9. Frequently asked questions

• When should we inform the SCC committee about code modifications to the benchmark source codes?

  If you modify the source codes for any of the benchmark applications, such modifications should be shared with the SCC25 committee by Nov 10, 2025 (Mon). This is to make sure that the committee has sufficient time to review and approve the changes. Code modifications after this date may be allowed by the SCC committee only in exceptional circumstances.

• We have novel hardware for which X benchmark has not been optimized. How can we still be competitive with other teams?

  Your raw benchmark output constitutes a fraction of your benchmarking score. You will also be evaluated on your overall efficiency, and your contributions to the community. The benchmarking judges will make every effort to perform fair comparisons across different hardware.

• Our team optimized the code for X. Do we get extra credit for that?

  If your team made significant code contribution to any benchmark or application, which is validated by the respective judges and shared publicly with the community before the start of the competition, then the judges can award bonus points to the team for that application or benchmark. Note that the awarding of bonus points will be at the discretion of the judge(s).

  To receive bonus points for the MLPerf Inference benchmark, teams must 1) publicly share their code changes with the community by creating a pull request to one or more of the following repositories: mlcommons/mlperf-automations, mlcommons/inference, mlcommons/inference_results_v5.1, at least 7 days before the start of the competition (Nov 10, 2025, 9:00AM CDT), 2) have their code changes reviewed by the MLCommons reviewer(s), 3) accept the Contributor License Agreement (CLA) to merge the PR into the repository.