FLUX.2 Klein 4B on a Core i5-650 from 2010, 8 GB DDR3 RAM and an RX 570 with 4 GB VRAM from 2017 (WTF :O) Model: FLUX.2 Klein 4B Q4_K_M Text encoder: Qwen3 4B Q4_K_M Backend: Vulkan Resolution: 512 × 512 Steps: 4 Generation time: 403.74 seconds Afterwards, the image was upscaled with Real-ESRGAN, also using Vulkan. Total model memory: 6199.46 MB VRAM: 2644.09 MB RAM: 3555.38 MB The Qwen text encoder uses 3555.38 MB of RAM (the poor dual core D:), while the FLUX.2 diffusion model uses 2483.65 MB of VRAM and the VAE another 160.43 MB of VRAM. And yes, it works! :D #EdgeAI #EdgeComputing #AI #LocalAI #AIOnTheEdge #Upcycling #HardwareUpcycling #SustainableComputing #EfficientComputing #FOSS #OpenSource #Linux #Vulkan #AMD #RX570 #FLUX2 #RealESRGAN #FluxKlein #ImageGeneration #GenerativeAI #AIArt #OldHardware
FLUX.2 Klein 4B on a Core i5-650 from 2010, 8 GB DDR3 RAM and an RX 570 with 4 GB VRAM from 2017 (WTF :O) Model: FLUX.2 ...
