Also known as: GPU, Graphics card

A graphics processing unit (GPU) is a processor built from many small cores that run the same operation across large blocks of data in parallel — originally for rendering graphics, now also for general computation.¹

Overview

Where a CPU has a few powerful cores tuned for sequential work, a GPU has hundreds or thousands of simpler cores that excel at data-parallel math: the same calculation applied to many pixels, vertices, or array elements at once. A discrete GPU is a card that plugs into a PCIe slot with its own high-bandwidth memory; integrated GPUs share the CPU’s memory. Modern GPUs are large integrated circuits whose transistor counts have ridden Moore’s law upward for decades.

What it’s for

Beyond 3D graphics, GPUs power AI training and inference, scientific simulation, and signal processing — any workload that maps onto wide parallel arithmetic. In SDR work a GPU can accelerate FFTs and filtering across many channels at once, complementing the streaming DSP GopherTrunk runs on the CPU. The trade-off is latency and complexity: GPUs shine on big batched workloads, less so on small, branchy, latency-sensitive tasks where the CPU wins.

Sources

1. Graphics processing unit — Wikipedia, on GPUs and parallel computation. ↩

See also

• Central processing unit (CPU)
• Integrated circuit
• PCI Express (PCIe)
• Instruction set architecture (ISA)
• Moore's law
• System bus