This is the article mentioned in the recent press release in PC Today magazine...
Floating-Point Calculations Boosted
A new floating-point data technique created by LogPoint Technology should significantly increase logic computation speeds within a computer's processor for use in handheld and network computers, signal processors, and laser printers. The new advance is a completely
logarithmic data format that provides precise logic computations at ultra-high speeds, enabling faster processing of graphics, audio, and interaction in multimedia environments.
According to Lester Pickett, chief technologist for LogPoint, this innovation is based on exponential floating point technology, which increases the speeds of logic computations in the processor, saving silicon space and decreasing computation time. Unlike other floating point technology, Pickett says, LogPoint's floating point multiplication actually is performed as an integer addition. This results in an increase in speed by six to 10 times over conventional
floating point techniques for such graphical computations as reciprocal, divide, and square root functions. In fact, the company says its tests indicate 6% to 44% speed performance increases over
competing conventional software floating point math. Company officials say LogPoint initially will target the handheld PC market for units using Hitachi SH-3, ARM, MIPS 4001, and Motorola PowerPC 800 series of processors. The technology should be licensed to ASIC, microprocessor, and DSP manufacturers within the year, and Pickett says other embedded markets will soon follow. |
