CUDA Hardware

Overview of each generation of CUDA hardware from Tesla through Ampere

[VoidloniXaarii]

Such amazing insights and historic commentary! Thank you very much!

[manojkumar-yied]

Thanks you for the detailed explanation.

[CarlCallmer]

Thank you for this, very informative!

[Philip]

Nice any update for newer generations?

[JRis]

Dang thanks for the education. Still have so much to learn and gain understanding on. This was a great start.

Id like to know if there is a way i can re-purpose my 980m GPUs that they stopped pushing driver updates on a lil bit ago.

[ihgnmah]

Thank you Professor for a wonderful lecture.

[STRAGGLER]

I really enjoyed the clarity and simplicity of your description of the function of a c u d a core. I would however like you to explain tensor cores more clearly and explain the neural net function of them

[lohphat]

With so many cores, at what point is the communiocation and scheduling overhead consume too many resources?

How complex can the function __GLOBAL__ module be before it doesn't sit in the cache effectively? Are there CUDA profilers which help guide the developer how to construct an efficient CUDA function so that it maximizes the arch and keeps overhead low so that the GPU is doing mostly heavy lifting vs. bookkeeping?

[shawnshaw]

DRAM on GPU is for calculation, also for video pixel memory? How do they co-exist

[maloukemallouke]

thanks for video,
unfortunately it's too expensive the Cuda Graphic cards for AI