RTX 3090 Ti vs RTX 3060 Ultimate Showdown for Stable Diffusion, ML, AI & Video Rendering Performance

Summary And Conclusions PDF ⤵️

Playlist of StableDiffusion Tutorials, Automatic1111 and Google Colab Guides, DreamBooth, Textual Inversion / Embedding, LoRA, AI Upscaling, Pix2Pix, Img2Img ⤵️

Technology & Science: News, Tips, Tutorials, Tricks, Best Applications, Guides, Reviews ⤵️

The GitHub gist file shown in the video ⤵️

Whisper tutorial ⤵️

How to install Python and Automatic1111 Web UI tutorial ⤵️

Whisper github ⤵️

Davinci Resolve tutorial ⤵️

Best DreamBooth training settings ⤵️

How to install Torch 2 for Stable Diffusion Automatic 1111 Web UI ⤵️

0:00 Box opening of Gainward #RTX3090 Ti and Cougar #GEX1050
0:51 Installation of RTX 3090 GPU and Cougar GEX1050 PSU into the computer case
5:03 Final view of the installed case
5:23 CPU, Ram and other hardware overview of the used PC
6:34 Used gist file explanation in this video
7:04 How to install latest Nvidia GeForce driver
7:32 What is the difference between Nvidia Game Ready drivers and Studio drivers?
8:43 OpenAI Whisper speech to text transcription benchmarks
9:52 How to verify installed and used PyTorch, CUDA and cuDNN versions via my custom script
10:30 How to update Whisper to latest version
10:53 Testing command used for Whisper
11:20 Demo of Whisper transcription benchmarks
12:32 How to install Torch version 2 on main Python installation
13:13 How to install cuNDD latest DLL files
14:24 Benchmark results of all Whisper tests
17:00 When RTX 3090 and #RTX3060 transcribing speech at the same time
18:01 4K Video rendering tests in Davinci Resolve
19:10 How to change rendering GPU in Davinci Resolve
19:35 Rendering results of Davinci Resolve benchmarks
20:22 Bug in Davinci Resolve, RTX 3060 is not used
23:00 Where to download FFmpeg with hardware acceleration - CUDA and GPU support
24:00 How to set default FFmpeg via environment variables path
25:27 Testing setup of the FFmpeg 8k video rendering
27:19 Demo demonstration of FFmpeg benchmark
27:58 Final results of FFmpeg benchmarks on both RTX 3060 and RTX 3090
29:45 Starting to benchmark Stable Diffusion via Automatic1111 Web UI
30:06 How to see used Torch, CUDA and cuDNN DLL version of your Web UI
30:38 How to update Web UI xFormers version
31:55 it/s iteration per second testing
32:20 Demo of testing methodologies that will be used for Stable Diffusion benchmarks
36:57 Starting result analysis of Stable Diffusion benchmarks Torch 1.13
42:48 Used DreamBooth training settings for benchmarking
44:00 Stable Diffusion benchmarks with Torch 2.0
46:48 How to make sure that Web UI uses second device in all cases
48:26 opt-sdp-attention benchmark results with Stable Diffusion
51:19 The discovery I made about optimizers used in Stable Diffusion Web UI
53:32 Solution for Stable Diffusion NansException : A Tensor with all NaNs in Unet

The world of artificial intelligence and machine learning is rapidly growing, and as it expands, the demand for powerful and efficient hardware is skyrocketing. Among the most critical components of this hardware are graphics cards, which play a pivotal role in the performance and capability of machine learning applications. The Nvidia RTX 3090 and RTX 3060 are two notable examples of this new generation of graphics cards, designed with machine learning in mind. This article will explore the features of these two cards, and discuss the importance of graphics cards in the field of machine learning.

