Hardware

TOPS

Tera (trillion) operations per second; the unit used to describe the peak performance of NPUs and AI accelerators.

// EXPLANATION

TOPS is typically quoted for low-precision (INT8) operations and is a peak figure; real performance depends on memory bandwidth, software and the precision used. TOPS values from different vendors cannot always be compared directly.

// ALSO KNOWN AS

TOPS

// RELATED TERMS

NPU (neural processing unit)A dedicated processing unit designed to accelerate neural-network computation (matrix/tensor operations) at very low energy. GPU (graphics processing unit)A processor with thousands of small cores that performs the same operation on lots of data in parallel, used for graphics and AI computation. QuantizationA technique that lowers memory and speed cost by storing model weights with fewer bits (for example 8 or 4 instead of 16).

// ARTICLES THAT MENTION THIS TERM

MOBILE MOBILE September 11, 2026 iPhone 18 Pro vs iPhone Duo: Pre-Order Hardware and AI Analysis While the smartphone industry has long been stuck in a vicious cycle of "the same chassis, just a little shinier", Apple's latest move has shaken things up. The traditional flagship iPhone 18 Pro … 4 MIN READ #SMARTPHONES

ARTIFICIAL_INTEL ARTIFICIAL_INTEL September 11, 2026 NVIDIA's End-to-End AI: DRIVE Thor and Autonomous Driving Although driverless car technologies have been around for years, our dream of fully autonomous (Level 5) driving has never quite come true. The biggest reason was the software forming the vehicle's brain… 4 MIN READ #HARDWARE

HARDWARE HARDWARE September 10, 2026 Battle for the Silicon Throne: Can the NPU Replace the GPU? As AI models have made their way down to our desktops and laptops, a new power struggle has broken out in the world of silicon. Discrete graphics cards, for years the undisputed leaders of AI computation… 4 MIN READ #HARDWARE

HARDWARE HARDWARE September 10, 2026 The Secret of Supersonic Speed: AMD Zen and AI on Silicon Processor performance is no longer decided by clock speed alone, but by architectural efficiency, IPC and AI accelerators built into the silicon. Here is how AMD's new Zen architecture does it… 4 MIN READ #AMD_RYZEN

ARTIFICIAL_INTEL ARTIFICIAL_INTEL September 9, 2026 Farewell to the Cloud: AMD Brings AI Straight to Your Desk Every cloud AI request crosses oceans and data centers, raising privacy and latency concerns. With Ryzen AI and the XDNA NPU, AMD runs AI locally on your own silicon… 4 MIN READ #AMD

← Whole glossary