3 ms·
Mainly I think it's that you need a LOT of VRAM to handle long context - server-class hardware is pretty much a requirement to work with more than ~10,000 token
by simonw 1y ago
Mainly I think it's that you need a LOT of VRAM to handle long context - server-class hardware is pretty much a requirement to work with more than ~10,000 tokens.
- ranger_danger 1y agoOn my i9 desktop with 128GB RAM and only 8GB VRAM, using llama.cpp I can split the work between both CPU/GPU and get the max 200k context to run on Qwen3 at a decent (human-reading) speed.