Flex‑RV: An Open‑Source Sub‑$1 Flexible RISC‑V Microprocessor with Integrated Machine‑Learning Accelerator
The article presents Flex‑RV, a sub‑dollar, bendable 32‑bit RISC‑V microprocessor built with IGZO TFT technology that integrates a programmable machine‑learning accelerator, achieves up to 60 kHz operation at less than 6 mW, and demonstrates reliable performance under mechanical stress for emerging wearable and smart‑packaging applications.
Recent advances have produced a flexible, non‑silicon 32‑bit microprocessor called Flex‑RV, fabricated using low‑temperature IGZO thin‑film transistors on a polyimide substrate, enabling sub‑$1 unit cost and a thin, bendable form factor.
Flex‑RV implements an open‑source RISC‑V ISA (RV32E) using the minimal Serv CPU core, and extends it with four custom instructions to control an on‑chip SIMD machine‑learning accelerator, allowing execution of compiled C programs and ML workloads.
The chip occupies 17.5 mm², contains roughly 12,600 logic gates, and operates at up to 60 kHz while consuming under 6 mW. It was tested both on‑wafer and after assembly onto flexible printed circuit boards (FlexPCB), showing stable functionality when bent to a 3 mm radius.
Mechanical testing revealed a modest performance variation of +2.3 % in tensile bending and –4.3 % in compressive bending, with average clock frequency remaining above 52 kHz across 13 FlexPCB samples.
By leveraging the open, royalty‑free RISC‑V architecture and low‑cost IGZO TFT manufacturing, Flex‑RV addresses the cost and form‑factor barriers that prevent microprocessors from being used in emerging applications such as smart labels, wearable medical devices, and disposable sensors.
The work demonstrates that flexible, programmable processors with integrated ML acceleration can be produced at scale, opening a path toward ubiquitous, low‑power computing in everyday objects.
Architects' Tech Alliance
Sharing project experiences, insights into cutting-edge architectures, focusing on cloud computing, microservices, big data, hyper-convergence, storage, data protection, artificial intelligence, industry practices and solutions.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.