Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Visualizing CUDA Cores: How do Graphics Cards Work? Exploring GPU Architecture

Mohit Mishra

25,657 subscribers

66,751 views • 1 year ago •via X (Twitter)

Health & Wellness Education Science & Technology

Anya Rossi• Live Now

Private livecam show

7 Comments

Mohit Mishra1 year ago

More about GPU Memory Data Flow Hierarchy

Mohit Mishra1 year ago

Source: Feel free to join our discord server for more such low-level stuff discussions and work together

Bill Wallace, PhD1 year ago

rtfm

Dani Ihza Farrosi1 year ago

Great YT channel, their animation and visualization is amazing

Mohit Mishra1 year ago

true, one of the best one out there for understanding concepts with visualization

عبد الحافظ1 year ago

@ezdubs_bot english to Arabic

HeyGen Labs1 year ago

@chessMan786 @ezdubs_bot @_iraqi2022 Here's your translated video: Try HeyGen, our translations include lip-syncing & voice cloning.

Related Videos

Visualizing architecture by standing inside the idea

Visualizing architecture by standing inside the idea

Science girl

16,896 views • 3 months ago

Introducing GeForce RTX 50 Series graphics cards & laptops. Powered by NVIDIA Blackwell architecture & AI, the RTX 50 Series enables new experiences & next-level graphics fidelity. Learn More →

Introducing GeForce RTX 50 Series graphics cards & laptops. Powered by NVIDIA Blackwell architecture & AI, the RTX 50 Series enables new experiences & next-level graphics fidelity. Learn More →

NVIDIA GeForce

1,028,522 views • 1 year ago

How DDA raymarching for graphics work.

How DDA raymarching for graphics work.

Elliot Arledge

98,202 views • 3 months ago

I feel it should be illegal to have two processors that share the same name (Z2 A and Z2 Extreme) have completely different GPU cores, the Z2 A is an updated version of the Steam Deck chip with RDNA 2 graphics, whereas the Z2 Extreme is equipped with RDNA 3.5 graphics. Source: Linus Tech Tips

I feel it should be illegal to have two processors that share the same name (Z2 A and Z2 Extreme) have completely different GPU cores, the Z2 A is an updated version of the Steam Deck chip with RDNA 2 graphics, whereas the Z2 Extreme is equipped with RDNA 3.5 graphics. Source: Linus Tech Tips

NikTek

146,928 views • 7 months ago

You can run CUDA, on a Mac ARM GPU, in the browser. It sounds ridiculous but it actually works. HipScript chains CUDA, to OpenCL, to Vulkan, to Tint (Google’s shader translator), to a WASM WebGPU. I got a plasma simulation in running in just a few minutes, no NVIDIA GPU!

You can run CUDA, on a Mac ARM GPU, in the browser. It sounds ridiculous but it actually works. HipScript chains CUDA, to OpenCL, to Vulkan, to Tint (Google’s shader translator), to a WASM WebGPU. I got a plasma simulation in running in just a few minutes, no NVIDIA GPU!

LaurieWired

160,223 views • 1 year ago

China just broke the internet with the most insane IPO ever Moore Threads, founded by Nvidia’s former China chief, raised bids worth $4.5 trillion for a $1.1b IPO (4,126x oversubscribed!). US chip bans created a vacuum… and now China’s building its own GPU empire with MUSA architecture, MUSIFY (CUDA killer), & 10,000-GPU superclusters. Nvidia may still rule performance today, but tomorrow? Shreya Chandra gets you an answer #MooreThreads #ChinaTech #NvidiaKiller #AI #GPU #CNBCTV18Digital #ChineseIPO

China just broke the internet with the most insane IPO ever Moore Threads, founded by Nvidia’s former China chief, raised bids worth $4.5 trillion for a $1.1b IPO (4,126x oversubscribed!). US chip bans created a vacuum… and now China’s building its own GPU empire with MUSA architecture, MUSIFY (CUDA killer), & 10,000-GPU superclusters. Nvidia may still rule performance today, but tomorrow? Shreya Chandra gets you an answer #MooreThreads #ChinaTech #NvidiaKiller #AI #GPU #CNBCTV18Digital #ChineseIPO

CNBC-TV18

271,382 views • 6 months ago

Experimental "No Graphics API" path tracer - PTX to instruct the GPU. No OpenGL, VK, DX (no graphics api) - No tris/verts, only planes - Bindless, raw pointers to scene data Maintainability 1 out of 10, but I do enjoy breaking free of graphics APIs and their paradigms.

Experimental "No Graphics API" path tracer - PTX to instruct the GPU. No OpenGL, VK, DX (no graphics api) - No tris/verts, only planes - Bindless, raw pointers to scene data Maintainability 1 out of 10, but I do enjoy breaking free of graphics APIs and their paradigms.

