CUDA Calc
Given your register, shared memory, thread, and block requirements, CUDA Calc will determine resource bottlenecks.
Inputs
Program Requirements
(32-Bit Registers Per Thread)
(Shared Memory Per Block, KB)
(Threads Per Block)
Hardware Limits
Compute Capability:
(Max 32-Bit Registers Per Thread)
(Max 32-Bit Registers Per MP)
(Max Threads Per Block)
(Max Resident Threads Per MP)
(Max Resident Warps Per MP)
(Max Resident Blocks Per MP)
(Max Shared Memory Per MP, KB)
(Threads Per Warp)