Similar to SIMD. Single instruction multiple threads, the goal here is to run on a vector of data but the vector is spread out over the GPU.
Similar to SIMD. Single instruction multiple threads, the goal here is to run on a vector of data but the vector is spread out over the GPU.