Start with the technical problem worth solving.
Cerebral Chips welcomes focused conversations about edge-AI workloads, architecture, software enablement, verification, and open engineering.
No form, tracking funnel, or intermediary is required
Write to the closest technical context.
Each address reaches a distinct conversation path. A short, concrete introduction is enough to start.
Company questions, updates, and general introductions.
Technical collaboration across workloads, software, architecture, and validation.
Help the first reply reach the substance.
- 01The workload, model family, or system constraint you are exploring
- 02Where the current software or hardware path becomes limiting
- 03The evidence, platform, or artifact already available
- 04What a useful first technical exchange would establish
Where technical conversations can connect.
- Quantized inference workloads
- RISC-V vector and matrix execution
- Compiler and runtime enablement
- Memory-aware architecture modeling
- RTL and verification methodology
- Reference platforms and open tooling
Before you reach out.
01What is Proton LPU?
A planned RISC-V-based accelerator architecture for quantized transformer and language-model inference.
02Can I use the hardware now?
You can build and run Proton NPU in RTL simulation using its public repository. It runs a bare-metal scalar, vector and matrix program. Proton LPU-E0 remains a separate research direction; no physical chip or FPGA release is offered by this milestone.
03Is Cerebral Chips publishing repositories yet?
Yes. Proton NPU is public on the Cerebral Chips GitHub organization, with source, build instructions, architecture documentation, licenses and verification records. The Open Source page links to the repository and contribution guide.
04What kinds of collaboration are relevant?
Focused technical work around edge-AI workloads, quantized inference, compiler and runtime enablement, RISC-V execution, architecture modeling, RTL, verification, and reference platforms.
05Are benchmark results available?
Proton NPU publishes functional RTL verification results, including eight passing combined workloads. These are correctness checks, not AI throughput benchmarks or silicon performance measurements.
06Where will company social links appear?
Our confirmed GitHub organization and Proton NPU repository are linked on the Open Source page. Other social profiles will be added when confirmed.
Every machine should think.
Tell us what the machine needs to do where it operates.