A YouTuber known as bitluni has recently amazed the tech world with his latest creation – a 256-core RISC-V megacluster that he built from scratch. In a video shared on YouTube, bitluni showcased the entire process of designing, producing, assembling, and testing this miniature supercomputer, providing viewers with a glimpse into the intricate world of tech gadget building.
bitluni’s innovative design involved combining 16 RISC-V superclusters to form a single megacluster. Each supercluster comprised 16 CH32V003 RISC-V microcontrollers connected by an 8-bit bus, with each cluster boasting its own LED for displaying text. To address potential challenges associated with constructing a large and power-intensive PCB, bitluni strategically placed the superclusters in pairs on eight ‘cluster blade’ designs, utilizing additional CH32V203 microcontrollers to bridge the individual superclusters with the main bus.
The video captures the mesmerizing assembly process as bitluni meticulously attaches the microcontrollers to the PCB, solders the GPIO headers, and brings the megacluster to life for testing. While showcasing the functionality, bitluni encountered a minor setback – the LEDs blinked out of sync due to the absence of an internal clock source, resulting in a visually captivating but unsynchronized light show.
Addressing the technical aspects, bitluni explained how he managed bus collisions as all the microcontrollers communicated simultaneously. The megacluster boasted an impressive configuration with 256 RISC-V microcontrollers running at 48 MHz and 17 chips operating at 144 MHz, offering 640 GPIO pins and 256 ADC circuits. Despite the modest 14.7 GHz combined single-core clock rate, bitluni’s creation stands as a remarkable feat, especially considering its compact size and cost-effectiveness compared to traditional supercomputers.
bitluni’s 256-core RISC-V megacluster serves as a testament to DIY innovation in the tech industry, inspiring enthusiasts to push the boundaries of possibility and explore the realm of supercomputing from the comfort of their homes.
