- This topic has 5 replies, 2 voices, and was last updated 1 month, 1 week ago by
.
-
Posts
-
-
Hello
Since 24gb VRAM is the recommended RAM size to have enough room to process workflows in Stable Diffusion. We also have various series of NVidia GeForce RTX cards 3000, 4000 and 5000 (32gb (2025)), these have different architectures, Ampere, Ada Lovelace & Blackwell 2.0. That have 24gb + VRAM on these series of cards, so what exactly is the architecture doing within these different series when it comes to Stable Diffusion.
We know when it comes to games the cards are faster each series, render’s more detail (this area is a bit ambiguous) and produces a higher FPS for game play for various games. Can you apply the same logic to AI & Stable Diffusion, given that you have an image to decode and output to a pixelated image. Everything is the same for this very high resolution image, at the highest, crispest and highly detailed image contained in latent space. All that needs to be done is to generate the image at the GPU to produce the highest possible quality on each of these cards.
NVidia GeForce RTX cards 3090: This card will produce the highest detail and resolution this GPU can produce, at a rendering time suitable for this GPU. With Ampere architecture
NVidia GeForce RTX cards 4090: OK, we know that this card will out perform the 3090 speed wise, but will the Ada Lovelace architecture add any extra detail to the final image. Whereby increasing it’s resolution, crispness, smoothness of a high resolution image.
NVidia GeForce RTX cards 5090: OK this card has not been release yet, but can the Blackwell 2.0 architecture spec tell us anything? Will this card perform in much of the same way as the 4090 over the 3090. We can assume that this card will be faster, we know it has extra bus bandwidth and an extra 8gb VRAM over it’s predecessor totalling 32gb VRAM (reportedly). All good for speed, but resolution and that finer detail we all love in out generated images, will there be an improvement.
We know that the NVidia GeForce RTX 3090, is a really good card and that’s the reason why it’s held it’s value over the 4090. But apart from speed wise, does the 4090 tower over the 3090 when it comes to detail in the final outputted image. Can I apply what I know about game cards to AI & Stable Diffusion, if I apply what I know about the graphic engine, then I’ll get my answer (maybe). We know that the GPU is used differently in gaming than in the application use in AI-DL-ML. When we get to the graphics engine (used in gaming), does Stable Diffusion use this engine in exactly the same way as used in gaming. As in renders 3D environments, textures, lighting effects, shadows, and other visual elements to create the final image displayed on the screen.
Logically I know the answer, but my limited knowledge of how AI-DL-ML works is creating an issue, knowing that gaming and AI are used in completely different ways doesn’t help. However the use of the graphics engine maybe the same and the conclusion maybe based on that. I just don’t know for sure as an absolute, so maybe those that have been using Stable Diffusion for a while. Maybe able to shed some light between application of use between the RTX 3090 vs 4090 image quality wise, forgetting the speed of image generation. Am battling with Ampere vs Ada Lovelace at an GPU, I know that with the release of RTX 5000 series the RTX 3090 will come down in price to meet my budget possibly.
So here’s where I’m at:
NVidia GeForce RTXÂ 3090 24gb vs NVidia GeForce RTX 4080 Super Ti 16gb
Both cards will be in my budget range when the RTX 5000 series is released in 2025, with the RTX 3090 dropping to the RTX 4080 Super Ti price value (I’m predicting).
Is it really about the VRAM???
Which card is it and what are the downsides of each, if its time with no image loss I have time, but as we know it’s not as simple as that.
I’m just trying to work out which one I should go for…
Kind Regards
Zandebar
-
The newer architectures have some new optimization techniques and can be faster in training and using models.
SD models use GPU differently from gaming applications. A GPU card’s FLOPS number (floating number operations per second) is a good gauge for performance.
4090 is for sure faster than 3090 but they should generate the same image with the same setting. The only difference is how long you wait.
I would only consider 24GB+ VRAM if I buy a GPU card now. Consider it an investment to future-proof your system. A slower card means you need to wait longer. A low-VRAM card means you cannot run certain models at all. (Or you need to jump through hoops to do it)
But if you are happy with the toolset now – SD 1.5, SDXL, Flux, getting a 16GB card is not a bad idea to save some money.
-
Hello
Great and Thank You! You kind of confirmed what I was thinking.
Right: I have a bottle neck, I’m based in the UK I only have a £1000 GBP to spend on a GPU, I’m a hobbyist and will not be making any money from this to justify the expense and the outlay. However I’m also not sure where I’ll be going with this so I’m looking for a hybrid solution to GPU needs.
Let’s get this straight; in 12 months time when you get the RTX 5090 with 32GB VRAM, you’ll be saying Wow at the speed and recommending 32gb VRAM and not 24gb, when asked the very same question.
Granted if your a pro then you’ll need the FLAGSHIP option. When your not a pro (like me) justifying the expense becomes hard when your on a tight budget and you have household bills to pay, I can only dream of owning the latest and greatest GPU. There is a compromise a cheaper option or rent a GPU from a render farm, I’m actually looking at both at the moment.
