GPUs & CPUs & Enthusiast hardware: Questions, Discussion and fanboy slap-fights - Nvidia & AMD & Intel - Separe but Equal. Intel rides in the back of the bus.

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
Cross-posting for my homie @The Ugly One and anyone else interested

AMD Intros “SD 3 Medium”, The World’s First BF16 NPU Model Designed For XDNA 2 AI NPUs, Offer Reduced Memory Footprint & Faster Text-To-Image Gen (archive)
https://www.amd.com/en/blogs/2025/worlds-first-bf16-sd3-medium-npu-model.html (archive)

 
Cross-posting for my homie @The Ugly One and anyone else interested

AMD Intros “SD 3 Medium”, The World’s First BF16 NPU Model Designed For XDNA 2 AI NPUs, Offer Reduced Memory Footprint & Faster Text-To-Image Gen (archive)
https://www.amd.com/en/blogs/2025/worlds-first-bf16-sd3-medium-npu-model.html (archive)

How will this affect the competitive masturbation scene
 
  • Thunk-Provoking
Reactions: StacticShock
How will this affect the competitive masturbation scene
You are going to be lying down with your thin-and-light laptop with the all-new bargain bin laden quad-core Ryzen AI 5 330 with XDNA 2.0 on top of you, and pumping your cock twice as fast while using the new BF16 SD3M in "Amuse" to generate images. Or maybe not if you need more specialized software than that.
 
A real addict would be using the 5090 and doing text-to-video-to-gooncave.
A true addict would have an RTX 6000 workstation GPU. You can get one for about $7000 which is not bad considering a lot of 5090s are hovering around $3000.

The budget-minded goonlord, the temporarily embarrassed goonionaire, can get by with a couple of 5060 Ti 16GBs though. Not as fast but you can gen text on one and images on another for an ascended interactive smut experience.
 
You are going to be lying down with your thin-and-light laptop with the all-new bargain bin laden quad-core Ryzen AI 5 330 with XDNA 2.0 on top of you, and pumping your cock twice as fast while using the new BF16 SD3M in "Amuse" to generate images. Or maybe not if you need more specialized software than that.
Is the 330 even usable for ai
 
Is the 330 even usable for ai
It's a fast Zen 5 quad-core (all modern CPUs are fast), with a shit 2 CUs RDNA3.5 iGPU (hasn't been tested yet, it was just announced), but the full enchilada XDNA 2.0 at 50 TOPS, which is what the newly optimized version of the Stable Diffusion 3 Medium model is intended to use according to AMD's announcement. It should work just fine for generating images.
 
It's a fast Zen 5 quad-core (all modern CPUs are fast), with a shit 2 CUs RDNA3.5 iGPU (hasn't been tested yet, it was just announced), but the full enchilada XDNA 2.0 at 50 TOPS, which is what the newly optimized version of the Stable Diffusion 3 Medium model is intended to use according to AMD's announcement. It should work just fine for generating images.
I see the profitable opportunity in making a Ryzen Ai 320 with 2 cores and 4 threads to rob retards
 
I see the profitable opportunity in making a Ryzen Ai 320 with 2 cores and 4 threads to rob retards
I think that's actually plausible if the rumor is true that it uses a smaller "Krackan Point 2" die that only has 4 cores max. But that's unconfirmed for now.

If it's regular Krackan Point, then they are disabling it down from 8 cores so a dual-core would be unlikely. But there is precedent for AMD doing this: the dual-core Ryzen 3 5125C is a Chromebook chip based on the 8-core Cezanne die. Below the bottom of the barrel.

Ryzen AI 5 330 is probably not to be considered a direct replacement for Mendocino since it uses the FP8 socket. I predict pricing for 330 laptops could start at $500 and fall to around $300. I'd buy that for a dollar!
 
I think that's actually plausible if the rumor is true that it uses a smaller "Krackan Point 2" die that only has 4 cores max. But that's unconfirmed for now.

If it's regular Krackan Point, then they are disabling it down from 8 cores so a dual-core would be unlikely. But there is precedent for AMD doing this: the dual-core Ryzen 3 5125C is a Chromebook chip based on the 8-core Cezanne die. Below the bottom of the barrel.

Ryzen AI 5 330 is probably not to be considered a direct replacement for Mendocino since it uses the FP8 socket. I predict pricing for 330 laptops could start at $500 and fall to around $300. I'd buy that for a dollar!
That's actually pretty cheap for what you're getting
 
How will this affect the competitive masturbation scene
It won't. Look, I'm not gonna hide my main motivator behind getting a GPU with 24GB of VRAM, so here's the quick rundown.

For a base model to be usable for NSFW, it first needs to be on an open license. The last Stability AI model that's on such a license was SDXL, as with SD3 they've introduced a restricrive license so no one trains merges on it. For Flux.1, only the Schnell model is on an open license, and right now the only NSFW merge being trained on that is Chroma.

However most sloppers are still using SDXL merges. One, they're lighter, two, they're better. The previous meta was Pony Diffusion, but that had a major issue where you had to use LoRA's for everything as the base model only had one shitty artstyle and couldn't handle characters too well. The current meta however are Noob/Illustrious based models, and those handle just about every artstyle and all you need to do is prompt it e621/Danbooru tags to get what you want. No LoRA'd needed for the most part. It's also insanely good at hands and multi character scenes with just direct prompting. In comparison, Chroma doesn't handle artstyles at all, Lodestones doesn't plan on adding them to the dataset, and Chroma's dataset is still smaller than that of Noob's. Chroma also doesn't handle e6/Dan tags as well.

In short, it won't, because SD3 and Stability AI sucks dick.
 
Back