How do Graphics Cards Work? Exploring GPU Architecture

Interested in working with Micron to make cutting-edge memory chips? Work at Micron: https://bit.ly/micron-careers
Learn more about Micron's Graphic Memory! Explore Here: https://bit.ly/micron-graphic-memory
Curious about AI memory and HBM3E? Take a look: https://bit.ly/micron-hbm3e

Graphics Cards can run some of the most incredible video games, but how many calculations do they perform every single second? Well, some of the most advanced graphics perform 36 Trillion calculations or more every single second. But how can a single device manage these tens of trillions of calculations? In this video, we explore the architecture inside the 3090 graphics card and the GA102 GPU chip architecture.

Note: We chose to feature the 30 series of GPUs because, to create accurate 3D models, we had to tear down a 3090 GPU rather destructively. We typically select a slightly older model because we're able to find broken components on eBay. If you're wondering, the 4090 can perform 82.58 trillion calculations a second, and then we're sure the 5090 will be even more.

Table of Contents:
00:00 - How many calculations do Graphics Cards Perform?
02:15 - The Difference between GPUs and CPUs?
04:56 - GPU GA102 Architecture
06:59 - GPU GA102 Manufacturing
08:48 - CUDA Core Design
11:09 - Graphics Cards Components
12:04 - Graphics Memory GDDR6X GDDR7
15:11 - All about Micron
16:51 - Single Instruction Multiple Data Architecture
17:49 - Why GPUs run Video Game Graphics, Object Transformations
20:53 - Thread Architecture
23:31 - Help Branch Education Out!
24:29 - Bitcoin Mining
26:50 - Tensor Cores
27:58 - Outro

We're working on more ambitious subjects like computer architecture and graphics cards. Any contribution would greatly help make these videos. https://www.patreon.com/brancheducation
Branch Education Website: https://www.branch.education
Branch Education Facebook: https://www.facebook.com/BranchEducation/

Key Branches from this video are: How do Video Game Graphics Work? https://youtu.be/C8YtdC8mxTU

Animation: Mike Radjabov, Sherdil Davronov, Adrei Dulay, Parvesh Khatri
Research, Script and Editing: Teddy Tablante
Twitter: @teddytablante
Modeling: Mike Radjabov, Prakash Kakadiya
Voice Over: Phil Lee
Sound Design by Drilu: www.drilu.world
Sound Design and mix: David Pinete
Additional Sound Design: Raúl Núñez
Supervising Sound Editor: Luis Huesca

Erratum:
04:50 Ubuntu is a type of Linux
08:45 3080 has 10GB, not 12GB.

Image Attribution:
Williams, George. Jacquard Loom 6/6 https://www.flickr.com/photos/....ghwpix/18056523/in/a
Eiermann, Georg. A close up of a weaving machine in a room. Jacquard Loom. https://unsplash.com/photos/a-....close-up-of-a-weavin

Wikipedia contributors. "Embarrassingly Parallel", "Graphics Processing Unit", "Parallel Computing" , " SIMD", " Single Instruction, Multiple Threads" , "Thread block (CUDA Programming)". Wikipedia, The Free Encyclopedia. Wikipedia, The Free Encyclopedia, Visited October 18th 2024

Textbooks and Papers
A TON of information was found in Nvidia's white papers. You can find them here:
https://research.nvidia.com/publications
We recommend the GA102 architecture white paper and the fermi architecture white paper.

#gpu #graphicscard #micron