NVidia GeForce RTX 4080 Super Ti 16gb VRAMÂ (I can afford right now), I’ll be able to learn SD and do a fair bit with 16GB VRAM. When I hit that wall and need extra VRAM I’ll out source the GPU to a render farm, with the render farm option I’ll just pay for what I use. This isn’t a good place to be really with the new RTX5000 series coming out, where you were only 2 thirds of the max VRAM the the 5000 series comes out your half the max size. Where the model size will only get bigger, I was bouncing around and saw a model size (flux) 14GB. Ouch! not much room for everything else that gets stored in VRAM. Chance are this size model would work in 16GB VRAM, and its only going to get bigger. We know that because of the increase of VRAM in the 5000 series, if you make more space people will fill more space. You can’t win being a hobbyist.
I was also thinking, wait long enough the 3090 may fit in my budget:
Do you get this craziness where you are?
EVGA GeForce RTX 3090 Ti FTW3 ULTRA GAMING, 24G-P5-4985-KR, 24GB GDDR6X, iCX3, ARGB LED, Backplate, Free eLeash
£2,094.22
MSI GeForce RTX 4090 VENTUS 3X E 24G OC Gaming Graphics Card – 24GB GDDR6X, 2550 MHz, PCI Express Gen 4, 384-bit, 2x DP v 1.4a, HDMI 2.1a (Supports 4K & 8K HDR)
£1,749.99
GIGABYTE GeForce RTX 4090 GAMING OC 24GB Graphics Card – 24GB GDDR6X, PCI-E 4.0, Core 2535Mhz, RGB fusion, Anti-sag bracket, Metal back plate, DP 1.4, HDMI 2.1a, NVIDIA DLSS 3, GV-N4090GAMING OC-24GD
£1,899.00
Where the 4090 is cheaper than the 3090 CRAZY! OK the 3090 is not a toaster like the 4090 with the power socket issue. But still you would of thought there’ll be some rest bite for us hobbyist with an older series of card, Nah!! So where stuck at the next generation down, the 4080 ti super.
And wait for it, Nvidia are not doing themselves any favours with the next generation of cards now that they have no competition. Look at this…
RTX 5080
TDP: 350W
GPU Name: GB203
GPCs: 7
TPCs: 42
SMs: 84
Cores: 10752Tensor Cores: (likely) 384 (half the number of RTX 5090)
Memory Configuration: 256-bit GDDR7 (16GB VRAM)Boost clock speed around 2.8 GHz
RTX 4080
Architecture: Ada Lovelace
Process node: 4nm TSMC
CUDA cores: 9,728
Ray tracing cores: 76
Tensor cores: 304
Base clock speed: 2,205 MHz
Maximum clock speed: 2,505 MHz
Memory size: 16GB GDDR6X
Memory speed: 21 Gbps
Bus width: 256-bit
Bandwidth: 912 GBps
TBP: 320W4080 Difference: +9.3% with the 5090
NVidia have got there head some where I can’t say here, but logically with the uplift in performance of the 5090, you would have thought a shift in the other models.
GeForce RTX 5000 to resemble something like this in vram: 12gb (5060), 16gb (5070), 24gb (5080) and 32gb (5090)
And the CUDA core count is not much higher, would have thought they’ll match the 4090 cores with the 5080. Core count @10752 you would have thought they’ll match at @16384 CUDA Cores. Given that’s its rumoured that the 5090 is having 21,760 CUDA cores. And the Tensor cores have dropped, maybe a good reason there but out of my scope.
Logically that makes more sense, it just leaves us users of the products’ frustrated, plus if the 5080 with imaginary 24gb VRAM and 16384 CUDA Cores. This would almost match the 4090 and cause a price drop of remining units of 4090. Everyone wins, but NO…
That’s why am waiting to see what the market does and see if these rumoured specs are true, and make a decision then. Either way the consumer is going to be at a dis-advantage give Nvidia previous history.
In the meantime: Checking out GPU farms and what they can offer is looking like a good idea and could in principle be more beneficial. That’s out of scope for this thread, I’ll make one on GPU farms…
Kind Regards
Zandebar
-
-
-
It’s kind of where I’m at, with the move to cloud computing and Nvidia moving away from desktop products due to miniaturisation (mini pc’s, tablets, laptops). I feel that am going to get screwed, as what I can afford to buy a GPU at with the budget I have, will always be entry level. I’ll be always playing catchup, I don’t know yet if the 90% of the rubbish / trash I’ll produce will matter with a cloud GPU. That’s what I’m trying to work out, plus apart from generative ai, I have no barrier with my current old GPU (rtx2070). I know that I can’t afford a GPU that has 24gb VRAM locally for my present budget (they might drop in price), to get the benefit for all the modules / workflows. With that, VRAM going to always go up in size as modules / workflows get bigger, its better to work smart and know your limitations. Seeking out that better financial option for that creative data file, where craving for, plus am looking to go down that rabbit hole of video.
I’m not sure where this journey is going to take me and there’s going to be surprizes along the way, who knows where I’ll end up.
Its just about getting that experience in generative ai and being able to make smart discissions along the way.
-
- You must be logged in to reply to this topic.