| Version 2 (modified by , 5 years ago) ( diff ) |
|---|
Supported Features
- CUDA kernels with the
__global__specifier - The dim3 struct type
- Use of the CUDA variables
threadIdx,blockIdx,gridDim, andblockDim __syncthreads__shared__- Enqueuing multiple kernel calls with the null stream
cudaMalloccudaMemcpycudaFreecudaDeviceSynchronize
Major Limitations
- Lacks support for other stream types
Missing Features
- Use of the
warpSizevariable - Atomic functions (e.g.
atomicAdd)
Note:
See TracWiki
for help on using the wiki.