The Nvidia RTX 3090 and RTX 3060

Nvidia's GeForce RTX 3090 and RTX 3060 are part of the company's Ampere architecture, which aims to provide a significant leap in performance and efficiency compared to previous generations. The RTX 3090, known as the "BFGPU" (Big Ferocious GPU), is the flagship model, boasting 24 GB of GDDR6X memory, 10,496 CUDA cores, and a memory bandwidth of 936 GB/s. This card delivers unparalleled performance, making it ideal for high-end machine learning applications, rendering, and gaming.

The RTX 3060, on the other hand, is a more budget-friendly option, but still packs a punch in terms of performance. With 12 GB of GDDR6 memory, 3,584 CUDA cores, and a memory bandwidth of 360 GB/s, the RTX 3060 provides excellent value for money, while still offering enough power to handle many machine learning tasks.

[SECourses]

Please join discord, mention me and ask me any questions. Thank you for like, subscribe, share and Patreon support. I am open to private consulting with Patreon subscription.

[outcast]

You're a legend to have figured out some of these out of memory error work arounds, they drive me just crazy. I don't care about extra speed compared to just getting it working reliably.

[valorantacemiyimben]

Stable Diffusi da yüz değiştirmeyi kaliteli şekilde yapabilmek için rtx 3060 12gb iş görür mü hocam?

[lefourbe]

You deserve this 3090-ti !
I'm glad to be able to be a patreon. We Can see where the money goes and it benefit us all.

[pkflyfishing]

Nice video. Now I have a question, when using CPU core i5 13600K and GPU rtx 3060 12GB and RAM DDR5 32 or 64GB, will my timeline play smoothly or jerky in DaVinci Resolve Studio? Videos will be from FHD to 4K 60fps from the DJI Osmo Action 4 camera

[halilibrahimakkaya]

It is amazing to see someone interesting to SD deeply, thank you for great knowledge share.

[matthewbennett]

This video holds a ton of great knowledge! Currently messing around with torch 2.0, and I will rewatch this many times. 🎉

[frzenisshadowbanned]

That's a lot of Watts.
Most Cards respond very well to a power limit of up to -25% whilst still archiving 92.5-95% of the computing power.
Try MSI Afterburner. It's literally just moving a Slider :P

Thanks for all the Data :)

[ifritirius]

what you think about rtx 4060ti 16 gb for sd ? becaus i m worry about bus 128 bit may will be not handle if sd use more 12gb vram . so is bus matter for sd?

[Raketenclub]

this video is outstanding in terms of evrthng.... so much information and input. subscribed and like .... this is indeed an awesome howtodo benchmarks. .... i still use sd on cpu... old dual xeon.. 12 cores + ht ... 24 fake cores... deb 9... takes houuuurs.... cant wait for the rtx 3090 24g... ill do benchmarks too. lol.

[greendsnow]

Hello! Does CPU cache affect the SD performance, given all other components remain the same?
I'm going to buy a 5600x3d which has a big volume of cache, but a 13600K is at the same price with more cores..
I'm mainly going to use the PC for AI things

[MehmeterTem]

Su watt ilcer bendede var ama calismiyor. Nesi vardir acaba.

[nakkeomsu]

abi bunu kullanabilmek icin pc de stabble diffusionun yüklü olması mı gerekiyor?

[MathewRenfro]

This is the video I have been waiting for

[vhsdude]

If the Whisper Benchmarks includes the loading of the models etc then you really need a longer audio file.

[muhammadiqbal-onze]

What's the advantage using stable diffusion on pc compare using Google collab?

[zapznox]

i try to update the xformers but whenever i open the .bat webui it says a lower version is installed

[mtnmecca_ej]

Great information! I mostly curious how to improve the quality of training and image generation to match the trained subject. Does the 3090 and bf16 plus other settings allow for improved training quality? I have another computer with rtx 3090 16GB ram but have not tried any SD tasks yet since I want to learn as much as possible before I decide to upgrade and how much quality can be gained. I am currently running SD on RTX 3060 12GB ram.

[fahd]

SECourses doing gods work over here keep up the good work champ

[flusterzero]

Im surprised this doesnt have more views