Older GPUs abandoned by new inference engines: are they still worth keeping?
pipecanon · reddit · 2026-11-17
A Reddit discussion on newer inference engines raising minimum hardware requirements and dropping older GPUs: realistic options include running lighter backends like llama.cpp that still support old architectures, using small quantized models to fit limited VRAM, and demoting old cards to low-priority background batch workers instead of production traffic. Managing mixed old/new GPU fleets across machines requires per-node backend configuration; platforms like gpustack centralize this, though no management layer bypasses hardware limits. The post asks the community where the practical cutoff line lies between maintaining old cards and buying new hardware.
More from Infra
- Engineer runs Gemma 3 270M entirely inside an office laser printer at 60s per token — osanseviero · 2026-11-17
- Blogger argues total compute controls create blind spots and accelerate adversaries — TinfoilTricorn · 2026-11-17
- How rollout staleness hurts on-policy distillation, and how to build a faster pipeline — burkov · 2026-11-17
- SK hynix's Nvidia revenue doubled to $24B, but US customer concentration is the real risk — tengyanAI · 2026-11-17
- macOS TCP Stack Breaks After Exactly 49.7 Days of Uptime Due to 32-bit Timestamp Overflow — DanielLockyer · 2026-11-17
- Running a 2M-param embedding model on an ESP32 with no_std Rust at 830 embeddings/s — qdrant_engine · 2026-11-17