XLNS Research Purchases Patent to Speed Up Math on DSP and Embedded Processors.

BOULDER, Colo.--(BUSINESS WIRE)--June 18, 1999--

XLNS Research announced today that it has recently purchased a patent that uses logarithms to speed up computer arithmetic.

This invention has the potential to enable cheap fixed point DSP and embedded processors to perform single precision calculations at speeds much faster than conventional floating point software. The patent was purchased from an independent inventor, Mark Arnold, who is also a member of the Computer Science department at the University of Wyoming.

The patent is titled "Method and Apparatus for Fast Logarithmic Addition and Subtraction" (U.S. Patent 5,337,266 issued on August 9, 1994). Acquisition of this patent is the first step in XLNS Research's plan to develop a portfolio of proprietary technology that can be sold or licensed to established microprocessor and software firms in the embedded controller and digital signal processing (DSP) industries. The patent also enables XLNS to reassure potential customers that its future software will be free of patent infringement issues.

XLNS Research is pleased to join the ranks of firms that own patented inventions that use a logarithmic representation of numbers to speed up calculations. Motorola, VLSI Technology and Log Point Technologies are some of the firms that have been granted patents in this field. Although university and corporate research in this field extends back to at least the 1970's, inventions capable of supporting 32-bit logarithmic calculations only became practical in the last decade because of advances in memory capacity. In the academic and patent literature, this field is commonly referred to as Logarithmic Number Systems (LNS). Log Point Technologies prefers to use the term "exponential floating point" (refer to their press release of March 17, 1999).

Because other firms have LNS patents, and because the public domain contains many LNS inventions that have been published in research articles, XLNS Research's patent does not enable it to prevent others from entering the LNS field. The common features shared by all LNS inventions are single cycle operations (multiplication, division, squares, square roots), dynamic range and precision comparable to floating point, and minimal CPU silicon (if the logarithm tables needed for addition and subtraction can be stored in external memory). LNS technology thus offers the potential to speed up the "multiply-accumulate" operation that is prevalent in DSP applications. Of course each LNS invention also offers a unique combination of precision, speed, space, and power consumption. XLNS Research's recently acquired patent enables XLNS to offer a proprietary 32-bit LNS at a competitive price.

XLNS Research hopes to make the majority of its revenues by selling or licensing its present and future proprietary LNS technology to established microprocessor and software firms. In order to stimulate demand for these products, XLNS is presently developing LNS software that is based on public domain LNS technology. This software will eventually be distributed at low / no cost as per the "open source software" business model. As noted by Mr. Arnold, "The GNU and Linux operating systems have achieved considerable success with open source software, and I hope that this strategy will also enable XLNS Research to achieve comparable success with my patent."

Mr. Arnold teaches courses in computer architecture at the University of Wyoming. He is also the author of a recent textbook about integrated circuit design ("Verilog Digital Computer Design: Algorithms into Hardware," Prentice-Hall, 1998, www.phptr.com/ptrbooks/ptr_0136392539.html). He has written several research papers on LNS technology. Mr. Arnold believes that his 1982 thesis was the first published work to suggest the use of logs and antilogs for interpolation of the function required for logarithmic addition.

XLNS Research is a Colorado sole proprietorship that uses consultants in lieu of a permanent staff. The founder, Mark Winkel, is a software engineer formerly with one of Hewlett-Packard's R&D labs. This is a very speculative enterprise and will remain so until LNS technology is used in a few high volume products, attracts customers, and generates profits. XLNS hopes that there will eventually be some lucrative niche markets for its products. XLNS is investigating financing options so that it can overcome severe capital constraints that are impeding product development.

XLNS Research believes that its products will coexist in the market with those from Log Point Technologies (OTC BB:LGPT). Each firm's patents have different footprints and are thus apt to appeal to different segments of the market. For more information about Log Point's products, refer to www.logpoint.com.

The patent was purchased by XLNS Research on May 13. On June 17, XLNS received word from its patent attorney confirming that the associated paperwork has been sent to the U.S. Patent and Trademark Office (PTO). It is anticipated that it will take the PTO two to three months to update its records accordingly.

Although the primary audience for this press release is the embedded controller and DSP industries, XLNS Research realizes that this press release might also be of interest to LGPT investors. XLNS therefore voluntarily discloses that throughout 1998 (i.e., before XLNS Research was established), its founder accumulated shares in Log Point Technologies on the open market. After establishing XLNS Research, Mr. Winkel ceased trading in LGPT stock (with the exception of the purchase of 5,000 shares between 4/8/99 and 4/12/99). Presently, Mr. Winkel's holdings are less than 1% of LGPT's issued stock. This position in LGPT stock will eventually be sold in full or part. Mr. Winkel has not, and will not, short LGPT stock.