Joined: Fri Aug 30, 2002 1:09 am
Posts: 8521
Location: Southern California
|
Welcome to the forum. I'm glad to see you took the time to read the whole thread—which is why it's kept.
Quote: There is a motivation to make a "be all end all" product that "suits all needs". Everyone has different wants and desires out of something like this. The problem is that once you make an all powerful system, you end up with something vast and complicated like the PC is today.
One of the challenges for those who really want it to be the "be all, end all" product is to do that without creating another PC mess. Frankly I think you were waaaaay too nice to the PC industry and most especially Microsoft. I just found out that Linux-based Lindows is becoming available. You can see them at http://www.lindows.com. It must be awfully hard to get such a company off the ground with the ugly green giant standing over you, but Lindows has in its favor all the people who are sick of MS's programming inefficiency and bugs, their worse-than-lousy support, the way they've taken away user control, the time and expense of PC maintenance, MS big-brother tactics, the way they've blackmailed software retailers and so on. I have very little good to say about MS. Lindows, OTOH, is based on the open-source Linux, and is committed to avoiding the MS problems I just mentioned. I understand their system is much more reliable and their customer support is outstanding.
[Edit, many years later:] MS demanded that Lindows (from "Linux windows," and certainly there were windows before MS came out with Windows) change its name. IIRC, MS lost in court, but paid Lindows to change its name, which they did, to Linspire. I bought a new PC with Linspire installed, and no MS Windows. It was wonderful. Everything just worked. Linspire had its Click-n-Run software library too where almost all the software was free and it was much easier to install than Windows software is. After a while though, Xandros bought Linspire, then dropped support. About that time, Ubuntu seemed to be emerging as the leader in desktop Linux, so I got that (again, free). That was great too, although I don't like what they've been doing in the last few years, and I'm thinking I ought to try Linux Mint.]
Quote: The PC is truly a generic machine that can do many things well
...when the PnP works right, when the computer isn't crashing because of MS bugs, etc..
Quote: But for the user experience, it is difficult for its generic nature to not show through. For example, it has long boot times
...due to programming inefficiency
Quote: so that it can figure out what hardware is in the system
My HP-71 hand-held computer had PnP 10 years before they started talking about it in PCs, and it always works right. In my case for example, within the first fraction of a second after turn-on, it already knows which modules I have plugged in, how much memory of each kind there is, what parts of memory I have freeported for independent file chains, it has checked the integrity of the file chains, and it knows that the interface loop is plugged in. It issues the automatic device-addressing command on the loop (which is a serial, more-intelligent implementation of IEEE-488). If there are no devices out there, it times out in two seconds. If there are devices plugged in and turned on, it will almost instantly know how many disc drives, printers, test bench instruments, display devices, modems, interface converters, etc are there and what the basic capabilities of each one are, and has assigned each one an address. Maximum number of peripherals with simple addressing is 30. If you use extended addressing, it's 930. Then it checks to see if any alarms came due, either action-type alarms that could not be carried out because the hardware was not left connected when they were due, or message alarms you didn't acknowledge because you weren't near the computer to hear it beeping at the time. How much this thing does in the first second or two with a 700kHz clock speed puts the PC to shame!
Quote: Then it must run through a generic procedure to locate code to make the system actually usable.
Why not put the code to run these things in ROM on the various peripherals themselves? When my hard disc on this computer went down this summer, the shop could not re-install the sound board driver software because the company they expected to be able to get it from over the internet was out of business.
Quote: It's almost impossible to take a stock "white box" machine and make it "instant on".
My HP-71 is almost instant-on, but as you say below, the code is always there—some in ROM, some in RAM—not having to be loaded from a disc.
Quote: And all of that flexibility leads to insane complexity trying to get the disparate pieces to work with each other socialably.
It is indeed extremely complex; but the same thing could be done with a small fraction as much memory, loading time, and execution time if the programming were not so wasteful. That's where we would hope to do much better.
Quote: The more generic you make a system, the longer it takes to stabilize and become usable. Making a generic platform is a waste of time, we already have a decent one (yea, there are Issues with the PC, but it's an amazing platform today).
I think the biggest obstacle to making it generic is that by definition, you have to be able to look ahead and come up with a system where you don't keep finding that you've painted yourself into a corner and can't do something else you just thought of without redesigning. The PC is still not well suited to real-time work, in spite of astronomical clock speeds.
Quote: This is the motivation for having a language in ROM. It makes the machine readily usable. You don't have to wait for it to load from a disk. You don't even need a disk.
