>This sounds remarkably similar to the benefits attributed to i2O. Just wondering if Intel has incorporated VxWorks into the core processor, or is this a related development that reduces the need for I960 processors?
----------------------------------------------------------------------------------------------
The dual bus Intel has just introduced is an acknowledgment that processors and rich data types (e.g. graphics, multimedia) are out-pacing traditional input/output processing in desktop computers. However, it appears to me that the dual bus is simply a stop-gap measure, similar to using cache memory or pipelined instruction set, rather than a more general solution of the type possible using I2O. But this in only a guess, since the announcement mentioned two buses, not an I/O processor.
The advantages of I2O are more than just speed derived from off-loading most I/O from the CPU.
I2O can be programmed to handle rich data types without even bothering the CPU. This includes a zillion forms of compession, complex communication protocols, security, complex, hard real-time multi-media data streams, and no doubt many, many more.
Once I2O is broadly used, complicated device drivers specific to the hardware and operating system can be replaced with much simpler robust drivers, not only eliminating high development and maintenance costs which burden hardware and software developers, but also eliminating a primary source of problems encountered by end-users.
Ultimately even more important, I2O helps enable creative restructuring of both computer and network architecture. Disks and all other I/O devices (including I/O components normally considered part of a computer) no longer need to be slaved to a computer and only then perhaps shared over a network. I2O helps open up decentralization of computer processing, as exemplified by its appeal for use in clustering disparate computers.
This breakup, or decentralization, of computers will not occur overnight, due to anyone's insight into a new computing paradigm. Rather, the PC paradigm (which includes the NC), will be worn down at an ever-increasing pace, caused by the dismemberment of the computer into its component parts only to be reconnected in creative ways over the network. As the dismemberment proceeds, exotic means of making computers sensitive to human needs will emerge, the end result of which will be the death-knell of the PC paradigm.
To the extent that the dual bus trick appears to prolong the life of desktop computer designs, I don't regard the announcement as good for WIND, but even so there are a multitude of non-server I2O applications in the hopper, and many, many more to come. I still expect to see I2O on high-end PCs and workstations in the near term, and on all PCs in time.
Allen |
