Thompson’s repeated assertion that my book is a waste of time isn’t shared by everyone. I don’t think it’s necessary to point out his self-contradictions, and it’s not necessary to state that his experience with one computer is statistically insignificant compared to my experience with hundreds. That said, I’ll just post some mail and shut up. I’ve got other business to tend to tonight. Friends who need things, plans to make. I’ll be back tomorrow with some Linux stuff.
From: Dan Bowman
…and that’s a nice piece on the old machines. I think I’ll have to say something about my work P-200 and Athena here at home,
You can read his bit here.
From: Steve DeLassus
Nice post for Monday. I have to slightly disagree that MS doesn’t believe in programming for performance. They’ll program in just enough to keep the common user hooked and not try an alternate OS. They don’t want you to think that their new OS just ate up all of the “gains” you made by purchasing a CPU with more cycles (“more cycles = faster computer, right?”). The problem here is consumer education, and most consumers relate more raw compute power to better performance, not realizing that better *system design* means better performance.
From: John Lowell
About a year ago, I e-mailed you introducing myself, attempting to engage you in a dialogue about something about which you seemed quite expert at the time. I never received even the courtesy of a reply. Looking at Bob Thompson’s website today, it would seem that one would have to count silence from you as a blessing.
For the record, I don’t remember Mr. Lowell’s mail. But considering the volumes of mail I sometimes receive, it’s not uncommon for one to fall through the cracks.
You can’t please everyone.
I enjoyed your commentary. I’m the MIS at a joint that would still using IBM PS2 computers if we could get them to use Office 2000.
So, we do what we do with what we have. Robert, as well as Jerry, don’t have any problems getting the free stuff; ours cost us. So we build our own units and upgrade until the motherboards burn out.
Keep up the good info.
Received from Steve DeLassus, via e-mail. (Steve is an electrical engineer-turned software developer, so he *KNOWS* this stuff.)
CPUs have been overkill (for the most part) from the beginning. Slow I/O is the ultimate bottleneck. Slow hard drives, slow memory, slow buses – they all play a BIG part in how much data the CPU can be fed. Caches aren’t helping much anymore. There needs to be a breakthrough in storage technology for computers to really truely utilize the CPU. Probably a non-mechanical, high-density mechanism.
Some defense of the software folks: it’s hard to make hardware play nice. But there is plenty of literature out there about how higher level software *should* act. [A] decked-out machine is still going to crawl if he uses software written by monkeys. Any idiot can write an inefficient algorithm
that’ll bring a computer to its knees.
And to that I should add the Story of Mel:
Real Programmers Write in Machine Code