What makes system calls expensive: pipeline flushes and predictor resets, explained
abhi9u · x · 2026-11-04
Abhinav Upadhyay breaks down why system calls are far costlier than they look on Linux x86-64: the direct cost is the small bit of kernel code per call, but the bigger hit comes from microarchitectural side effects — every user/kernel transition makes the CPU drop its optimizations, flush pipelines, and reset predictor state, then rebuild them on return. That's why iouring and eBPF exist to minimize kernel crossings. A deep walkthrough of the kernel entry/exit path and its performance profile footprint.
More from Infra
- SpecAgent: multi-session Claude Code pipeline with measured 3x context cost savings — AdOtherwise9458 · 2026-11-05
- SoftBank cloud arm IDC Frontier hit by ransomware in 'seven minutes,' taking ~500 organizations offline in Japan — ChuckDBrooks · 2026-11-05
- A $2.5k local rig runs Qwen3.8 27B at 150 tokens/s, benchmarks show — casper_hansen_ · 2026-11-05
- A $2.5k RTX 3090 box runs 328GB GLM-5.3-Flash locally via expert offloading — casper_hansen_ · 2026-11-05
- Arm earnings: data center royalties double as selling own chips pits it against licensees — tengyanAI · 2026-11-05
- Open-source Local Office adds Snapdragon NPU support for local AI on Windows ARM64 — HowDevelop · 2026-11-05