Subject

    Computer Science

    Topic

    Computer Hardware

    Which GPU compute model runs many lightweight threads with the same kernel over data?

    ক)
    MIMD only
    খ)
    SIMT (Single Instruction Multiple Threads)
    গ)
    SISD
    ঘ)
    Distributed MPP only

    Explanation

    SIMT groups threads into warps/wavefronts that execute in lockstep on vector units. Divergence reduces efficiency when threads branch differently. It maps well to graphics and data-parallel workloads. Toolchains like CUDA/OpenCL expose the model.

    Find Tutors

    View all Tutors