NVIDIA SANA is Now Available on Windows! Step-by-Step Installation
Added 2025-02-20 15:00:11 +0000 UTCSANA IS NOW AVAILABE FOR WINDOWS!! – easily one of the most underrated applications of the year, the text-to-image tool that brings creativity to everyone, no matter your hardware! With versions tailored to different GPUs, Sana delivers stunning visuals up to 4096×4096 resolution, faster and more efficiently than ever.
Your Choice, Your Power:
🌟 Sana-1.6M: Perfect for GPUs with 12GB of VRAM, delivering high-quality images quickly and smoothly.
🌟 Sana-6M: Designed for GPUs with less than 9GB of VRAM, making powerful image generation accessible to even more creators.
🌟 Both Multilingual: Generate images from prompts in multiple languages—breaking barriers for creators around the world.
Why Sana Stands Out:
🖼️ Crystal-Clear Results: High-resolution, high-quality images with excellent text alignment.
💻 Accessible for All Hardware: Whether you’re on a powerful setup or a modest laptop, there’s a Sana version for you.
⚡ Fast and Lightweight: Sana creates 1024×1024 images in under a second on laptops, all while being smaller and faster than traditional models.
🌍 Affordable and Inclusive: No need for expensive GPUs—just pick the version that suits your setup.
Github repo: https://github.com/gjnave/Sana-for-Windows
A Get-Going-Fast Installer can be found here: https://www.patreon.com/posts/nvidia-sana-for-122669029
********************************
Membership & Intstallers: https://www.getgoingfast.pro
Website: https://www.cognibuild.ai
Discord: https://discord.gg/jRYeUbBt6P
LinkedIn: https://www.linkedin.com/in/gjnave
Reddit: https://www.reddit.com/user/FitContribution2946/
Personal Music: www.daysinging.com
Instagram: https://www.instagram.com/cognibuild/
Spotify: https://open.spotify.com/artist/0oH5qisu13DpnT7DucnV9d
* Helpful Video Tutorials *
VS Build Tools 2019 - Cleanest / Easiest Install Ever: https://youtu.be/TUS4A-XsAPo
Installing Cuda: https://youtu.be/_GpyoEbs3Hs
Installing CuDNN: https://youtu.be/RY2mEbi7PJc
Fresh AI Setup (Windows): https://youtu.be/4aCtOVcvcj4
Installing Miniconda: https://youtu.be/ULAVtKBOKXE
Installing Python: https://youtu.be/1CdeBbk9A2k
Insalling Git: https://youtu.be/xxVOm3ndeSU
Installing WSL: https://youtu.be/3GNqg3GdEKA
Installing Llama.cpp: https://youtu.be/r-05yuXTEPE
Comments
C:\Windows\System32>nvcc --version nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2025 NVIDIA Corporation Built on Wed_Jan_15_19:38:46_Pacific_Standard_Time_2025 Cuda compilation tools, release 12.8, V12.8.61 Build cuda_12.8.r12.8/compiler.35404655_0
Wei Choong Lee
2025-02-22 03:06:41 +0000 UTCokay, this is an issue of you have CUDA ( different than cudatoolkit) which is lower than 12.6. go your windows aterminal and type: nvcc --version and tell me whate you see
Gary
2025-02-21 18:59:43 +0000 UTCWhen I executed this line: conda install nvidia/label/cuda-12.6.0::cuda-toolkit, I got this error:Channels: - defaults - nvidia/label/cuda-12.6.0 Platform: win-64 Collecting package metadata (repodata.json): done Solving environment: failed LibMambaUnsatisfiableError: Encountered problems while solving: - nothing provides cuda-version >=12.6,<12.7.0a0 needed by cuda-nvml-dev-12.6.37-2 Could not solve for environment specs The following package could not be installed └─ cuda-toolkit is not installable because it requires └─ cuda-nvml-dev 12.6.37.* , which requires └─ cuda-version >=12.6,<12.7.0a0 , which does not exist (perhaps a missing channel).
Wei Choong Lee
2025-02-20 16:06:46 +0000 UTC
