Skip to content

Conversation

Anteru
Copy link

@Anteru Anteru commented Jul 1, 2025

Allow i8 and u8 types, which are useful for pre-processing/post-processing data used in cooperative matrix/vector operations. Right now it's really ugly to unpack/pack via i32. This proposal enables i8 and u8 as native types in the shader core.

Copy link
Member

@damyanp damyanp left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Couple of notes remaining from me.

@Anteru
Copy link
Author

Anteru commented Aug 27, 2025

Can someone please re-run the CLA thing? It should be resolved on my end.

@damyanp
Copy link
Member

damyanp commented Aug 27, 2025

Can someone please re-run the CLA thing? It should be resolved on my end.

Please follow the instructions in the comment it posted.

@Anteru
Copy link
Author

Anteru commented Oct 1, 2025

CLA bot should be pleased now? I should have been added.

Copy link
Member

@damyanp damyanp left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks, CLA bot seems happy now.

Will need the header updated and the text formatted to 80 columns before we can accept this.

The comment around tying it to cooperative vectors can take or leave for now.


## Introduction

This proposal introduces 8-bit integer types to HLSL, to align it more closely with C/C++ and other languages, and allow tight interop with the 8-bit types used in the cooperative vector proposal.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please format to 80 columns.


## Introduction

This proposal introduces 8-bit integer types to HLSL, to align it more closely with C/C++ and other languages, and allow tight interop with the 8-bit types used in the cooperative vector proposal.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As the cooperative vector proposal has been rejected it might be better reframe this proposal. Does it still make sense when combined with the linear algebra matrix proposal? Does this proposal bring enough benefits on its own without tying it to another one?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's somewhat orthogonal (8-bit types have value on their own), but it's obviously help LinAlg too instead of using packed types there. I'll rephrase.

Co-authored-by: Damyan Pepper <damyanp@microsoft.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

Status: No status

Development

Successfully merging this pull request may close these issues.

3 participants