Siemens EDA boosts chip design agent score from 55% to 95% with NVIDIA NeMo Gym
NVIDIAAI · x · 2026-10-31
NVIDIA published a tech blog detailing how Siemens EDA used NeMo Gym to improve its Solido Fuse chip design agent: a domain-specific verifier grades final Liberty artifacts (coverage, validity, accuracy, integrity, zero reward on failure), and diagnostics from repeated rollouts drive refinements to the agent's skills and MCP tool descriptions. On a standard-cell characterization benchmark, the composite verification score rose from 55% to 95%, agent turns dropped from 155 to 53 (66% fewer), and output tokens fell from 81K to 29K (64% fewer). The same evaluate-and-optimize approach applies to engineering agents in EDA, CAE and other domains.
More from coding & agent
- LlamaIndex's Agentic Plus adds max effort mode, lifting extraction F1 from 77.7 to 92.9 — llama_index · 2026-10-31
- Set it on Sunday: a daily prompt that triages Gmail and briefs you via 45-second AI avatar video — toolstelegraph · 2026-10-31
- 2 million lines of Rust in 2 weeks: FrankenSim adds full physics, agent-built suite matures — doodlestein · 2026-10-31
- Next.js prototyping native Go support for Route Handlers and Proxy — rseroter · 2026-10-31
- Tech paradigm shifts reprice mundane infra: 1Password could become the identity layer for agents — signulll · 2026-10-31
- GitHub's Agent Bake Off compares coding models and harnesses head to head — 0xkarasy · 2026-10-31