Charles Moore's 144-core processor goes on sale for $ 20

Charles Moore, the creator of the programming language Forth (Forth), brought to the stage of industrial production of the unique product - multi-core processor GA144 . The 10x10 mm chip has already gone on sale at a price of $ 20 (when ordering from ten pieces), motherboards for it are also available. In fact, this is the hardware embodiment of the Fort programming language itself.

The extremely unusual processor in a number of parameters has no equal among the CPU:

• 144 independent cores, which are activated only upon receipt of instructions, that is, this processor does not have such a characteristic as “clock frequency”;
• speed of execution of instructions 1400 picoseconds (equivalent to 700 MHz);
• power consumption of 7 picojoules per instruction;
• power consumption in sleep mode less than 100 nanowatts;

Multilevel programming makes it possible to write very fast and simple microcode, or use a high-level programming language, or combine both of these methods in clusters of computational cores, indicating “hosts” and “coprocessors” among them.

Chuck developed this processor on his own using the OKAD II VLSI tools he created. GA144 development tools, including assembler / compiler and sample code, are distributed free of charge in a package under the general name arrayForth.

Experts are trying to understand what are the targeted areas of application for GA144. There are many options:

• Robotics (manipulators, prostheses, autonomous mobile robots);
• artificial intelligence and neural networks (classification, signal / image recognition);
• “On-board systems” (real-time status diagnostics, motion control);
• “Academic” application (hardware for digital signal processing courses, parallel programming, computer architecture);
• speech recognition and synthesis;
• modulators / demodulators of signals.

The processor developer himself supplements this list with various energy-efficient applications (wireless power reception module), portable devices, image processing systems, complex control systems, cryptography, high-performance signal processing, simulation and synthesis programs, and other applications that need mass parallelism.

GA144 Architecture (PDF)
arrayForth Guide (PDF)