[pull] main from Rust-GPU:main

.github/workflows/ci_linux.yml

@@ -163,7 +163,6 @@ jobs:
               --exclude cust
           '
 
-      # Exclude cust_raw because it triggers hundreds of warnings.
       - name: Check documentation
         run: |
           docker exec "$CONTAINER_NAME" bash -lc 'set -euo pipefail

.github/workflows/ci_windows.yml

@@ -106,7 +106,6 @@ jobs:
         run: cargo test --workspace --exclude blastoff --exclude cudnn --exclude cudnn-sys --exclude cust
 
       # Exclude crates that require cuDNN, not available on Windows CI: cudnn, cudnn-sys.
-      # Exclude cust_raw because it triggers hundreds of warnings.
       - name: Check documentation
         env:
           RUSTDOCFLAGS: -Dwarnings

.github/workflows/container_images.yml

@@ -33,9 +33,15 @@ jobs:
           - name: Ubuntu-24.04/CUDA-12.8.1
             image: "rust-gpu/rust-cuda-ubuntu24-cuda12"
             dockerfile: ./container/ubuntu24-cuda12/Dockerfile
+          - name: Ubuntu-24.04/CUDA-13.0.2
+            image: "rust-gpu/rust-cuda-ubuntu24-cuda13"
+            dockerfile: ./container/ubuntu24-cuda13/Dockerfile
           - name: RockyLinux-9/CUDA-12.8.1
             image: "rust-gpu/rust-cuda-rockylinux9-cuda12"
             dockerfile: ./container/rockylinux9-cuda12/Dockerfile
+          - name: RockyLinux-9/CUDA-13.0.2
+            image: "rust-gpu/rust-cuda-rockylinux9-cuda13"
+            dockerfile: ./container/rockylinux9-cuda13/Dockerfile
     steps:
       - name: Free up space
         # Without this the job will likely run out of disk space.
@@ -153,8 +159,12 @@ jobs:
         variance:
           - name: Ubuntu-24.04/CUDA-12.8.1
             image: "rust-gpu/rust-cuda-ubuntu24-cuda12"
+          - name: Ubuntu-24.04/CUDA-13.0.2
+            image: "rust-gpu/rust-cuda-ubuntu24-cuda13"
           - name: RockyLinux-9/CUDA-12.8.1
             image: "rust-gpu/rust-cuda-rockylinux9-cuda12"
+          - name: RockyLinux-9/CUDA-13.0.2
+            image: "rust-gpu/rust-cuda-rockylinux9-cuda13"
     steps:
       - name: Set artifact name
         run: |

README.md

@@ -2,116 +2,36 @@
   <h1>The Rust CUDA Project</h1>
 
   <p>
-    <strong>An ecosystem of libraries and tools for writing and executing extremely fast GPU code fully in
-    <a href="https://www.rust-lang.org/">Rust</a></strong>
+    <strong>An ecosystem of libraries and tools for writing and executing extremely fast GPU code
+    fully in <a href="https://www.rust-lang.org/">Rust</a></strong>
   </p>
-
-  <h3>
-    <a href="https://rust-gpu.github.io/rust-cuda/index.html">Guide</a>
-    <span> | </span>
-    <a href="https://rust-gpu.github.io/rust-cuda/guide/getting_started.html">Getting Started</a>
-    <span> | </span>
-    <a href="https://rust-gpu.github.io/rust-cuda/features.html">Features</a>
-  </h3>
-<strong>⚠️ The project is still in early development, expect bugs, safety issues, and things that don't work ⚠️</strong>
 </div>
 
-<br/>
-
 > [!IMPORTANT]
 > This project is no longer dormant and is [being
 > rebooted](https://rust-gpu.github.io/blog/2025/01/27/rust-cuda-reboot). Read the [latest status update](https://rust-gpu.github.io/blog/2025/08/11/rust-cuda-update).
 > Please contribute!
+>
+> The project is still in early development, however. Expect bugs, safety issues, and things that
+> don't work.
 
