If the CPU is the brain of your PC, the GPU is the artist. It’s the chip that draws every pixel you see on screen, and it does so by performing an absolutely staggering amount of math, very quickly, in parallel. If you’ve ever wondered why a graphics card can cost more than the rest of your PC combined, the answer is: because it’s doing the heaviest lifting in your machine.

GPU stands for Graphics Processing Unit

A GPU is a specialized processor designed to handle the kind of math that produces images. Where a CPU is a generalist with a handful of very capable cores, a GPU is a specialist with thousands of much simpler cores all running in parallel. That structure happens to be perfect for graphics because rendering a frame involves doing the same kind of calculation millions of times for millions of pixels.

There are two kinds of GPUs you’ll run into: integrated and dedicated.

  • Integrated GPUs share space (and memory) with the CPU. They’re small, low-power, and fine for office work, video playback, and very light gaming.
  • Dedicated GPUs sit on a separate card that plugs into your motherboard, with their own dedicated memory (VRAM) and their own cooling. This is what gamers and creators want.

What VRAM is and why it matters

VRAM is video memory — the dedicated RAM that lives on the graphics card itself. It stores all the textures, models, and frame buffers the GPU needs to access at high speed. The amount of VRAM your card has determines what resolutions and texture qualities you can run.

  • 8GB is the minimum for modern 1080p gaming, and even that is getting tight in some new titles.
  • 12GB is the comfortable sweet spot for 1440p in 2026.
  • 16GB or more is what you want for 4K gaming or for creative work like 3D rendering and AI workloads.

When a game runs out of VRAM, it can’t just borrow more from system RAM without performance falling off a cliff. That’s why a card with low VRAM can have great specs on paper but stutter horribly in modern games.

Clock speed, CUDA cores, stream processors — what do they mean?

Nvidia calls its parallel processing units CUDA cores. AMD calls them stream processors. Intel calls them Xe cores. They’re all roughly the same idea: tiny processors that work together. More of them, running faster, generally means more performance — but only when comparing within the same generation and brand. A 5000-core card from 2019 is not faster than a 3000-core card from 2025.

The honest truth is that the only number that matters for buying a GPU is real-world benchmark performance in the games or applications you actually use. Independent benchmark sites measure FPS (frames per second) at specific resolutions and settings. That number tells you more than any spec sheet.

Ray tracing, DLSS, FSR — the modern features

Three buzzwords dominate GPU marketing right now:

  • Ray tracing is a more realistic way of calculating lighting and reflections. It’s gorgeous in games that use it well, but expensive in performance.
  • DLSS (Nvidia) uses AI to upscale a lower-resolution image to a higher-resolution one, getting you more frames per second with minimal quality loss.
  • FSR (AMD) does the same thing as DLSS but works across all GPU brands.

These features matter a lot. A mid-range GPU with DLSS or FSR enabled can often match the raw performance of a much more expensive card without those features.

Power and cooling

Modern GPUs are power hungry. A high-end card can pull 350-450W on its own, and you need a power supply that can handle it. Always check the GPU’s recommended PSU wattage before you buy. You’ll also want a case with good airflow — these cards run hot.

Modern GPUs also need one or more PCIe power connectors plugged in from your PSU. Higher-end cards use the newer 12V-2×6 connector (formerly 12VHPWR). Make sure your PSU supports the right type, or use the adapter that comes in the GPU box.

How to pick a GPU

Here’s the realistic process I’d recommend:

  1. Decide what resolution and refresh rate you’ll game at (1080p 60Hz, 1440p 144Hz, 4K 60Hz, etc.).
  2. Pick a budget. GPU prices in 2026 range from around $200 for entry cards to $2000+ for flagships.
  3. Look up independent benchmarks for the games you play at your target resolution.
  4. Make sure your power supply can handle the card. Add some headroom.
  5. Check the card fits your case — modern GPUs are huge.

The honest take

Most people massively over-buy on GPUs. If you’re playing at 1080p and you’re happy at 60fps, you don’t need a top-of-the-line card. The sweet spot for most gamers in 2026 is the mid-range — these cards run almost everything beautifully at 1440p with all the modern features turned on, and they cost a fraction of the flagship.

Your GPU is probably going to be the most expensive part of your build, but it’s also the part that will most directly determine whether your PC is fun to use for gaming. Take your time with this one.

Related guide: How to Upgrade Your GPU: A Step-by-Step Guide