chaosink
Reputation: 1483
Why is cudaOccupancyMaxActiveBlocksPerMultiprocessor() independent of device?
Different devices may have different shared memory sizes, register counts.
Why is cudaOccupancyMaxActiveBlocksPerMultiprocessor() independent of device?
It doesn't need a device as a parameter.
Upvotes: 1
Views: 621
Answers (1)
Related Questions
- Random Occupancy values returned by the "cudaOccupancyMaxActiveBlocksPerMultiprocessor"
- cudaOccupancyMaxPotentialBlockSize API template usage
- Thread blocks active per SM
- CUDA device runtime api cudaMemsetAsync doesn't work
- cudaFuncSetCacheConfig - applies to all devices or current device?
- cudaOccupancyMaxActiveBlocksPerMultiprocessor is undefined
- cudaMemcpyFromSymbol on a __device__ variable
- Is cudaEvent device specific?
- CUDA kernel behavior depending on the number of blocks and threads even if not used
- Does cudaFreeHost care what device is active when cudaMallocHost is used to allocate memory?