I restarted a 10-year-old Xeon 174 times to delete 12 flags and gain 4 tps

I restarted a 10 year old Xeon 174 times to delete 12 flags and gain 4 tps

The author ablated all 25 flags from a previous Gemma 4 setup on a 2016 Xeon. Most did nothing. The real levers are flash attention, physical-core thread count, and workload-gated speculative decoding: on for code, off for long-doc summarization (54% faster without it). The auto-tune drafter setting is a crude per-request router that currently hurts long-context throughput.

Why it matters: A solid local-inference tuning experiment that ablates 25 launch flags one by one, delivering scene-dependent numbers on speculative decoding and the finding that most flags are placebo. Hits H and K, but narrow audience and zero industry conversation value mean R misses — lan...