docs.rs failed to build cumath-0.2.4
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure docs.rs builds.
If you believe this is docs.rs' fault, open an issue.
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure docs.rs builds.
If you believe this is docs.rs' fault, open an issue.
cumath
Cuda-based matrix/vector computations
Install nvcc before : https://developer.nvidia.com/cuda-toolkit
/!\ This library is still under developement!
/!\ The detection of cuda compatible c compiler is very basic and only works for linux, if it doesn't work for you, you can try to comment this line in build.rs : "std::env::set_var("CXX", get_ccbin().unwrap());"
Implemented :
- GPU memory management
- Vector
- Matrix
- CuBLAS
- CuRAND
- Cuda streams
To be implemented :
- Data type genericity (being able to use integer vectors)
- User-defined Cuda kernels
- More built-in functions
- Improve automatic cuda-compatible c compiler detection
Won't be implemented
- Backend choice with CPU (This would rather be a higher level library)
Getting started
Add Cumath to your Cargo.toml:
[dependencies]
cumath = "0.2.2"
Then in your main.rs :
extern crate cumath;
Examples
Simple vector addition
extern crate cumath;
use *;
Matrix multiplication using CuBLAS
extern crate cumath;
use *;
For more info, run 'cargo doc --open'