Crayz

107,980 views • 5 months ago

WOW, big news from NVIDIA 🔥 They just launches Jetson Orin Nano Super Developer Kit at $249, delivering 67 TOPS (Trillion Operations per second) of AI performance - a 70% boost over previous version through enhanced GPU, memory, and CPU clocks. → The kit packs an Ampere architecture GPU with 1024 CUDA cores, 32 tensor cores, and a 6-core ARM CPU, targeted at edge AI development and robotics applications. → Developers can achieve the performance boost through a software upgrade, making advanced AI capabilities more accessible for hobbyists and makers. 🔧 Key Highlights → Hardware specifications showcase serious edge computing muscle Features an Ampere GPU with 1024 CUDA cores and 32 tensor cores. The 6-core ARM Cortex-A78AE CPU paired with 8GB LPDDR5 memory delivers 102GB/s bandwidth. → The $249 price point makes advanced AI development accessible. Existing Jetson Orin Nano users can unlock the performance boost through a software update. → Comprehensive connectivity options enable diverse applications Two MIPI CSI-2 camera connectors, multiple USB 3.2 ports, and PCIe expansion slots provide flexibility for various sensors and peripherals. → Robust software ecosystem support Comes with NVIDIA's AI stack including Isaac for robotics, Metropolis for vision AI, and Holoscan for sensor processing. Supports TAO Toolkit for model fine-tuning. NVIDIA AI Developer

WOW, big news from NVIDIA 🔥 They just launches Jetson Orin Nano Super Developer Kit at $249, delivering 67 TOPS (Trillion Operations per second) of AI performance - a 70% boost over previous version through enhanced GPU, memory, and CPU clocks. → The kit packs an Ampere architecture GPU with 1024 CUDA cores, 32 tensor cores, and a 6-core ARM CPU, targeted at edge AI development and robotics applications. → Developers can achieve the performance boost through a software upgrade, making advanced AI capabilities more accessible for hobbyists and makers. 🔧 Key Highlights → Hardware specifications showcase serious edge computing muscle Features an Ampere GPU with 1024 CUDA cores and 32 tensor cores. The 6-core ARM Cortex-A78AE CPU paired with 8GB LPDDR5 memory delivers 102GB/s bandwidth. → The $249 price point makes advanced AI development accessible. Existing Jetson Orin Nano users can unlock the performance boost through a software update. → Comprehensive connectivity options enable diverse applications Two MIPI CSI-2 camera connectors, multiple USB 3.2 ports, and PCIe expansion slots provide flexibility for various sensors and peripherals. → Robust software ecosystem support Comes with NVIDIA's AI stack including Isaac for robotics, Metropolis for vision AI, and Holoscan for sensor processing. Supports TAO Toolkit for model fine-tuning. NVIDIA AI Developer

Rohan Paul

24,713 views • 1 year ago

New blackboard lecture w Reiner Pope How do chips actually work – starting with basic logic gates, and working up to why GPUs, TPUs, FPGAs, and the human brain each look the way they do. 0:00:00 – Building a multiply-accumulate from logic gates 0:16:20 – Muxes and the cost of data movement 0:25:59 – How systolic arrays work 0:39:00 – Clock cycles and pipeline registers 0:51:40 – FPGAs vs ASICs 1:03:14 – Cache vs scratchpad 1:07:16 – Why CPU cores are much bigger than GPU cores 1:11:49 – Brains vs chips 1:15:22 – A GPU is just a bunch of tiny TPUs Look up Dwarkesh Podcast on YouTube/Spotify/etc to watch. Enjoy!

New blackboard lecture w Reiner Pope How do chips actually work – starting with basic logic gates, and working up to why GPUs, TPUs, FPGAs, and the human brain each look the way they do. 0:00:00 – Building a multiply-accumulate from logic gates 0:16:20 – Muxes and the cost of data movement 0:25:59 – How systolic arrays work 0:39:00 – Clock cycles and pipeline registers 0:51:40 – FPGAs vs ASICs 1:03:14 – Cache vs scratchpad 1:07:16 – Why CPU cores are much bigger than GPU cores 1:11:49 – Brains vs chips 1:15:22 – A GPU is just a bunch of tiny TPUs Look up Dwarkesh Podcast on YouTube/Spotify/etc to watch. Enjoy!