-## Goal
-
-The Rust CUDA Project is a project aimed at making Rust a tier-1 language for extremely fast GPU computing
-using the CUDA Toolkit. It provides tools for compiling Rust to extremely fast PTX code as well as libraries
-for using existing CUDA libraries with it.
-
-## Background
-
-Historically, general purpose high performance GPU computing has been done using the CUDA toolkit. The CUDA toolkit primarily
-provides a way to use Fortran/C/C++ code for GPU computing in tandem with CPU code with a single source. It also provides
-many libraries, tools, forums, and documentation to supplement the single-source CPU/GPU code.
-
-CUDA is exclusively an NVIDIA-only toolkit. Many tools have been proposed for cross-platform GPU computing such as
-OpenCL, Vulkan Computing, and HIP. However, CUDA remains the most used toolkit for such tasks by far. This is why it is
-imperative to make Rust a viable option for use with the CUDA toolkit.
-
-However, CUDA with Rust has been a historically very rocky road. The only viable option until now has been to use the LLVM PTX
-backend, however, the LLVM PTX backend does not always work and would generate invalid PTX for many common Rust operations, and
-in recent years it has been shown time and time again that a specialized solution is needed for Rust on the GPU with the advent
-of projects such as rust-gpu (for Rust -> SPIR-V).
-
-Our hope is that with this project we can push the Rust GPU computing industry forward and make Rust an excellent language
-for such tasks. Rust offers plenty of benefits such as `__restrict__` performance benefits for every kernel, An excellent module/crate system,
-delimiting of unsafe areas of CPU/GPU code with `unsafe`, high level wrappers to low level CUDA libraries, etc.
-
-## Structure
-
-The scope of the Rust CUDA Project is quite broad, it spans the entirety of the CUDA ecosystem, with libraries and tools to make it
-usable using Rust. Therefore, the project contains many crates for all corners of the CUDA ecosystem.
-
-The current line-up of libraries is the following:
-
-- `rustc_codegen_nvvm` Which is a rustc backend that targets NVVM IR (a subset of LLVM IR) for the [libnvvm](https://docs.nvidia.com/cuda/nvvm-ir-spec/index.html) library.
-  - Generates highly optimized PTX code which can be loaded by the CUDA Driver API to execute on the GPU.
-  - For the near future it will be CUDA-only, but it may be used to target amdgpu in the future.
-- `cuda_std` for GPU-side functions and utilities, such as thread index queries, memory allocation, warp intrinsics, etc.
-  - _Not_ a low level library, provides many utility functions to make it easier to write cleaner and more reliable GPU kernels.
-  - Closely tied to `rustc_codegen_nvvm` which exposes GPU features through it internally.
-- [`cudnn`](https://github.com/Rust-GPU/rust-cuda/tree/master/crates/cudnn) for a collection of GPU-accelerated primitives for deep neural networks.
-- `cust` for CPU-side CUDA features such as launching GPU kernels, GPU memory allocation, device queries, etc.
-  - High level with features such as RAII and Rust Results that make it easier and cleaner to manage the interface to the GPU.
-  - A high level wrapper for the CUDA Driver API, the lower level version of the more common CUDA Runtime API used from C++.
-  - Provides much more fine grained control over things like kernel concurrency and module loading than the C++ Runtime API.
-- `gpu_rand` for GPU-friendly random number generation, currently only implements xoroshiro RNGs from `rand_xoshiro`.
-- `optix` for CPU-side hardware raytracing and denoising using the CUDA OptiX library.
-
-In addition to many "glue" crates for things such as high level wrappers for certain smaller CUDA libraries.
-
-## Related Projects
-
-Other projects related to using Rust on the GPU:
-
-- 2016: [glassful](https://github.com/kmcallister/glassful) Subset of Rust that compiles to GLSL.
-- 2017: [inspirv-rust](https://github.com/msiglreith/inspirv-rust) Experimental Rust MIR -> SPIR-V Compiler.
-- 2018: [nvptx](https://github.com/japaric-archived/nvptx) Rust to PTX compiler using the `nvptx` target for rustc (using the LLVM PTX backend).
-- 2020: [accel](https://github.com/termoshtt/accel) Higher-level library that relied on the same mechanism that `nvptx` does.
-- 2020: [rlsl](https://github.com/MaikKlein/rlsl) Experimental Rust -> SPIR-V compiler (predecessor to rust-gpu)
-- 2020: [rust-gpu](https://github.com/Rust-GPU/rust-gpu) `rustc` compiler backend to compile Rust to SPIR-V for use in shaders, similar mechanism as our project.
-
-## Usage
-```bash
-## setup your environment like:
-### export OPTIX_ROOT=/opt/NVIDIA-OptiX-SDK-9.0.0-linux64-x86_64
-### export OPTIX_ROOT_DIR=/opt/NVIDIA-OptiX-SDK-9.0.0-linux64-x86_64
-
-## build proj
-cargo build
-```
-
-## Use Rust CUDA in Container Environments
-
-The distribution related Dockerfile are located in `container` folder.
-Taking ubuntu 24.04 as an example, run the following command in repository root:
-```bash
-docker build -f ./container/ubuntu24-cuda12/Dockerfile -t rust-cuda-ubuntu24 .
-docker run --rm --runtime=nvidia --gpus all -it rust-cuda-ubuntu24
-```
+## Documentation
 
