Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Synthesize indirect dispatch workload calculations for data-dependent shapes #38

Closed
benvanik opened this issue Oct 13, 2019 · 1 comment
Closed

Synthesize indirect dispatch workload calculations for data-dependent shapes #38

benvanik opened this issue Oct 13, 2019 · 1 comment
Labels
compiler/dialects Relating to the IREE compiler dialects (flow, hal, vm) enhancement ➕ New feature or request next-gen ✨ Forward-looking work built on IREE's foundation

Comments

@benvanik
Copy link
Collaborator

When we aren't able to fully parameterize shapes based on input shapes (such as when a shape is sliced out of an arbitrary tensor) we'll need to synthesize a dispatch that computes the workload for indirect dispatch. We'll want to add bounds checking and compatibility checks and some way of communicating data loss errors (such as when the computed workload is not possible to perform), though we could possibly work around this by enqueuing multiple indirect dispatches and using as many as we need to satisfy the request.
@benvanik benvanik added the enhancement ➕ New feature or request label Oct 13, 2019
@benvanik benvanik added this to the Dynamic Shapes milestone Oct 13, 2019
@benvanik benvanik added the compiler/dialects Relating to the IREE compiler dialects (flow, hal, vm) label Mar 19, 2020
@benvanik benvanik modified the milestones: Dynamic Shapes, flow.stream modeling Nov 22, 2020
@benvanik benvanik added the next-gen ✨ Forward-looking work built on IREE's foundation label Nov 22, 2020
@benvanik benvanik removed this from the flow.stream modeling milestone Nov 15, 2021
@allieculp allieculp removed this from IREE Codegen May 3, 2022
@GMNGeoffrey GMNGeoffrey added this to IREE Jun 28, 2022
@benvanik
Copy link
Collaborator Author

This is just an idea; will file targeted issues when we get around to it.

@benvanik benvanik closed this as not planned Won't fix, can't repro, duplicate, stale Jun 30, 2022
@benvanik benvanik moved this to Done in IREE Jun 30, 2022
@GMNGeoffrey GMNGeoffrey mentioned this issue Mar 29, 2023
Merged
@dpackwood dpackwood mentioned this issue Sep 8, 2023
Closed
stellaraccident pushed a commit that referenced this issue Sep 24, 2023
@rsuderman
Support degenerate HostToBuffer cases (#38)
2d691ec 
It is possible for the transfer to be degenerate, specifically
transferring either a splat matrix or a matrix with a length-0
dimension. This also includes a util for computing the permutation for
when permuted examples appear.
@gabeweisz gabeweisz mentioned this issue Mar 25, 2024
Open
@deng-ShiFu deng-ShiFu mentioned this issue Jul 3, 2024
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
compiler/dialects Relating to the IREE compiler dialects (flow, hal, vm) enhancement ➕ New feature or request next-gen ✨ Forward-looking work built on IREE's foundation
Projects
No open projects
IREE
Archived in project
Milestone
No milestone
Development

No branches or pull requests
1 participant
@benvanik