|
Uni20
Asynchronous tensor and runtime library.
|
cuda_hello_world_example.cpp reports whether CUDA was enabled in the current Uni20 build. CUDA-enabled builds show the compile-time, runtime, and driver versions, enumerate visible devices, display their cached hardware capabilities, install the scoped process-wide CUDA runtime, and exercise each canonical per-device stream/completion pool. CPU-only builds show the CMake option and runtime requirements needed to enable the CUDA path.Build and run the example with:
See the examples index, the introductory CUDA Buffersguide", and the <a href="../../docs/backends/cuda/" >CUDA backend documentation index.