Why This RISC-V Microcontroller Matters: The Baochip-1x Story

The Baochip-1x represents a fascinating intersection of hardware innovation, open-source philosophy, and pragmatic engineering. Andrew 'bunnie' Huang's latest project isn't just another microcontroller—it's a deliberate challenge to decades of embedded computing conventions.

{{IMAGE:1}}

The MMU Question: Why Memory Management Matters

The most striking feature of the Baochip-1x is its inclusion of a Memory Management Unit (MMU), a component that's conspicuously absent from virtually every other microcontroller in its class. For those unfamiliar with OS terminology, the MMU is what separates the software running on your desktop or phone from what runs in your toaster. It enables secure, loadable applications by giving each its own virtual memory space.

Huang traces this absence back to the ARM7TDMI core from the 1990s, when transistors were scarce and memory even scarcer. Back then, virtual memory wasn't practical for devices with just a few kilobytes of RAM. The ARM7TDMI became a runaway success, shipping over a billion units and establishing ARM as the dominant player in embedded systems.

Fast forward thirty years, and we're dealing with transistors that are thousands of times more capable. A fleck of silicon smaller than your pinky nail today contains more transistors than a full-sized desktop PC from the 1990s. Yet the convention persists: small systems get flat memory spaces with no address isolation.

The root cause is surprisingly straightforward—MMUs are too valuable. Without one, you can't run Linux, BSD, or Mach. When ARM split their IP portfolio into A, R, and M-series cores, the low-cost M-series cores were forbidden from having an MMU to prevent price erosion of their high-end A-series cores. Instead, they introduced the MPU (Memory Protection Unit), a proprietary hack that provides some memory security but lacks features like swap memory.

The Open Source Timing Question

Why pursue a "mostly open RTL" SoC now rather than waiting for fully open source PDKs? Huang's answer cuts to the heart of practical open-source strategy. While he'd love to see a fully open silicon stack from fabs up, he estimates it could take more than a decade before we have a sufficiently robust open source silicon ecosystem to market economically competitive SoCs.

For developers creating embedded products today, that leaves a stark choice: continue using Cortex-M ARM devices and tune software to their proprietary MPU, or embrace partially open solutions that enable progress now. "Do I really want to spend my time porting Xous to use ARM's proprietary flavor of memory protection? Surely not."

This pragmatic approach acknowledges that software drives demand for silicon, not the other way around. By making partially open SoCs available today, we can begin de-leveraging ARM together, creating a mature application stack that will be ready when truly open silicon alternatives become economically viable.

The Hitchhiking Strategy

Perhaps the most ingenious aspect of the Baochip-1x is how it came to be. Huang didn't raise venture capital or rely on personal wealth. Instead, he "hitchhiked" on a 22nm chip primarily designed by Crossbar, Inc.

Crossbar wanted to build a high-performance secure enclave using their RRAM technology for non-volatile storage. They approached Huang for strategy advice, and he agreed on one condition: he could add a CPU core of his choice and sell a version under his own brand. This arrangement gave birth to the Baochip-1x.

The economics are compelling. Modern SoC costs are driven largely by peripherals and memory, with the CPU often just a couple percent of the area. All peripherals are "memory mapped," meaning the same physical silicon can effectively have two different personalities by switching out CPU cores.

What "Mostly Open" Actually Means

The Baochip-1x isn't fully open source, and Huang is transparent about what that means. Everything that can "compute" on data is available for simulation and inspection. The closed portions include the AXI bus framework, USB PHY, and analog components like PLLs, voltage regulators, and I/O pads.

Critically, none of the closed components are involved in data transformation—they're effectively "wires." While this isn't perfect transparency and can't definitively rule out back doors in black-box components, it's far better than the fully NDA SoCs we currently use to handle our secrets.

The Road Ahead

With wafers already fabricated and hundreds of chips distributed through an early sampling program, the Baochip-1x is transitioning from experiment to product. Two wafers are earmarked for production qualification, with a few thousand chips available through the Dabao evaluation board pre-order campaign.

Huang acknowledges there's still much work to be done: documentation to write, drivers to bring up, and more programming languages and OSes to support. While he champions the Rust-based Xous OS, he's clear that it doesn't have to be the only code base for Baochip.

The project represents more than just another microcontroller—it's an invitation to build a brighter, more secure, and more trustable open source future together. By making hardware security primitives accessible, Huang hopes to empower community leaders to create the foundation for an eventually fully open silicon-to-software ecosystem.

{{IMAGE:2}}

The Baochip-1x story illustrates a crucial principle in open-source hardware development: perfect can be the enemy of good. By embracing "mostly open" solutions today, we create the momentum and community needed to achieve truly open alternatives tomorrow.