What is the difference between a thread block and a warp (32 threads) ?
(CUDA 2.1 Programming Guide)
1 Like
A block is made up of warps. A warp is what executes on each SM at any given timestep.
Shared memory is shared among all threads in a block. Synchronization occurs at the block level. So the block is the ‘scope’ within which sets of threads can communicate.
A warp is a hardware detail which is important for performance, but less so for correctness.
Thanks a lot
What does “SM” stand for ?
EDIT : Streaming Multiprocessor
1 Like