Is Intel really going to rule the world?
Your post is an excellent argument for selling Wind River and investing the proceeds in Intel.
Rapidly increasing speed of microprocessors means that the capability of microprocessors continually expand to absorb applications previously considered beyond the capability of general processing. The zero marginal cost of production mandates migrating applications to software, where possible.
The marginal cost of absorbing an application in a general purpose computer is zero, as you indicate. However, an embedded system runs a single application by definition and there's a cost associated with making that functionality available. If "any given level of application functionality will eventually succumb to general-purpose processing", then Intel and Microsoft will rule the world and Motorola and the RTOS vendors will lose.
Historically, that has been the case. Intel has grown much faster than Motorola because they've focused their attention on the processor used in the PC and left the leadership in the embedded market to Motorola and the DSP market to TI. However, the pendulum has now swung in the other direction. TI's DSP business is growing at a 50% rate and they're building new fabs to accommodate the demand. Intel has taken note with the single-instruction multiple-data (SIMD) instructions in their MMX initiative (essentially a generalization of the multiply-accumulate DSP instruction), but their x86 revenue is still growing at a much slower rate than the TI DSP business. The embedded processor market is also growing faster than the x86 and other general purpose computer processor markets (SPARC, Alpha, PA-8000 etc.).
Essentially, this means that your "Law of General Processing" is wrong. Life is much more complex and dynamic than a simplistic mathematical theorem with "QED" at the bottom. Perhaps Intel isn't such a slam dunk investment after all.
While it is true that application-specific hardware devices will always be invented to handle demanding applications, you fail to understand the dual forces that constantly erode advantages of hardware, supplanting it with software.
This isn't true at all. I agree that hardware functions migrate to software. Someday we won't need any hardware at all! You fail to understand that a soft core is software that resides in an ASIC, hence the name "soft core". Calling a soft core "specialized hardware" is like labeling EPROM firmware "hardware" simply because it resides in a sliver of silicon. The hardware design languages (Verilog and VHDL) that describe the contents of the ASIC are easily understood by systems programmers because the syntax resembles the Pascal computer language. Hardware and system software engineers are always pointing their fingers at each other. You get a great deal of synergy by having the Verilog programmers across the street from the firmware programmers.
Did you know that VxWorks plays an integral role in the world's fastest, parallel-processing computer built by Sandia Labs along with Intel using over 9,000 microprocessors?
This is exactly the sort of consulting gig that Wind River should be pursuing. Unfortunately, the company is focusing it's limited resources on porting VxWorks to 20 processors to increase royalty income. The main competitive advantage that RTOS vendors have over Microsoft is that they support processors that are much more powerful and/or have better price/performance than the uniprocessor x86 PC systems supported by Microsoft. Deemphasizing consulting allows competitors like Mentor to attack niches such as the telecom apps of Alcatel and Nokia. Wind River will be left only with the apps that can't justify the rapidly falling cost of an ASIC. Even that limited market will disappear soon when ChipExpress and Orbit Semiconductor give you an ASIC within 48 hours of the receipt of an HDL description.
Why is VCSI migrating to software? Because their algorithms for voice recognition no longer requires specialized hardware, and therefore, by the Law of General Processing, the migration is mandated.
The Pentium has been out for a long time, but VCSI's migration is recent. Why? Because the price of DRAM has fallen dramatically rather than anything associated with processing. When the price of DRAM collapsed, somebody wrote an article for Barron's explaining which companies would derive the most benefit. The first company on the list was Kurzweil, a voice recognition company, and the stock made a huge jump that week based solely on the story!
While it is true that application-specific hardware devices will always be invented to handle demanding applications...
This is untrue because your "Law of Specific Processing" is false. Eventually, the processor soft cores and peripheral I/O soft cores will reside in the ASIC(s), the system software will reside in flash memory, and the application software, if needed, will reside in secondary storage such as a disk drive.
Someday, we won't need any hardware at all! |
