>> is it new info that they (5 listed vendors) are using the i960?
It is hard to pin down exactly who is using I2O for computers and I/O type products. First, I2O has not been a specific bragging point in advertising products. Second, intelligent I/O has been around for awhile, so differentiating between I2O and i2o is not easy. Examine carefully Intel's source listing under I2O and you will find a mixture of products that clearly use the i960 RP chip and others that use some version of the i960 chip to do smart I/O.
>>Also, there were numerous other vendors listed which announced I2O support,
including some biggies (Compaq, Dell, DEC, HP). Is there any chance these
ones make use of the i960 as well?
I2O is an open standard. Anyone is free to implement it either in hardware or software, or any combination, without ever talking to Intel or Wind River. As indicated above, many vendors have implemented intelligent I/O devices without the so-called I2O chip or Wind River's IxWorks.
HOWEVER, at the moment, as far as I know there is only one I2O chip on the market, and that is Intel's i960 RP, and every bloody one comes with IxWorks, like it or not. I doubt that any vendor is implementing anything that smacks of I2O (notice the capital I) without using the I2O chip. It is almost inconceivable to me that anyone in the business of making add-in cards, or motherboards, or RAID controllers, etc. would want to make their own version of I2O. In other words, I believe that every item on the list you referred to involves the i960 RP chip. (Let me make this clear. If they say I2O then today they must be referencing the i960 RP chip. If they say "smart" or "intelligent" and imply off-loaded, efficient I/O, then they might be referencing a design that pre-dates I2O.)
It is not inconceivable to me that someday someone will introduce a competitive I2O chip. In fact it is expected. I have been predicting for some time that I2O functionality someday will be integrated onto a multifunction chip. But when that happens, I would expect the producer to talk first to WIND about porting IxWorks to the new chip.
If you are thinking that I2O could in fact be very big, you are right. Reread my post on valuing WIND stock, where I indicated that I2O alone conceivably could be worth $150 per share of WIND stock.
If you haven't seen the most recent copy of OEM Magazine (or EE Times), track it down and get one. Intel inserted a 15 page advertisement about I2O, with lots of comments by various vendors. Be sure and notice WIND's advertisement on the last page.
Notice that Intel has set up a special home page for tracking I2O developments.
Microsoft is showing real interest finally, too. It is not necessary that vendors wait for version 5.0 on Windows NT before I2O can be used with Windows NT. The slight amount of interaction necessary with the operating system can be bridged with special software until the Microsoft addition is complete and available. I believe that primary need that Microsoft wants to address as soon as possible, is PC clustering. Clustering has become viewed as the best way to fashion mission critical, large servers that have the robustness to compete with high-end Unix boxes, and even mainframes. However, PC clustering requires advanced I2O capability (although NetFrame and possibly others are doing it with existing I2O).
By the way, recall Michael Greene's post when he said, when asked their opinion on WIND's I2O deal with Intel, INTS responded: "He said that he could not evaluate it since WIND will not give out any details. However, he claims that Motorola believes that the I2O standard will not fly."
To put this in context, note that INTS is a member of the I2O SIG, and therefore should know all about I2O. Motorola is not a member, and so far is not participating in the I2O endeavor. Why would INTS defer to Motorola on the subject of I2O?
It seems to me that the argument in favor of a standardized I/O processor with the feature set established for I2O is compelling. It also seems to me that if Wintel wants it they can have it, since they control the lion's share of all desktop computer categories, including servers, and soon workstations.
A final, very important point about I2O. WIND did not simply port VxWorks to the i960 RP chip, and then turn I2O development over to Intel, waiting impatiently to begin collecting royalites. Actually VxWorks was already running on the i960 family, so porting per se was not an issue. But lots of fundamental questions about how to structure the RTOS was at issue, and WIND turned loose Dave Wilner, one of the co-founders, to personally design and see to the development of what became IxWorks. For this extra special effort, all stockholders in WIND owe Dave a hearty thanks. But because of this contribution, in my mind WIND is as much an owner of I2O as is Intel, and is irreplaceable. No matter how big the royalty checks from Intel to WIND get with the future success of I2O, WIND deserves every penny. (By the way, upon reading an DMG analyst report on WIND, I discovered they think the I2O royalty is $1.50. You may recall that I guessed it was somewhere between $1 and $3, so I used $2. Michael Green quoted WIND as saying the royalty is on the $1 side of a range of $1 to $5. Conclusion: I think it is $1.50, which is what I am henceforth using in my model.)
Hope this helps about I2O.
Allen |
