Compiler for Fast, Accurate Mathematical Computing on Integer Processors, Phase I

The proposers will develop a computer language compiler to enable inexpensive, low-power, integer-only processors to carry our mathematically-intensive comptutations at high speed, with excellent computational accuracy. This will enable space systems designers to select from a wide range of radiation-hard processors for math-intensive command and data handling applications such as spacecraft attitude control, advanced sensing, instrument data processing and calibration, and autonomous operation. Integer-only processors are capable of accurate math-intensive computing if properly programmed using a fixed point computational model. This presents a major challenge to the program designer. Software tools are available to help the programmer analyze fixed point implementations, but not to create them. This kind of trial-and-error design cycle can be expensive, time-consuming and error-prone. Our compiler will analyze system specifications, such as input data ranges and formats, architectural constraints, and a description of the computational algorithm. It will insert data scaling operations into the integer instruction stream to make the most effective use of the internal data representation. The resulting code will be an order of magnitude faster and more compact than a software floating point implementation, with very competitive computational accuracy. This technology will be available for NASA programs immediately after Phase 2.

This is a mature technology (TRL 7+) — the realistic path in is usually NASA's Technology Transfer Program: licensing an existing NASA patent, or a Space Act Agreement to use NASA facilities/expertise directly. NASA also runs a startup licensing program with no upfront fee for companies formed to commercialize a specific NASA technology.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.