Pairing a decade-old RX 480 with RX 7900 GRE boosts llama.cpp MoE inference 36% over single GPU
tabletuser_blogspot · reddit · 2026-09-08
A Reddit user benchmarked dual-GPU llama.cpp inference (Vulkan build 10453) with an RX 7900 GRE 16GB plus an old RX 480 8GB.
- Gemma4 26B-A4B Q4KM: 52.03 t/s generation on two GPUs vs 37.93 t/s on the 7900 GRE alone (36% faster); prefill rose from 238 to 321.6 t/s.
- Dense 27B models crawl on this setup: Qwen35 27B Q6K at 5.96 t/s, Q5KM at 11.73 t/s.
- Flash Attention made no measurable difference for the MoE model (52.03 vs 51.94 t/s).
- Even the low-bandwidth RX 480 (256 GB/s, 1.8 t/s alone) meaningfully helps MoE inference when pooled.
More from Infra
- HBM demand shifts: low-stack inference plans shelved, Rubin to go 12-high only — BenBajarin · 2026-10-26
- CircuitHub unveils The Grid, a factory-scale robot making PCBs in the US twice as fast as Shenzhen — garrytan · 2026-10-26
- Ubicloud publishes "Fixing Tool-Call Accuracy in vLLM" — petrusenko_max · 2026-10-26
- Meta approves nonvoting Class C shares as 2026 capex guidance hits $135B–$145B — thione · 2026-10-26
- Unsloth Ships MiniMax-M3.1-Flash Quants: 2-bit Retains 91.5% Accuracy, Runs in 150GB RAM — danielhanchen · 2026-10-26
- Engineer with thousands of fabricated devices pours cold water on photonic crystal modulator yields — jwt0625 · 2026-10-26