microsoft / microsoft/onnxruntime

[Feature Request] Fast HalfToFloat functions for c#

Open
#15,391 2 comments 0 reactions 0 assignees View on GitHub
feature request
Dominant language
C++
Stars
21.9k
Forks
4.2k
Avg merge
4d 11h
Merged PRs (30d)
184

Description

### Describe the feature request

I haven't found any good libraries that do fast HalfToFloat conversions (ie. Float16 to float)

Currently I am just using [the Unity function](https://docs.unity3d.com/ScriptReference/Mathf.HalfToFloat.html):

It would be good to get some fast functions HalfToFloat and FloatToHalf that work on tensors or arrays.

### Describe scenario use case

to convert float16 to float and vice versa.

Contributor guide

Open the contributing guide

Research direction

Start by comparing the requested behavior with Unity's linked Mathf.HalfToFloat reference and locating ONNX Runtime's relevant tensor or array conversion entry points. The issue is complete when fast HalfToFloat and FloatToHalf functions support the requested float16 conversions for tensors or arrays, with performance and correctness checked.

Written by the indexing model from the issue text.

Assessment

Tech stack
csharp
Domain
machine-learning, performance
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
32/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.