Skip to content

Commit

Permalink
Update Readme + Cuda version (#87)
Browse files Browse the repository at this point in the history
  • Loading branch information
nathanielsimard authored Aug 29, 2024
1 parent aebf106 commit 238f71e
Show file tree
Hide file tree
Showing 3 changed files with 16 additions and 2 deletions.
8 changes: 7 additions & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -4,8 +4,14 @@
<br />
<br />

[![Rust Version](https://img.shields.io/badge/Rust-1.75.0+-blue)](https://releases.rs/docs/1.75.0)
[![Discord](https://img.shields.io/discord/1038839012602941528.svg?color=7289da&&logo=discord)](https://discord.gg/KSBSPhAUCc)
[![Current Crates.io Version](https://img.shields.io/crates/v/cubecl.svg)](https://crates.io/crates/cubecl)
[![Test Status](https://github.com/tracel-ai/cubecl/actions/workflows/ci.yml/badge.svg)](https://github.com/tracel-ai/cubecl/actions/workflows/test.yml)
[![Rust Version](https://img.shields.io/badge/Rust-1.79.0+-blue)](https://releases.rs/docs/1.79.0)
![license](https://shields.io/badge/license-MIT%2FApache--2.0-blue)
<br />
[![NVIDIA](https://img.shields.io/badge/nvidia-cuda-84b629)](https://github.com/tracel-ai/cubecl/tree/main/crates/cubecl-cuda)
[![GPU](https://img.shields.io/badge/any_gpu-wgpu-ef604b)](https://github.com/tracel-ai/cubecl/tree/main/crates/cubecl-wgpu)

---

Expand Down
2 changes: 1 addition & 1 deletion crates/cubecl-cuda/Cargo.toml
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,7 @@ cubecl-core = { path = "../cubecl-core", version = "0.2.0" }

half = { workspace = true }
bytemuck = { workspace = true }
cudarc = { version = "=0.11.5", features = ["cuda-12030"] }
cudarc = { version = "=0.11.5", features = ["cuda-version-from-build-system"] }

log = { workspace = true }
derive-new = { workspace = true }
Expand Down
8 changes: 8 additions & 0 deletions crates/cubecl-cuda/README.md
Original file line number Diff line number Diff line change
@@ -1,3 +1,11 @@
# Cuda runtime

The runtime uses the lower level primitives from [cudarc](https://github.com/coreylowman/cudarc) to compile generated CUDA code into a ptx and execute it at runtime.

## Setup

By default, this runtime will attempt to detect the appropriate CUDA version during build time. To specify a fixed CUDA version, you can add `cudarc` to your dependencies with the corresponding feature flag:

```toml
cudarc = { version = "same as burn", features = ["cuda-11040"] }
```

0 comments on commit 238f71e

Please sign in to comment.