Onnxruntime v1.8.1 with GPU/CUDA 11 support

.github/workflows/general.yml

@@ -73,21 +73,17 @@ jobs:
       # ******************************************************************
       - name: Download prebuilt archive (GPU, x86_64-unknown-linux-gnu)
         uses: actions-rs/cargo@v1
-        env:
-          ORT_USE_CUDA: "yes"
         with:
           command: build
-          args: --target x86_64-unknown-linux-gnu
+          args: --target x86_64-unknown-linux-gnu --features cuda
       - name: Verify prebuilt archive downloaded (GPU, x86_64-unknown-linux-gnu)
         run: ls -lh target/x86_64-unknown-linux-gnu/debug/build/onnxruntime-sys-*/out/onnxruntime-linux-x64-gpu-1.*.tgz
       # ******************************************************************
       - name: Download prebuilt archive (GPU, x86_64-pc-windows-msvc)
         uses: actions-rs/cargo@v1
-        env:
-          ORT_USE_CUDA: "yes"
         with:
           command: build
-          args: --target x86_64-pc-windows-msvc
+          args: --target x86_64-pc-windows-msvc --features cuda
       - name: Verify prebuilt archive downloaded (GPU, x86_64-pc-windows-msvc)
         run: ls -lh target/x86_64-pc-windows-msvc/debug/build/onnxruntime-sys-*/out/onnxruntime-win-gpu-x64-1.*.zip
 
@@ -141,7 +137,6 @@ jobs:
         uses: actions-rs/cargo@v1
         with:
           command: build
-          # Use --manifest-path instead of --package. See https://github.com/actions-rs/cargo/issues/86
           args: --manifest-path onnxruntime/Cargo.toml --features model-fetching
       - name: Test onnxruntime-sys
         uses: actions-rs/cargo@v1

README.md

@@ -11,8 +11,8 @@ This is an attempt at a Rust wrapper for
 
 This project consist on two crates:
 
-* [`onnxruntime-sys`](onnxruntime-sys): Low-level binding to the C API;
-* [`onnxruntime`](onnxruntime): High-level and safe API.
+- [`onnxruntime-sys`](onnxruntime-sys): Low-level binding to the C API;
+- [`onnxruntime`](onnxruntime): High-level and safe API.
 
 [Changelog](CHANGELOG.md)
 
@@ -21,25 +21,25 @@ which provides the following targets:
 
 CPU:
 
-* Linux x86_64
-* macOS x86_64
-* macOS aarch64 (no pre-built binaries, no CI testing, see [#74](https://github.com/nbigaouette/onnxruntime-rs/pull/74))
-* Windows i686
-* Windows x86_64
+- Linux x86_64
+- macOS x86_64
+- macOS aarch64 (no pre-built binaries, no CI testing, see [#74](https://github.com/nbigaouette/onnxruntime-rs/pull/74))
+- Windows i686
+- Windows x86_64
 
 GPU:
 
-* Linux x86_64
-* Windows x86_64
+- Linux x86_64
+- Windows x86_64
 
 ---
 
 **WARNING**:
 
-* This is an experiment and work in progress; it is _not_ complete/working/safe. Help welcome!
-* Basic inference works, see [`onnxruntime/examples/sample.rs`](onnxruntime/examples/sample.rs) or [`onnxruntime/tests/integration_tests.rs`](onnxruntime/tests/integration_tests.rs)
-* ONNX Runtime has many options to control the inference process but those options are not yet exposed.
-* This was developed and tested on macOS Catalina. Other platforms should work but have not been tested.
+- This is an experiment and work in progress; it is _not_ complete/working/safe. Help welcome!
+- Basic inference works, see [`onnxruntime/examples/sample.rs`](onnxruntime/examples/sample.rs) or [`onnxruntime/tests/integration_tests.rs`](onnxruntime/tests/integration_tests.rs)
+- ONNX Runtime has many options to control the inference process but those options are not yet exposed.
+- This was developed and tested on macOS Catalina. Other platforms should work but have not been tested.
 
 ---
 
@@ -58,11 +58,21 @@ To select which strategy to use, set the `ORT_STRATEGY` environment variable to:
 3. `compile`: To compile the library
 
 The `download` strategy supports downloading a version of ONNX that supports CUDA. To use this, set the
-environment variable `ORT_USE_CUDA=1` (only supports Linux or Windows).
+feature `cuda` in `Cargo.toml`.
 
 Until the build script allow compilation of the runtime, see the [compilation notes](ONNX_Compilation_Notes.md)
 for some details on the process.
 
