CUDA Support

Creating GPU Tensors

C API:

size_t shape[] = {1000, 1000};
Tensor* t = tensr_zeros(shape, 2, TENSR_FLOAT32, TENSR_CUDA);

C++ API:

auto t = tensr::Tensor::zeros({1000, 1000}, tensr::DType::Float32, tensr::Device::CUDA);

Device Transfer

C API:

tensr_to_device(t, TENSR_CUDA, 0);

C++ API:

t.to(tensr::Device::CUDA, 0);

Synchronization

C API:

tensr_synchronize(TENSR_CUDA, 0);

C++ API:

tensr::synchronize(tensr::Device::CUDA, 0);

Device Count

C API:

int count = tensr_device_count(TENSR_CUDA);

C++ API:

int count = tensr::device_count(tensr::Device::CUDA);