Hardware-aware Ollama fork with legacy NVIDIA support, controllable runtime tuning, Q4/Q8 KV cache, Flash Attention, CUDA Graphs and optimized multi-GPU inference.
-
Updated
Nov 15, 2026 - Python
Hardware-aware Ollama fork with legacy NVIDIA support, controllable runtime tuning, Q4/Q8 KV cache, Flash Attention, CUDA Graphs and optimized multi-GPU inference.
AMD Polaris RX 580 / gfx803 Windows AI toolkit: ZLUDA, HIP/ROCm, rocBLAS, DirectML, optional ComfyUI. / AI-стек для RX 580 на Windows.
Local LLM desktop for Tesla P40 Pascal, CUDA MMQ and Ivy Bridge AVX1. / Локальные LLM на Tesla P40 и AVX1, Swift 256K и LAN-медиа.
ultra-fast vulkan renderer for minecraft 1.21.11 with support for older GPUs
KoboldCpp for Intel Haswell AVX2 and AMD RX 580 Polaris Vulkan. / Портативный KoboldCpp для Haswell, RX 580 и локальных GGUF-моделей.
llama.cpp CUDA DP4A/MMQ tuning for NVIDIA Pascal (SM 6.1), Tesla P40 and GTX 10. / Патчи llama.cpp для Pascal и Tesla P40.
Fix NVIDIA 390.157 DKMS build failures and Xorg crash on kernel 6.8 / Ubuntu 24.04 / Linux Mint 22
Ollama with PARAMETER think support + CUDA 3.5/3.7 (Tesla K80) + ROCm gfx803 (RX 580) legacy GPU support
GDM + GNOME configuration patches for AMD/ATI ES1000 (R100) GPUs — making the login screen usable on legacy hardware
Run selected PyTorch CUDA 13 operations on older NVIDIA GPUs, starting with Maxwell sm_52. Independent and evidence-first; not NVIDIA's CUDA Compatibility package.
Unofficial low-VRAM fork of wiltodelta/remove-ai-watermarks for legacy NVIDIA GPUs with 4 GB VRAM, face-free processing, and OCR-guided text restoration.
CUDA 8 Tiny YOLOv2 on fnOS Docker with NVIDIA Fermi GT 520M and driver 390.157. Reproducible scripts and benchmark evidence.
FeraVolt · Local LLM inference, GGUF and generative media on older hardware. / Локальный ИИ на доступном железе.
To associate your repository with the legacy-gpu topic, visit your repo's landing page and select "manage topics."