+### Note on using CUDA
+
+To use CUDA you will need to set the feature `cuda` but also to set your session with the method `use_cuda` as such:
+
+```
+    let mut session = environment
+        .new_session_builder()?
+        .use_cuda(0)?
+```
+
 ### Note on 'ORT_STRATEGY=system'
 
 When using `ORT_STRATEGY=system`, executing a built crate binary (for example the tests) might fail, at least on macOS,
@@ -76,9 +86,9 @@ dyld: Library not loaded: @rpath/libonnxruntime.1.7.1.dylib
 
 To fix, one can either:
 
-* Set the `LD_LIBRARY_PATH` environment variable to point to the path where the library can be found.
-* Adapt the `.cargo/config` file to contain a linker flag to provide the **full** path:
-  
+- Set the `LD_LIBRARY_PATH` environment variable to point to the path where the library can be found.
+- Adapt the `.cargo/config` file to contain a linker flag to provide the **full** path:
+
   ```toml
   [target.aarch64-apple-darwin]
   rustflags = ["-C", "link-args=-Wl,-rpath,/full/path/to/onnxruntime/lib"]
@@ -232,7 +242,7 @@ Generate the bindings:
 ```sh
 ❯ docker exec -it --user "$(id -u)":"$(id -g)" rustbuilder /bin/bash
 ❯ cd onnxruntime-sys
-❯ cargo build --features generate-bindings
+❯ cargo build --features 'generate-bindings, cuda'
 ```
 
 ### Generating Bindings for Windows With Vagrant
@@ -245,8 +255,8 @@ Windows can build both x86 and x86_64 bindings:
 ```sh
 ❯ rustup target add i686-pc-windows-msvc x86_64-pc-windows-msvc
 ❯ cd onnxruntime-sys
-❯ cargo build --features generate-bindings --target i686-pc-windows-msvc
-❯ cargo build --features generate-bindings --target x86_64-pc-windows-msvc
+❯ cargo build --features 'generate-bindings, cuda' --target i686-pc-windows-msvc
+❯ cargo build --features 'generate-bindings, cuda' --target x86_64-pc-windows-msvc
 ```
 
 ## Conduct
@@ -259,9 +269,9 @@ instead of the Rust moderation team.
 
 This project is licensed under either of
 
-* Apache License, Version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or
+- 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
+- MIT license ([LICENSE-MIT](LICENSE-MIT) or
   http://opensource.org/licenses/MIT)
 
 at your option.

onnxruntime-sys/Cargo.toml

@@ -34,6 +34,7 @@ default = []
 disable-sys-build-script = []
 # Use bindgen to generate bindings in build.rs
 generate-bindings = ["bindgen"]
+cuda = []
 
 [package.metadata.docs.rs]
 # Disable the build.rs on https://docs.rs since it can cause

onnxruntime-sys/build.rs

@@ -28,8 +28,6 @@ const ORT_ENV_STRATEGY: &str = "ORT_STRATEGY";
 /// Name of environment variable that, if present, contains the location of a pre-built library.
 /// Only used if `ORT_STRATEGY=system`.
 const ORT_ENV_SYSTEM_LIB_LOCATION: &str = "ORT_LIB_LOCATION";
-/// Name of environment variable that, if present, controls wether to use CUDA or not.
-const ORT_ENV_GPU: &str = "ORT_USE_CUDA";
 
 /// Subdirectory (of the 'target' directory) into which to extract the prebuilt library.
 const ORT_PREBUILT_EXTRACT_DIR: &str = "onnxruntime";
@@ -54,7 +52,6 @@ fn main() {
     println!("cargo:rustc-link-search=native={}", lib_dir.display());
 
     println!("cargo:rerun-if-env-changed={}", ORT_ENV_STRATEGY);
-    println!("cargo:rerun-if-env-changed={}", ORT_ENV_GPU);
     println!("cargo:rerun-if-env-changed={}", ORT_ENV_SYSTEM_LIB_LOCATION);
 
     generate_bindings(&include_dir);
@@ -280,17 +277,6 @@ enum Accelerator {
     Gpu,
 }
 
-impl FromStr for Accelerator {
-    type Err = String;
-
-    fn from_str(s: &str) -> Result<Self, Self::Err> {
-        match s.to_lowercase().as_str() {
-            "1" | "yes" | "true" | "on" => Ok(Accelerator::Gpu),
-            _ => Ok(Accelerator::None),
-        }
-    }
-}
-
 impl OnnxPrebuiltArchive for Accelerator {
     fn as_onnx_str(&self) -> Cow<str> {
         match self {
@@ -353,6 +339,12 @@ impl OnnxPrebuiltArchive for Triplet {
 }
 
 fn prebuilt_archive_url() -> (PathBuf, String) {
+    let accelerator = if cfg!(feature = "cuda") {
+        Accelerator::Gpu
+    } else {
+        Accelerator::None
+    };
+
     let triplet = Triplet {
         os: env::var("CARGO_CFG_TARGET_OS")
             .expect("Unable to get TARGET_OS")
@@ -362,7 +354,7 @@ fn prebuilt_archive_url() -> (PathBuf, String) {
             .expect("Unable to get TARGET_ARCH")
             .parse()
             .unwrap(),
-        accelerator: env::var(ORT_ENV_GPU).unwrap_or_default().parse().unwrap(),
+        accelerator,
     };
 
     let prebuilt_archive = format!(

onnxruntime-sys/examples/c_api_sample.rs

@@ -49,8 +49,6 @@ fn main() {
     };
 
     // Optionally add more execution providers via session_options
-    // E.g. for CUDA include cuda_provider_factory.h and uncomment the following line:
-    // OrtSessionOptionsAppendExecutionProvider_CUDA(sessionOptions, 0);
 
     //*************************************************************************
     // create session and load model into memory