microsoft / microsoft/onnxruntime
Minimal build documentation does not include all important limitations
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 21.9k
- Forks
- 4.2k
- Avg merge
- 4d 8h
- Merged PRs (30d)
- 179
Description
Is your feature request related to a problem? Please describe.
Official documentation on Minimal build is found exclusively here. The documentation lists some limitations of minimal build but doesn't mention that it disables avx512 instructions (see implementation here) - which is important for users to be aware.
For instance, on a x86_64 platform that has avx512, disabling the use avx512 instructions brings down inference performance up to 47% on sample models. I'd like to know what other features minimal build disables that is not listed in its documentation, but could impact inference latency.
System information
- ONNX Runtime version (you are using): 1.11.0
Describe the solution you'd like
Update the documentation with all limitations whether it's low- or high- level. This way, users can make informed judgement on whether to use minimal build or not without searching all occurrence of defined(ORT_MINIMAL_BUILD) in code base.
Describe alternatives you've considered
Manually search all occurrences of defined(ORT_MINIMAL_BUILD) in source code to understand what will be disabled for minimal build.
Additional context
n/a
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the Minimal build documentation at custom.html#minimal-build and review the linked platform.cpp implementation. Search the source for defined(ORT_MINIMAL_BUILD) to identify limitations that affect users, including AVX512 behavior. Done means documenting the relevant low- and high-level limitations so users can judge the build’s inference impact without manually searching the code.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- build-system, documentation, performance
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100