Hardware

A set of wires (or differential lanes) that carries data, addresses, or control signals between components inside a computer.

The component of a computer that fetches and executes instructions — the place where programs actually run.

A processor designed for massive data-parallel work — originally for rendering graphics, now also the workhorse of machine learning, simulation, and crypto.

Tiny electronic circuits that implement boolean operations — the physical building blocks of every digital chip.

Fast, temporary storage where a running program keeps the data it is actively using.

Long-term, non-volatile storage where files and programs are kept when the computer is off.

The microscopic electronic switch that, replicated by the billions, forms every modern integrated circuit and CPU.

The steady electronic pulse that synchronizes a digital circuit — every "tick" advances the CPU and other components through their next step, and its rate is measured in hertz.

A storage device that records data magnetically on spinning platters read by a moving head — cheap per gigabyte and high-capacity, but far slower than an SSD.

The main circuit board that connects and powers every component of a computer — CPU, memory, storage, and peripherals — and lets them communicate over shared buses.

Any device attached to a computer that isn't the core CPU and memory — keyboards, mice, displays, printers, drives, cameras — through which the machine takes input and produces output.

A storage device that keeps data in flash memory chips with no moving parts — far faster than a spinning hard drive, and now the default storage in most computers.

The two main kinds of volatile memory — DRAM is dense and cheap but slow and needs refreshing; SRAM is fast and stable but bulky and expensive, so it's used for CPU caches.

Non-volatile memory that stores data by trapping charge in transistor cells — retaining it without power, and the technology inside SSDs, USB sticks, and phone storage.

Gordon Moore's 1965 observation that the number of transistors on a chip doubles roughly every two years — the empirical trend that drove 50 years of computer performance improvements and is now slowing.

Google's custom ASIC for accelerating machine learning workloads — a matrix multiplication engine built as a systolic array, providing far higher throughput and power efficiency than GPUs for transformer and convolution operations.

A chip designed for a single specific purpose — providing the maximum possible performance and power efficiency at the cost of inflexibility and multi-million dollar fabrication investment, used for Bitcoin mining, AI accelerators, and mobile SoCs.

The empirical rule that as transistors shrink, their power consumption decreases proportionally — allowing faster clock speeds at the same power budget. It broke down around 2004, ending the era of ever-faster single-core CPUs.

An integrated circuit containing an array of programmable logic blocks connected by configurable interconnects — allowing hardware circuits to be defined in software, reprogrammed after fabrication, and used for custom acceleration.