Free Resources. No Email Required.
(Well, mostly.)
We believe in earning your attention, not gating it. Most resources are available immediately. For detailed technical documentation, we ask for NDA—not because we're secretive, but because our customers expect confidentiality.
Free Educational Content
Download immediately, no forms to fill.
NPU Architecture Fundamentals
Understanding the building blocks of neural processing units—from PEs to memory hierarchies.
Best for: Technical managers, architects new to NPU design
Build vs. Buy Analysis Framework
How to evaluate make/buy decisions for NPU IP, including hidden costs most teams miss.
Best for: Engineering managers, procurement teams
IP Evaluation Checklist
30+ questions to ask any NPU IP vendor. Use it on us too.
Best for: Technical evaluators, due diligence teams
Sparsity in Practice
How hardware-accelerated sparsity actually works, and when it helps (and when it doesn't).
Best for: Hardware architects, ML engineers
Technical Deep-Dives
Detailed documentation for serious evaluation.
PE Array Architecture
Detailed look at our 32×32 processing element design and local accumulation strategy.
Verification Methodology
How we achieve >95% coverage and what that means for your integration.
Memory Subsystem Design
Three-tier memory hierarchy architecture and interface specifications.
Coming Soon
Content we're working on.
- Case study: NPU integration at a fabless startup
- Webinar: Common NPU IP integration mistakes
- Guide: DFT considerations for NPU blocks
- Technical brief: Memory bandwidth optimization
Want to know when new content is available?
Why We Share This Freely
The semiconductor industry has a knowledge-hoarding problem. Information that should be freely available gets locked behind paywalls, NDAs, and "contact sales" buttons.
We think that's counterproductive. If you're evaluating NPU options, you should be able to learn the fundamentals without talking to a salesperson. If you're building your first NPU-based chip, you shouldn't have to reinvent wheels that others have already figured out.
Our bet: teams that understand NPU architecture deeply will make better decisions—and some of those decisions will be to work with us. But even if they don't, we've contributed something useful to the industry.