Intelligent I/O (I2O) is Intel’s solution to a major drawback to the PC’s ability to handle the data flows necessary in the modern multi-media world of networking. While end users may complain about slow speeds, both for wide-area and local networks; the problem currently does not rest with the client PC. For wide-area networks, particularly the Internet, PC’s generally are retarded by the limited speed of common modems (14.4, 19.2 and 28.8 bits per sec). For Local Area Networks network administrators point to network servers and network traffic. Unfortunately for both Intel, Microsoft and Novell, the current PC platform with Windows NT or Netware is not capable of replacing the Unix server for high-volume activity. If you doubt this opinion, check the Investor Business Daily on March 1, when it ran an article in which Intel is having to figure out why the Pentium gets bogged down under certain circumstances serving data.
The problem is not with Windows NT. In a way, the problem is not with the Pentium chip. The problem is with the very idea of a highly pipelined CPU being frequently interrupted to help handle data, and still being expected to perform other operations quickly. While Intel may find a work-around to lessen the problem in some circumstances, in general a much more powerful solution is needed. I2O is that solution.
I2O gives the PC platform mainframe I/O capacity, better than Unix servers. Not only will the I/O capacity expand data handling capacity by orders of magnitude, leapfrogging the competition, it can be synchronized routinely for real-time audio/video transfer. Moreover, I2O is truly plug-and-play. This alone makes the approach worthy of general use. The days of I/O driver mysteries and incompatibilities are gone with I2O, since detailed driver requirements are replaced with a high-level I/O API.
What is going to happen, therefore, is that Intel, Microsoft, and Novell are going to encourage the production and availability of I2O boards to upgrade Pentium servers as fast as possible. Next, I would guess that Compaq and HP, followed by all other PC makers, will include I2O on the motherboard (made easy by Intel being the world’s largest motherboard manufacturer). First high-end client PCs, then all PCs, i.e. not just servers, will begin to add I2O to handle very high-speed network traffic and real-time multi-media.
Finally, peripheral devices, particularly printers, communication servers, routers, and ultimately high-speed modem-type connectors, such as the cable-modem should quickly add I2O.
So, I2O is a very important for Intel, Microsoft and Novel, not to mention Compaq, HP and numerous network communications companies. Why is it such a big deal for WIND?
The reason is that I2O requires an embedded chip with a specially engineered RTOS. Intel’s version, which is hitting the street now for evaluation, consists of their specially made chip, the i960RP, coupled with WIND’s specially prepared RTOS, called IxWorks. Every copy of Intel’s i960RP contains a run-time copy of IxWorks, with a run-time license fee to WIND.
The full license revenue stream stemming from Intel’s shipments of the i960RP chip will only be felt after vendors complete their designs and ramp up production. I would guess the numbers ought to get exciting starting in 1997. However, each company having to engineer products using I2O will need to debug connectivity with the chip. The practical way to accomplish this is to license WIND’s Tornado Integrated Development Environment tailored to IxWorks. Intel is making this convenient by shipping a CD-ROM containing Tornado in each evaluation package. Hundreds of companies will be receiving evaluation packages.
Finally, companies being encouraged in this way to learn and utilize Tornado and WIND’s RTOS, are apt to realize that WIND’s state-of-the-art tools are vastly superior to home-grown methods as well as many tools supplied by competitors. Thus, there should be a flood of new design wins in addition to I2O product design wins.
This makes for yet another important deal developing for WIND.
By the way, everything mentioned here is my opinion formed from searching I2O-related home pages, reading IBD, and my reading between the lines during WIND’s analyst conference call March 1. I would be interested in other opinions, particularly any challenging my articulations of industry problems and solutions. |
