GPU//architecture

- The internal design of a GPU — streaming multiprocessors (NVIDIA) or compute units (AMD) grouped into warps/wavefronts executing in lockstep.

The internal design of a GPU — streaming multiprocessors (NVIDIA) or compute units (AMD) grouped into warps/wavefronts executing in lockstep.

Memory hierarchy includes registers, shared memory, L1/L2 caches, and high-bandwidth GDDR or HBM.

Architecture choices determine throughput for graphics, ML training, and scientific workloads.

discrete cards push bleeding-edge architectures; integrated units share die area with the CPU