-A sample `.devcontainer.json` file is also included, configured for Ubuntu 24.02. Copy this to `.devcontainer/devcontainer.json` to make additional customizations.
+Please see [The Rust CUDA Guide](https://rust-gpu.github.io/rust-cuda/) for documentation on Rust
+CUDA.
 
 ## License
 
 Licensed under either of
 
-- Apache License, Version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or http://www.apache.org/licenses/LICENSE-2.0)
+- Apache License, Version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or
+  http://www.apache.org/licenses/LICENSE-2.0)
 - MIT license ([LICENSE-MIT](LICENSE-MIT) or http://opensource.org/licenses/MIT)
 
 at your discretion.
 
 ### Contribution
 
-Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
+Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in
+the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any
+additional terms or conditions.

container/rockylinux9-cuda13/Dockerfile

@@ -0,0 +1,92 @@
+FROM nvcr.io/nvidia/cuda:13.0.2-cudnn-devel-rockylinux9 AS llvm-builder
+
+RUN dnf -y install \
+    --nobest \
+    --allowerasing \
+    --setopt=install_weak_deps=False \
+    openssl-devel \
+    pkgconfig \
+    which \
+    xz \
+    zlib-devel \
+    libffi-devel \
+    ncurses-devel \
+    libxml2-devel \
+    libedit-devel \
+    python3 \
+    make \
+    cmake && \
+    dnf clean all
+
+WORKDIR /data/llvm7
+
+# Download and build LLVM 7.1.0 for all architectures.
+RUN curl -sSf -L -O https://github.com/llvm/llvm-project/releases/download/llvmorg-7.1.0/llvm-7.1.0.src.tar.xz && \
+    tar -xf llvm-7.1.0.src.tar.xz && \
+    cd llvm-7.1.0.src && \
+    mkdir build && cd build && \
+    ARCH=$(uname -m) && \
+    if [ "$ARCH" = "x86_64" ]; then \
+        TARGETS="X86;NVPTX"; \
+    else \
+        TARGETS="AArch64;NVPTX"; \
+    fi && \
+    cmake \
+        -DCMAKE_BUILD_TYPE=Release \
+        -DLLVM_TARGETS_TO_BUILD="$TARGETS" \
+        -DLLVM_BUILD_LLVM_DYLIB=ON \
+        -DLLVM_LINK_LLVM_DYLIB=ON \
+        -DLLVM_ENABLE_ASSERTIONS=OFF \
+        -DLLVM_ENABLE_BINDINGS=OFF \
+        -DLLVM_INCLUDE_EXAMPLES=OFF \
+        -DLLVM_INCLUDE_TESTS=OFF \
+        -DLLVM_INCLUDE_BENCHMARKS=OFF \
+        -DLLVM_ENABLE_ZLIB=ON \
+        -DLLVM_ENABLE_TERMINFO=ON \
+        -DCMAKE_INSTALL_PREFIX=/opt/llvm-7 \
+        .. && \
+    make -j$(nproc) && \
+    make install && \
+    cd ../.. && \
+    rm -rf llvm-7.1.0.src* && \
+    dnf clean all
+
+FROM nvcr.io/nvidia/cuda:13.0.2-cudnn-devel-rockylinux9
+
+RUN dnf -y install \
+    --nobest \
+    --allowerasing \
+    --setopt=install_weak_deps=False \
+    clang \
+    openssl-devel \
+    fontconfig-devel \
+    libX11-devel \
+    libXcursor-devel \
+    libXi-devel \
+    libXrandr-devel \
+    libxml2-devel \
+    ncurses-devel \
+    pkgconfig \
+    which \
+    xz \
+    zlib-devel \
+    cmake && \
+    dnf clean all
+
+COPY --from=llvm-builder /opt/llvm-7 /opt/llvm-7
+RUN ln -s /opt/llvm-7/bin/llvm-config /usr/bin/llvm-config && \
+    ln -s /opt/llvm-7/bin/llvm-config /usr/bin/llvm-config-7
+
+# Get Rust (install rustup; toolchain installed from rust-toolchain.toml below)
+RUN curl -sSf -L https://sh.rustup.rs | bash -s -- -y --profile minimal --default-toolchain none
+ENV PATH="/root/.cargo/bin:${PATH}"
+
+# Setup the workspace
+WORKDIR /data/rust-cuda
+RUN --mount=type=bind,source=rust-toolchain.toml,target=/data/rust-cuda/rust-toolchain.toml \
+    rustup show
+
+# Add nvvm to LD_LIBRARY_PATH.
+ENV LD_LIBRARY_PATH="/usr/local/cuda/nvvm/lib64:${LD_LIBRARY_PATH}"
+ENV LLVM_LINK_STATIC=1
+ENV RUST_LOG=info

