiQ2Fdzx1hOAyc8et/edit?usp=drive_link&ouid=101768445160958513821&rtpof=true&sd=true) closes NVIDIA#360 Authors: - Alice Boucher (https://github.com/aliceb-nv) - Akif ÇÖRDÜK (https://github.com/akifcorduk) - Kumar Aatish (https://github.com/kaatish) - Rajesh Gandham (https://github.com/rg20) - Hugo Linsenmaier (https://github.com/hlinsen) - Ramakrishnap (https://github.com/rgsl888prabhu) - Ishika Roy (https://github.com/Iroy30) - James Lamb (https://github.com/jameslamb) Approvers: - Trevor McKay (https://github.com/tmckayus) - Rajesh Gandham (https://github.com/rg20) URL: NVIDIA#405
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR introduces a host-side implementation of Feasibility Jump 2.0 (Efficient Local Search), with CPU-specific tweaks and optimizations. This heuristic is run in parallel with GPU FJ during local search, and the best resulting solution is picked.
Also included in this PR:
build.shoption to emit lineinfo for host source filesBenchmark results:
Main branch 53d6e74:
Excel results
PR (cad9773):
Excel results
closes #360