NVIDIA / NVIDIA/cccl

[cuco][cudax] Enable map benchmarks with arbitrary key/value sizes

Open
#10,216 0 comments 0 reactions 0 assignees View on GitHub
cuco
Dominant language
C++
Stars
2.5k
Forks
486
Avg merge
2d 6h
Merged PRs (30d)
295

Description

cuco's map benchmarks currently have the restriction type combinations with `sizeof(Key) != sizeof(Value)` are skipped during the benchmark run. The anecdotal reason for this was to cut down on overall benchmark runtime.

The problem is that users may want to benchmark a certain type combination that is then automatically filtered out by through the applied constraint.
We may want to find a better mechanism for this.

Contributor guide

Open the contributing guide

Research direction

Start by locating cuco's map benchmark definitions and the constraint that skips key/value combinations with different sizes. Trace how the constraint is applied during the benchmark run, then define a mechanism that allows selected arbitrary combinations without forcing every combination to run; done means those combinations execute when requested while runtime remains manageable.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
performance
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
48/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.