stream_executor_example.cc

// Load GPU kernel
gpu::StreamExecutor stream_exec{PlatformKind::kCuda};
gcudacc::kernel::MyKernel my_kernel{&stream_exec};
bool ok = stream_exec.GetKernel(gcudacc::spec::MyKernelSpec(), &my_kernel);
if (!ok) { ... }

// Allocate some CPU memory for the output from the GPU
gpu::DeviceMemory<int> result = stream_exec.AllocateZeroed<int>();
if (result == nullptr) { ... }

// Perform a sequence of GPU operations using the Stream.
int host_result;
gpu::Stream my_stream{&stream_exec};
my_stream
 .Init()
 .ThenLaunch(ThreadDim{1024}, BlockDim{1}, my_kernel, result)
 .ThenMemcpy(&host_result, result, sizeof(host_result))
 .BlockHostUntilDone()
if (!my_stream.ok()) { ... }
printf("%d\n", host_result);