DD:CALLED THE COMPANY: IP Pt. 1
As stated earlier, I called the company twice on Wed. May 13, 1998. I first called and spoke with Mr. Vitucci, of public relations, for about 10 minutes. He was quite busy and we were interrupted on the phone twice. He was helpful but could not answer my technical questions.
So late in the afternoon I called Dr. Sam Shanks, President and CEO of Log Point. He was very willing to answer my questions and we spoke for almost an hour. I believe I understand the current status operandi much better than before.
The company at present has very little revenue inflow. They have developed important intellectual property (IP) in the design of a radically new floating point calculation algorithm that enables floating point calculations to be done much faster than before. For the mathematically challenged, integer math deals only with numbers without decimal points: 1, 2, 5, 10, 21, -97. All microprocessors are able to add and subtract and usually multiply integer numbers. These are rather easy to do and are required for some basic processor functionality. Floating point numbers on the other hand, have decimal points: 1.1, 10.723, 21.0, -97.62345231. These are much more difficult to deal with (try it yourself: which is easier 10/5 or 10.723/1.1?). Most lower tier processors do not have the capability to deal with numbers like this directly in hardware. Floating point math units are complicated, cost silicon real estate and current, and thus increase the cost and size of a processor. If you remember, even processors like Intel's 386 did not have the ability to deal with such numbers directly. In this case, if one has to use such numbers ( spread sheets, graphics, signal processing etc.), one either has to write software to add, subtract, multiply, and divide such numbers or one has to add a special floating point co-processor to do it. I have written such software myself. It is slow, takes code space, takes processing cycles etc. Major bummer. On the other hand, an external co-processor takes space, costs $, etc. With the "digital revolution" in progress the problem is getting worse and worse as applications for signal processing (DSP) explode.
Enter Log Point. Log Point's new algorithms have first been implemented as software routines and libraries (SuperSpeedT) that can be used on many different processors. There are various ways this could be done but is best if a C/C++ compiler is altered to be aware of the library and use it appropriately. These new libraries are much faster than any comparable software solution. Apparently, Mr. Lester Pickett, Chairman and Chief Technical Officer, has developed some further refinements and expansions of these algorithms and this has been incorporated into the software. But the patent for the latest additions hasn't been issued yet. It is expected shortly. In the mean time, the software is being beta-tested by major companies (see the list of processors in earlier post) under beta-agreements. When the patent is issued, apparently, the customers are ready to go and the revenues will commence. Expectations: $25 million in 1998 ($.75 eps).
But even better are the hardware solutions. In this scenario, Log Point takes the new algorithms and has SEVA design a hardware description of the new algorithms. There are low and high tier implementations of the algorithms providing a matrix of possible solutions and designs. These new hardware descriptions can be used in two different ways. 1. Log Point can hire fabs to build the parts as external floating point co-processors (much faster than previous processors) and sell them. 2. Log Point can sell the hardware descriptions to other companies as IP, so that they can incorporate the designs into their own chips. So company x is making a communications processor and wants floating point math. Log Point sells them the design, and receives a one time buy out for the product lifetime, or receives continuing royalties on sales of company x's product. Cool. The expectations of this field: $90 million in two years. I asked Mr. Shanks how this was calculated. He said it was a conservative estimate and was without doubt that once they get the hardware descriptions designed by SEVA and parts made etc. that they could do that amount of business in two years.
So much for IP
End part 1
Mark Nispel
Embedded Software Engineer
Not yet a share holder
Not affiliated with the company in any way |
