Skip to main content

GPU

A graphics processing unit. Hardware with many parallel cores, well suited to large array and matrix operations.

Well suited to large array arithmetic and poorly suited to small or branch-heavy work. Moving data to the GPU costs time, so a small operation can measure slower there than on the CPU. Verify a speedup rather than assuming one.

GPU & acceleration

Related terms

  • GPU offloading: Moving computationally intensive operations onto a GPU. Eugo applies this automatically when a GPU is available and the data transfer is worth its cost.
  • Compute node: A machine in a cluster that executes tasks. Compute nodes do the work; the head node coordinates it.
  • options(): The method that sets per-call resource requirements on a distributed function, such as `num_cpus` or `num_gpus`.