Developing a conceptual model of a physical system, it is natural to create a mathematical model that will enable you to estimate the quantitative behavior of the system, whether it is a skyscraper, spaceship to Mars, forecasting financial data, computer animation or artificial intelligence.
Optimization seeks to identify the best solution given a set of constraints, including time, cost, materials used and sources. High Performance Computing is the practice of using massive amounts of computing resources to automate these calculations.
Sunfish Studio, LLC (“Sunfish”) is an innovator, and they are working with the semiconductor companies as early adopters to revitalize a stagnant industry. Sunfish is an early stage company creating new and radical solutions for High Performance Computing by using the untapped branch of applied science called Modal Interval Arithmetic (MIA). It is a privately owned Minnesota limited liability company located in Minneapolis, Minnesota.
Nathan Hayes, Inventor of MIA Applications
Nathan Hayes, President, Chairman and CTO, served as an active member of the committee that produced the recently published IEEE Standard 1788 for Interval Arithmetic. He is also the inventor and author of a portfolio of 15 issued and 17 pending patents related to key commercial applications of Modal Interval Arithmetic (MIA).
Mr. Hayes says, “Introducing a new technology is a big task for an entire industry. As the market pain grows worse, our patented MIA technology becomes more relevant and valuable.” Sunfish believes in patience, faith and perseverance.
According to Hayes, present-day technology is floating-point arithmetic, which is used to solve optimization problems, but rounding errors can accumulate over time to produce catastrophic failures, which makes it unable to get accurate, efficient or reliable results.
MIA, Key to Optimization problems
MIA provides guaranteed accuracy, and can prove the existence or nonexistence of a solution – even if it is highly nonlinear – to within any user specified tolerance.
In 2005, the semiconductor industry hit a wall when the clock speed plateaued while miniaturization continued. The resulting trend increases the number of processors on a chip but not the computational efficiency.
With floating-point arithmetic, a well-known law of diminishing returns known as “Amdahl’s Law” makes it hard to keep all of these processors busy. A typical result is 5% efficiency, while 95% consume energy and produce heat while sitting idle.
MIA is the key to a new way of designing software in a highly parallel manner. Software designed with MIA can avoid the law of diminishing returns and keep thousands of processors busy producing maximum speedups in such a way speed can increase proportionately to the number of available processors.
Simple and Cost Effective
Sunfish came up with the vision to harness the radical power of MIA to solve the problems formerly believed to be unsolvable, with the mission to revitalize the industry by adopting MIA technology.
Sunfish’s current focus is the semiconductor companies. MIA can be inserted into existing hardware processors as a low cost and low risk incremental improvement that is backwards compatible with current computing standards.