container/ubuntu24-cuda13/Dockerfile

@@ -0,0 +1,89 @@
+FROM nvcr.io/nvidia/cuda:13.0.2-cudnn-devel-ubuntu24.04 AS llvm-builder
+
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -qq -y install \
+    build-essential \
+    clang \
+    curl \
+    libffi-dev \
+    libedit-dev \
+    libncurses5-dev \
+    libssl-dev \
+    libtinfo-dev \
+    libxml2-dev \
+    cmake \
+    ninja-build \
+    pkg-config \
+    python3 \
+    xz-utils \
+    zlib1g-dev && \
+    rm -rf /var/lib/apt/lists/*
+
+WORKDIR /data/llvm7
+
+# Download and build LLVM 7.1.0 for all architectures.
+RUN curl -sSf -L -O https://github.com/llvm/llvm-project/releases/download/llvmorg-7.1.0/llvm-7.1.0.src.tar.xz && \
+    tar -xf llvm-7.1.0.src.tar.xz && \
+    cd llvm-7.1.0.src && \
+    mkdir build && cd build && \
+    ARCH=$(dpkg --print-architecture) && \
+    if [ "$ARCH" = "amd64" ]; then \
+        TARGETS="X86;NVPTX"; \
+    else \
+        TARGETS="AArch64;NVPTX"; \
+    fi && \
+    cmake -G Ninja \
+        -DCMAKE_BUILD_TYPE=Release \
+        -DLLVM_TARGETS_TO_BUILD="$TARGETS" \
+        -DLLVM_BUILD_LLVM_DYLIB=ON \
+        -DLLVM_LINK_LLVM_DYLIB=ON \
+        -DLLVM_ENABLE_ASSERTIONS=OFF \
+        -DLLVM_ENABLE_BINDINGS=OFF \
+        -DLLVM_INCLUDE_EXAMPLES=OFF \
+        -DLLVM_INCLUDE_TESTS=OFF \
+        -DLLVM_INCLUDE_BENCHMARKS=OFF \
+        -DLLVM_ENABLE_ZLIB=ON \
+        -DLLVM_ENABLE_TERMINFO=ON \
+        -DCMAKE_INSTALL_PREFIX=/opt/llvm-7 \
+        .. && \
+    ninja -j$(nproc) && \
+    ninja install && \
+    cd ../.. && \
+    rm -rf llvm-7.1.0.src*
+
+FROM nvcr.io/nvidia/cuda:13.0.2-cudnn-devel-ubuntu24.04
+
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -qq -y install \
+    build-essential \
+    clang \
+    curl \
+    libssl-dev \
+    libtinfo-dev \
+    pkg-config \
+    xz-utils \
+    zlib1g-dev \
+    cmake \
+    libfontconfig-dev \
+    libx11-xcb-dev \
+    libxcursor-dev \
+    libxi-dev \
+    libxinerama-dev \
+    libxrandr-dev && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY --from=llvm-builder /opt/llvm-7 /opt/llvm-7
