Loading…
NVIDIA Nemotron 3 Ultra Leads Open Models on Accuracy and Efficiency in Agentic RTL Coding
NVIDIA Developer BlogNirmal Kumar Juluru
Summary
Register transfer level (RTL) development relies heavily on iterative verification and tool feedback, which poses challenges for standard code generation models. To address realistic hardware design tasks, the ACE-RTL agent pairs an iterative generate-test-reflect workflow with NVIDIA Nemotron 3 Ultra. The model utilizes a hybrid Mamba-Attention Mixture-of-Experts architecture trained on a synthetic dataset encompassing code generation, editing, and error debugging. Across the comprehensive Verilog design problems benchmark, ACE-RTL with Nemotron 3 Ultra attained a 97.1% average pass rate, outperforming GLM 5.2 and Kimi K2.6. Additionally, Nemotron 3 Ultra achieved its accuracy while consuming an average of 6,629 tokens per iteration, which is substantially lower than competing models.
Context
Modern chip design faces bottlenecks because RTL development and verification require specialized hardware knowledge, precise temporal reasoning, and repeated interaction with EDA tools to resolve bugs exposed during tool-based validation.
Approach / What changed
NVIDIA combined the ACE-RTL agent (featuring generator, reflector, and coordinator components) with Nemotron 3 Ultra, a 550B total-parameter (55B active) hybrid Mamba-Attention MoE model trained on an RTL synthetic data generation pipeline covering spec-to-RTL, code editing, and debugging tasks.
Takeaways
- ACE-RTL with Nemotron 3 Ultra achieved a 97.1% average pass rate across nine CVDP task categories, outperforming Kimi K2.6 (95.2%) and GLM 5.2 (92.1%).
- Nemotron 3 Ultra averaged 6,629 tokens per iteration on CVDP tasks, representing roughly 28% fewer tokens than GLM 5.2 (9,156) and 71% fewer than Kimi K2.6 (22,579).
- Nemotron 3 Ultra uses a 550B total, 55B active parameter hybrid Mamba-Attention MoE architecture supporting a 1M-token context length.
Related reading
Integrate NVIDIA Omniverse RTX Sensor Simulation Into Existing Apps
NVIDIA Omniverse libraries, included in NVIDIA Agent Toolkit, provide modular components for incorporating physical AI features into current tools and services. As part of this stack, the ovrtx prerelease library offers a lightweight C and Python SDK for real-time RTX sensor simulation and visualization using OpenUSD scenes. Host applications retain architectural and workflow control by stepping the renderer directly from their internal loops. Integration relies on ovstage for shared OpenUSD data management alongside RenderProducts and RenderVars to configure camera, lidar, radar, and semantic segmentation outputs. PTC Onshape Render Studio is exploring ovrtx to embed RTX rendering directly into cloud-native CAD and PDM workflows.
Tanya LenzDeveloping Healthcare Robotics with GPU-Native Medical Physics Simulation