Raspberry Pi Pico 2 at 873.5MHz with 3.05V Core Abuse
A determined tinkerer embarked on a quest to push the Raspberry Pi Pico 2's RP2350 microcontroller to its absolute limits, leveraging extreme voltage abuse and dry ice cooling. This deep dive into hardware overclocking showcases meticulous methodology and impressive results, reaching an astounding 873.5MHz. It's a classic Hacker News tale of curiosity, engineering defiance, and the surprising resilience of inexpensive hardware.
The Lowdown
Mike, the author, initiated a festive deep dive into the extreme overclocking capabilities of the Raspberry Pi Pico 2 (RP2350), aiming to discover its maximum stable frequency. Building on prior Raspberry Pi overclocking exploits, he meticulously documented his process and findings.
- Initial experiments with increased core voltage (from 1.1V to 1.7V) revealed a significant increase in clock speed (from 312MHz to 570MHz) but also a rapid rise in chip temperature.
- Adding a small heatsink and fan improved cooling, allowing speeds up to 678MHz at 2.2V, though the onboard voltage regulator proved insufficient for higher current demands.
- Identifying a test point (Test Point 7) on the Pico 2 allowed direct external voltage injection, bypassing the regulator's limitations.
- A plan for extreme cooling with dry ice (-80C) was formulated to manage heat, with a strict warning about safety precautions for handling solid CO2.
- A rigorous test setup was devised using the CoreMark benchmark, dual-core operation, a 1MHz reference clock, and custom firmware for accurate, stable measurements, with results displayed on a PicoVision monitor.
- With dry ice cooling, the Pico 2 easily achieved 700MHz using the internal regulator. Employing an external power supply, it reached 800MHz at 2.8V.
- Further pushing the voltage to 3.05V and testing multiple units, the highest stable frequency recorded was 861.6MHz (864MHz requested), with a brief, unstable peak of 873.5MHz.
- Experiments with the chip's RISC-V cores showed a slightly higher performance per MHz (nearly 5% better) compared to the ARM cores.
- An attempt with a large copper heatsink proved ineffective at low temperatures, possibly due to poor thermal contact.
In conclusion, the RP2350 proved remarkably robust, surviving high voltages and extreme cold without permanent damage. The author noted diminishing returns above 700MHz, suggesting even better performance might be possible with more effective cooling like liquid nitrogen. For practical, less extreme overclocking, around 1.6V could achieve 500MHz without additional cooling, making the Pico 2 a fun, low-risk platform for hardware experimentation due to its affordability.