36,657 views
Graphics cards can run the most incredible games, but how many calculations do they perform every second? The most advanced GPUs perform thirty-six trillion or more calculations every second. But how can a single device handle tens of trillions of calculations? In this video, we explore the architecture inside the thirty-ninety graphics card and the architecture of the GA-102 chip. Note: We decided to show the 30 series graphics cards because we had to tear down a 3090 graphics card quite destructively to create accurate 3D models. We usually choose slightly older models because we can find broken components on ebay. In case you were wondering, the 4090 can perform 82.58 trillion calculations per second, and we are sure that the 5090 will be even more powerful. High-quality AI-controlled video translation from the channel - @BranchEducation About the translation. This time, Bes had a hard time with numbers. And not only problems with pronunciation. As an example, he persistently read 1996 as 1906. Well, and the classic 80 he reads as 2 words 8 and 10, putting 2 stresses. He also constantly confused the bits, constantly calling them bits. Original video — • How do Graphics Cards Work? Explorin... Contents: 00:00 - How many calculations do graphics cards perform? 02:15 - Difference between GPU and CPU 04:56 - GA-102 GPU architecture 06:59 - GA-102 GPU manufacturing 08:48 - CUDA core structure 11:09 - Video card components 12:04 - GDDR6X and GDDR7 graphics memory 15:11 - All about Micron 16:51 - Single instruction multiple data architecture 17:49 - Why GPUs process game graphics, object transformations 20:53 - Thread architecture 23:31 - Help Branch Education develop! 24:29 - Bitcoin mining 26:50 - Tensor cores 27:58 - Conclusion