Cédric Augonnet

Port MiniWeather over the CUDASTF programming model

3

This PR introduces a new version of the MiniWeather benchmark based on the CUDASTF programming model. CUDASTF is shipped in NVIDIA's CCCL project, and implements task-parallelism as a C++ header...

[STF] Stackable optimizations

6

## Description Experiments on top of the stackable PR (not to break code using the branch), do not merge. closes ## Checklist - [ ] New or existing tests cover...

[STF] stackable stf resources

69

## Description This introduces helper methods to improve how we nest contexts to better leverage CUDA Graphs ## Checklist - [x] New or existing tests cover these changes. - [...

[STF] Add a test where we use parallel_for without logical data

1

## Description closes ## Checklist - [ ] New or existing tests cover these changes. - [ ] The documentation is up to date with these changes.

[STF] Add concepts for STF classes

3

## Description If we want to compose context implementations, we may add some concepts to avoid programming mistakes (and maybe specify the interface) closes ## Checklist - [ ] New...

[STF] Add a unit test with a logical data of an mdspan with static size extents

9

## Description We want to make sure we do not loose the nice properties of mdspan when manipulating thing by the means of logical data facilitie closes ## Checklist -...

### Is this a duplicate? - [x] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cccl/issues) for this bug and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md) ### Type...

[WIP] Generate CUDA graphs with CUDASTF

7

### PR content/description This PR introduces CUDA graphs by the means of the CUDASTF library TODO * [x] fetch sources from cmake * [ ] add an option in cmake/cargo...

[STF] Use a reduce access mode to compute residuals

2

Compute residuals using a reduce access mode rather than a sequential code on the host ### Which kernels are implemented? - [ ] synch_p2p (p2p) - [ ] stencil -...

[DO NOT MERGE] Experiment to leverage CUDASTF

4

This is intended to explore how we can use CUDASTF within cuGraph