+RUN ln -s /opt/llvm-7/bin/llvm-config /usr/bin/llvm-config && \
+    ln -s /opt/llvm-7/bin/llvm-config /usr/bin/llvm-config-7
+
+# Get Rust (install rustup; toolchain installed from rust-toolchain.toml below)
+RUN curl -sSf -L https://sh.rustup.rs | bash -s -- -y --profile minimal --default-toolchain none
+ENV PATH="/root/.cargo/bin:${PATH}"
+
+# Setup the workspace
+WORKDIR /data/rust-cuda
+RUN --mount=type=bind,source=rust-toolchain.toml,target=/data/rust-cuda/rust-toolchain.toml \
+    rustup show
+
+# Add nvvm to LD_LIBRARY_PATH.
+ENV LD_LIBRARY_PATH="/usr/local/cuda/nvvm/lib64:${LD_LIBRARY_PATH}"
+ENV LLVM_LINK_STATIC=1
+ENV RUST_LOG=info

crates/nvvm/src/lib.rs

@@ -312,8 +312,10 @@ pub enum NvvmArch {
     /// This default value of 7.5 corresponds to Turing and later devices. We default to this
     /// because it is the minimum supported by CUDA 13.0 while being in the middle of the range
     /// supported by CUDA 12.x.
-    // WARNING: If you change the default, consider updating the `--target-arch` values used for
-    // compiletests in `ci_linux.yml` and `.github/workflows/ci_{linux,windows}.yml`.
+    // WARNING: If you change the default, consider updating:
+    // - The `--target-arch` values used for compiletests in `ci_linux.yml` and
+    //   `.github/workflows/ci_{linux,windows}.yml`.
+    // - The CUDA versions used in `setup_cuda_environment` in `compiletests`.
     #[default]
     Compute75,
     Compute80,

guide/src/SUMMARY.md

@@ -1,8 +1,6 @@
 # Summary 
 
-- [Introduction](README.md)
-- [Supported Features](features.md)
-- [Frequently Asked Questions](faq.md)
+- [Introduction](introduction.md)
 - [Guide](guide/README.md)
   - [Getting Started](guide/getting_started.md)
   - [Compute Capability Gating](guide/compute_capabilities.md)
@@ -18,3 +16,9 @@
   - [Types](nvvm/types.md)
   - [PTX Generation](nvvm/ptxgen.md)
   - [Debugging](nvvm/debugging.md)
+
+----
+
+[Supported Features](features.md)
+[Frequently Asked Questions](faq.md)
+