itsahedge
2.8K posts


this is full setup on assuming you have fresh linux / WSL and nvidia-smi works: ``` sudo apt update && sudo apt install -y git build-essential cmake ninja-build nvidia-cuda-toolkit && \ git clone github.com/spiritbuun/lla… && \ cd llama-cpp-turboquant-cuda && \ git checkout feature/turboquant-kv-cache && \ mkdir -p build && cd build && \ cmake .. -DGGML_CUDA=ON -DGGML_NATIVE=ON -DGGML_CUDA_FA=ON -DGGML_CUDA_FA_ALL_QUANTS=ON -DCMAKE_BUILD_TYPE=Release -G Ninja && \ ninja -j$(nproc) && \ cd bin && \ echo "✅ Build complete on your RTX 3090! Starting TurboQuant server with Qwen3.5-27B abliterated (262k context)..." && \ TURBO_LAYER_ADAPTIVE=1 ./llama-server \ -hf mradermacher/Huihui-Qwen3.5-27B-abliterated-GGUF:Q4_K_M \ -c 262144 \ -ngl 99 \ -ctk turbo3 \ -ctv turbo3 \ -fa on \ -b 512 \ --port 8080 ```




After 5 years, parsec is shutting down. Not how we wanted our story to end, but we are proud of what we built and the value we provided along the way. We are eternally grateful to those that traversed the ups and downs onchain with us. It was quite the ride 🔭



Btw, you can use the new Wallet API's /funded-by endpoint to check if a wallet was funded by an exchange, mixer, or bridge Powered by Orb's database of 7,000+ tagged accounts and programs






We raised $500M at an $11B valuation to transform how people interact with technology.









