AgentSkillsCN

hardware-software-codesign

适用于量化研究、系统实施及生产管控的软硬件协同设计工作流。当任务涉及生产交易系统中的硬件与软件相关工作流时,可选用此类工作流。

SKILL.md
--- frontmatter
name: hardware-software-codesign
description: "Hardware Software Codesign workflows for quantitative research, implementation, and production controls. use when tasks involve hardware and software workflows in production trading systems."

Hardware Software Codesign

objective

Execute hardware software codesign work with reproducible research, explicit controls, and deployable outputs.

workflow

  1. define end-to-end latency budget and deterministic performance targets.
  2. instrument each stage from feed ingress to order egress.
  3. optimize kernel, memory, and network path for tail-latency reduction.
  4. stress packet bursts, failovers, and capacity saturation scenarios.
  5. promote only after reproducible latency and recovery behavior is verified.

required diagnostics

  • stage-level p50, p99, and p999 latency decomposition.
  • jitter and throughput stability under sustained burst load.
  • packet-loss recovery time and replay correctness.
  • resource saturation signals before service-level breach.

risk controls

  • enforce hard latency and packet-loss service objectives.
  • enforce automatic failover and load-shedding thresholds.
  • enforce runbooks for exchange-connectivity incidents.

outputs

  • run python scripts/hardware_software_codesign_diagnostics.py input.csv --output diagnostics.json and keep the json artifact.
  • write an implementation memo using references/hardware-software-codesign-playbook.md with assumptions, tests, limits, and rollout plan.

resources

  • use scripts/hardware_software_codesign_diagnostics.py for deterministic diagnostics.
  • use references/hardware-software-codesign-playbook.md for the domain-specific checklist and delivery structure.