Full-Fidelity Simulation, up to 1000x Faster.

Product Overview

What does Vinci's software do for hardware engineering teams?: Vinci enables engineers to perform extremely fast thermal simulations of complex hardware designs, at any scale, from nanometers to meters, with guaranteed accuracy and full IP security.

What kinds of physics can Vinci model?: Currently, Vinci can simulate steady-state conduction, transient conduction, and steady-state thermoelasticity (warpage).

Who is Vinci intended for?: Vinci is designed for thermal engineers who need physics simulations of complex designs, from nanometer-scale to meter-scale. Vinci was designed to allow every engineer, not just simulation experts, to quickly and easily access high-fidelity simulation results.

Customers designing semiconductor or electronic products can use Vinci for individual components (DRAM, SRAM, CPU, BEOL, Interposer, PCB, etc.), advanced packaging (2.5D, 3D, 3.5D), PCB assemblies, electronics enclosures, and more. At larger scales, Vinci can simulate aerospace, automotive, industrial, or any other advanced applications requiring thermal engineering.

How is Vinci different from other CAE tools?: Vinci eliminates the need for manual model simplification and preprocessing, saving hours of tedious work while producing simulation results up to 1,000 times faster than traditional FEA tools. Additionally, Vinci can run simulations on designs several orders of magnitude of complexity beyond what traditional FEA tools can handle, operating at resolutions down to 2 nanometers.

What inputs does Vinci require?: Vinci uses industry-agnostic standards like STEP and STL, as well as semiconductor design files such as OASIS, GDSII, ECXML, or IPC-2581, along with standard documentation files for material properties, power maps, and boundary conditions.

How are layer materials and thicknesses configured?: Users can configure layers via a configuration file, allowing for complex setups including directional conductivity (X, Y, Z) and detailed layer mask definitions. It’s also possible to configure layers manually by entering materials, their thermal conductivities, and layer thicknesses.

What resolution and accuracy settings are available for simulations?: You can customize resolution for each simulation axis, samples per layer for depth granularity, and/or target accuracy.

Can I view and export simulation results?: Yes. Vinci provides interactive 2D and 3D visualizations of simulation results. You can also download results as CSV files for temperature summaries per layer or source, and/or as VTK files for full 3D data compatible with visualization tools like ParaView.

Many AI systems are prone to hallucinations. Can we trust Vinci’s results?: Vinci is trained on physics, not on user data. There’s no hallucination risk because it automatically confirms all results against physics fundamentals.

Is customer data used to train Vinci’s AI models?: No. Vinci never uses customer data to train or enrich its models. It simply works out of the box.

How is Vinci deployed?: Vinci is packaged as a Debian (.dpkg) application that installs a Docker container on the customer’s infrastructure. It can run on-premises or on cloud platforms like AWS, Azure, or GCP.

What platforms are supported?: Linux-based systems and GPU are required. Specifically, Ubuntu 22.04 is supported. Chrome-based browsers are recommended for user access.

Who owns the simulation input and output data?: All input and output data belongs to the customer. Vinci has no access to customer proprietary data.

What data does Vinci collect?: Only anonymized usage metrics are collected via a secure logging service. These include timestamps, GPU usage, model accuracy, and compute times.

How is the collected data secured?: Logs are encrypted at rest using AES256 and transmitted via HTTPS. Metadata access is restricted to Vinci engineers with multi-factor authentication.

How are system performance and accuracy monitored?: Metrics are captured during each simulation run, including compute time, GPU usage, and accuracy levels. These are analyzed to ensure consistent performance.

Can customers restrict backend connectivity?: Backend access currently uses GCP IPs and domains. Static IP support is under development.