User:Jtp15

jtp15

Some of my areas of interest are computer science and related areas, such as artificial intelligence and philosopy, especially the philosophies of logic, and the various tools that are used to employ them. These tools include computers and networks.

I have an M.S. in Computer Science, and a B.S. in Electrical Engineering. I started consulting in 1977 and have been putting food on the table by developing software and systems ever since. I joined Mensa as a teenager, but got bored with it rather quickly, I don't enjoy discussions about how smart we all are; rather, I enjoy discussions about what smart is.

I enjoy history, religion, philosophy, etymology and alternative energy, but mostly set theory and quantum states are among my many eclectic studies. I tend towards existentialism, but I also lean towards the "Big Bang" theory as causal. One of my prescriptions says "take 1 pill as needed for mania." If this means that I am a certified maniac, I am in denial, and repress it under a veil of wit.

As far as binary machines go, I am of the school of thought that we have reached the point where the circuits are small enough and spaced closely enough together that external forces, electro-magnetic radiation, electro-mechanical irregularities and stray energy from static or power supply discrepancies, combine to result in the dreaded "computer error." Add over-clocking, and you will get differences in one WORD (32 bit) rounding on symmetrically built computers.

In 1996, after building a two-tiered client-server-based financial system, in which the back-end performed the business logic and returned a consistent result set, my users showed me minor differences in the results on their work stations. This can not be, I thought. I ran tests on machines I had ordered in a lot of twenty from a major micro-computer manufacturer that had the exact same build specifications. The lot even had sequential serial numbers, though of course not the components.

After running many software tests, I ended up coding assembly language that ran directly on top of the BIOS (a mini-OS). I put five of the workstations in a controlled environment to stabilize entropy. Of course, there still could have been variations in the BIOS and the PROMs on the motherboard but, for practical purposes, the conditions were far more stable than a typical PC in a cubicle.

I wrote a few simple rounding routines, each of which used hard-coded constants, 32-bit WORDs of rational and irrational infinites. Also an n-depth stack that was n-depth wide, where n was MAX(WORD). The result was a bicubic array, x=y=n. After y gigaflops, the registers were read and I found that two of the machines were not in the same static state 23 percent of the time. In real-world working conditions, the accuracy was even less.

Thus, I will no longer blame the software for all errors. Though I never measured the frequency of incidence in more tightly controlled lab conditions, and did not narrow the results down to specific components or ROMs. My hypothesis is that even a HAL 9000 series is subject to a stray photon or two.

As you can see from this last blurb, I am cautious of absolutes. You might even call them one of my pet peeves.

So, except for a few minor idiosyncrasies, I would love to be of help with articles when I have the time or the need is great.

J. Ted Parker