Dwarkesh Patel

923,242 views • 16 days ago

3080 vs 4080 vs 5080 GPU Marvel Rivals Benchmark Brawl! We put a 3080, 4080, and 5080 NVIDIA GeForce RTX graphics card to the test in Marvel Rivals at high graphics settings with frame generation active. #newegg #gaming #pcgaming #MarvelRivals #GPU

3080 vs 4080 vs 5080 GPU Marvel Rivals Benchmark Brawl! We put a 3080, 4080, and 5080 NVIDIA GeForce RTX graphics card to the test in Marvel Rivals at high graphics settings with frame generation active. #newegg #gaming #pcgaming #MarvelRivals #GPU

Newegg

10,491 views • 1 year ago

Meet the new #IntelCoreUltra Series 3—with new E-cores and P-cores, a massive GPU, and a highly efficient NPU built into a smaller area. See it at #CES2026

Meet the new #IntelCoreUltra Series 3—with new E-cores and P-cores, a massive GPU, and a highly efficient NPU built into a smaller area. See it at #CES2026

Intel

31,913 views • 5 months ago

NEW: Microsoft just unveiled its most powerful Surface ever. Surface Laptop Ultra is a 15-inch NVIDIA-powered laptop built to run local AI models, creative work, and gaming. • NVIDIA RTX Spark chip with up to 20 Arm CPU cores • Blackwell GPU with 6,144 CUDA cores • Up to 128GB unified memory • Runs up to 120B parameter AI models locally • 15-inch mini-LED display with 2,000 nits peak brightness • Dual-fan cooling with 2.5x the thermal capacity of Surface Laptop 7 This could be a serious AI workstation.

NEW: Microsoft just unveiled its most powerful Surface ever. Surface Laptop Ultra is a 15-inch NVIDIA-powered laptop built to run local AI models, creative work, and gaming. • NVIDIA RTX Spark chip with up to 20 Arm CPU cores • Blackwell GPU with 6,144 CUDA cores • Up to 128GB unified memory • Runs up to 120B parameter AI models locally • 15-inch mini-LED display with 2,000 nits peak brightness • Dual-fan cooling with 2.5x the thermal capacity of Surface Laptop 7 This could be a serious AI workstation.

Ritwik Pavan

92,958 views • 6 days ago

One more.. visualizing how RSA encryption works.

One more.. visualizing how RSA encryption works.

Ado

74,404 views • 2 months ago

🎉 "This is the 20th anniversary of CUDA. We have been working on this architecture for 20 years ... to now have built up hundreds of millions of GPUs and computing systems around the world that run CUDA."

🎉 "This is the 20th anniversary of CUDA. We have been working on this architecture for 20 years ... to now have built up hundreds of millions of GPUs and computing systems around the world that run CUDA."

NVIDIA HPC Developer

27,642 views • 2 months ago

Nvidia announces the new RTX Spark, a new platform powered by the NX1 CPU, and shows off Spark laptops running 007 First Light and Forza 6. The CPU has 20 ARM based cores and a Blackwell RTX GPU with 6144 CUDA Cores. This is the same core count as a 5070, but with 128GB of unified LPDDR5X RAM memory sitting in the same package as the CPU and GPU. The entire Nvidia software stack is available, particularly CUDA, vital for AI. Nvidia's new laptops will likely be ideal for running local LLMs be cause the unified memory means you can load models up to 120-180B parameters (quantized). These laptops are expected to ship later this year and could become strong competitors to high-end MacBooks and even Mac Studios for local AI workloads, thanks to CUDA support and unified memory. Price is unannounced.

Nvidia announces the new RTX Spark, a new platform powered by the NX1 CPU, and shows off Spark laptops running 007 First Light and Forza 6. The CPU has 20 ARM based cores and a Blackwell RTX GPU with 6144 CUDA Cores. This is the same core count as a 5070, but with 128GB of unified LPDDR5X RAM memory sitting in the same package as the CPU and GPU. The entire Nvidia software stack is available, particularly CUDA, vital for AI. Nvidia's new laptops will likely be ideal for running local LLMs be cause the unified memory means you can load models up to 120-180B parameters (quantized). These laptops are expected to ship later this year and could become strong competitors to high-end MacBooks and even Mac Studios for local AI workloads, thanks to CUDA support and unified memory. Price is unannounced.

Grummz

30,481 views • 7 days ago

If you’re exploring CES this week, check out NVIDIA Showcase at Fontainebleau and enter our Explore to Win sweepstakes for a chance to win prizes including playing cards and a signed GPU. 🎉 Learn more 👉 A special thanks to Shin Lim for bringing our NVIDIA playing cards to life. ✨

If you’re exploring CES this week, check out NVIDIA Showcase at Fontainebleau and enter our Explore to Win sweepstakes for a chance to win prizes including playing cards and a signed GPU. 🎉 Learn more 👉 A special thanks to Shin Lim for bringing our NVIDIA playing cards to life. ✨

NVIDIA

32,091 views • 5 months ago

Spliting and merging number cards can be helpful in visualizing regrouping when subtracting. Learn more in this Polypad Pointer video:

Spliting and merging number cards can be helpful in visualizing regrouping when subtracting. Learn more in this Polypad Pointer video:

Mathigon

18,215 views • 2 years ago

Visualizing US electric power transmission lines. Learn how:

Visualizing US electric power transmission lines. Learn how:

Kyle Walker

44,835 views • 4 months ago