And having things like this in ROM protects it from virii and dumb mistakes that could wipe it all out. Even if it's in RAM, you'd get some degree of protection if you leave the option to do a hardware write-protect (via a DIP switch, for example) on different portions of it.
Quote: If anything, I like the idea of some power friendly non-volatile storage, simply battery backed up RAM.
I'm sure the value of battery-backed RAM is underestimated by those who have never seen a nice operating system make use of it. I normally have scores of files in my HP-71 all the time, ready to run without having to "load" them into another part of RAM or any of that. For this reason, I seldom need to connect to a mass-storage device. Beyond just having all your files right there however, it's better than just having the OS in ROM, because things like user preferences are retained in OS variables and flags.
Quote: Instant on and off is SUCH a wonderful feature
An editorial in one of the electronics industry magazines said that's an area where calculators have an advantage that the PC will have a hard time overcoming. Engineers, businessmen, and others will always reach for a calculator to do certain things when they have a PC right there. Even very sophisticated ones won't keep you waiting for something to load. Just turn it on and go.
Quote: So, if anything, today I think it would be nice to have something that is a cross between the simple machine of yesterday with the pervasive, instantaneous abilities of modern palmtops and PDAs, but in a desktop format. If we're dreaming, let's dream of a desktop Newton.
How 'bout mixing some newer technology (start with a 16MHz 65816 with megabytes of 15ns memory) and the care with which software used to be written back when programmers could not afford to waste memory and processor speed. I was "underwhelmed" however by the Newton. The technology to recognize handwriting (when it works) is impressive— but I thought, "Somebody has forgotten that we used to use typewriters before we had computers because typing was faster than writing by hand—a lot faster!"
Quote: Regarding video, how is that done? If you put aside the hi tech accelerated graphics cards, what do you need to drive a VGA monitor? I guess for electronic savvy folks, this isn't that hard of a task, but for naive folks like me, I haven't a clue.
Even there, Samuel Falvo (kc5tja) here has some good information on what is done with video, the fact that the PC industry standardized on certain GUI factors before studies showed that those high-overhead factors did not increase operator efficiency.
Quote: With sound, the original Mac had little more than a D/A converter for its sound, correct? vs the specialized chips of the Atari and Commodore. The chips are nice to off load the processor, and for perhaps as MIDI instruments.
I've heard some very impressive music that was generated on Commodore 64's. You'd never know from the sound of it that it came from a C64. Offloading jobs like sound synthesis certainly leaves more processing power for other things, but the 6502's interrupt response is such a strong point of this processor that you can do fast, high-fidelity audio sampling just on interrupts with minimal hardware and still be able to do plenty of other things at the same time. I'm trying to collect information on various processors' interrupt performance. It has been hard to get complete info. Howard Spiegel commented that he suspects many of the manufacturers try to partially hide the numbers because their processors' interrupt performance is so bad. I have some information on each of 28 processors—mostly 8-bit—and so far, not a single one of them gives a modern WDC 65c02 any serious competition. I have complete information on very few processors though. If anyone can direct me to appropriate web pages, please do.
Quote: I'm not much of a sound guy, relying mostly on MP3s and such which seem to simply be D/A streams, so I don't really know what a modern "sound card" actually does today.
MP3 is a lossy data-compression algorithm. It allows recording music with less than 10% of the memory space that would be required to simply lay all the actual sample values down in memory. With good speakers or headphones, you can definitely tell that the audio quality has been degraded by the MP3 process. I think the .wav files are probably just uncompressed streams of samples to feed to a D/A converter. I'm sure someone else on the forum knows.
Quote: (Someone mentioned Pygmy FORTH earlier. That's a system that is almost self hosting, in that it's written in itself -- it just relies on DOS at the moment.)
I haven't used Pygmy Forth, but I am a strong proponent of Forth, for many reasons. It makes the programmer highly productive. Even the compiler is completely open to be extended and modified by the programmer. Right from the beginning of Forth, it has had the tools to do OOP, even before it was called OOP. Beyond being just a language, it's also a programming environment and sometimes the operating system too. It is extremely frugal with memory. It has gotten kind of a bad wrap partly because of a number of very poorly written, free, public-domain Forth kernels that have been circulated over the years; but even those poorly-written ones generally had the tools to make improvements to the kernel itself for anyone who wanted to put in the effort. That cannot be said for other languages. One person here on the forum is porting my 65816 Forth to the 65265.
_________________ http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?
|
|
Login
Register
FAQ
Search
But it resulted in our