Fun and Profit With Obsolete Computers

An anonymous reader writes "C|Net has a story about the value of aging computer hardware, and the subculture of people who collect them. The story details some of the more enthusiastic collectors currently participating in the hobby, as well as their old-school beautiful hardware. '[Sellam Ismail] recently brought a quarter century-old Xerox Star computer back to life to be used as evidence in a patent lawsuit. The pride of his collection is an Apple Lisa, one of the first computers (introduced in 1983) with a now standard graphical interface. Such items sell for more than $10,000. In an old barn in Northern California that also houses pigs, Bruce Damer, 45, keeps a collection that includes a Cray-1 supercomputer, a Xerox Alto (an early microcomputer introduced in 1973) and early Apple prototypes. '

Comments

Old DEC gear

[MichaelSmith]

...used to sum up my job. We used to get spare PDP/11 parts from people like the those in the article. The DEC maintenance guys at the time told me about a factory they knew about which relied absolutely on a PDP/8. Service calls there were a challenge, to say the least.

Towards the end of my stint at Vic Roads the foam padding stuck to the top of the slide out boxes on the 11/84's had turned to dust and collected around the base of all the mux cards where they go into the backplane. Swap out a card and spend the next couple of hours vacuming out the backplane to get it working again. Installing a SCSI card was a challenge. You slide out the CPU box and get yourself organised by lying flat on your back underneath it. Like taking the transmission out of a car. The you identify the wire wrap cable for the slot which is going to take the card and repatch the appropriate interrupt line. On some of them you were lucky, there would be little shorting patches which you could pull off, like on the back of an IDE disk. Don't muck up the backplane in the process because people need traffic lights, you know.

I've got an ohio scientific superboard 2 in my spare parts cabinet. As long as I can still find a TV which listens to an RF modulator I am free to run up the micro assembler and hack away. My son is 5 now. In 7 years he will be the same age as me when my dad built that machine up.

Re:Best for learning programming

[romiz]

To learn system programming, it is a bad deal compared to a microcontroller with an emulator, or even a refurbished GBA with a flash card:

• Interrupt handling - Check
  With only 15 interrupts lines, cascaded into 2 8-lines banks, the IBM PC is quite limited, and you still have the trouble to handle the cascaded handlers.
• Instruction timing-based optimization - Check
  But if the 8086 processor understands a subset of the complete assembly language from the current PC, the timings constraints are completely different: the cost of an instruction for a 8086 accessing directly the main memory completely changes as soon as you have cache, which is essential for modern computers. And with the mess that x86 assembly is, I'd prefer dabbling with ARM assembly instead.
• Drawing lines by directly altering video memory - Check
  OK - but it is not alone on that segment.
• Disk and memory data structures - Check
  Disk structures ? The cylinder/head/track abstraction that come with the floppy disks is compulsory on old IBM PCs. The LBA method is much more straightforward. No one should need to learn a complex, obsolete abstraction that doesn't even correspond to the reality anymore.
• And in complement to that, it is impossible to debug from the outside. With embedded platforms, you can write the code with your PC, test it in an simulator, and then test it on the platform with an In-Circuit Emulator to check for bugs. You can't do that on an old IBM PC.

Re:Best for learning programming

[arivanov]

You missed one of the GP points: instruction timing based optimisation. You cannot teach that on a modern machine (most you can no longer turn off the cache) even if you boot it in 16 bit mode. The last machine to allow this and have a well published instruction set was 286. 386SX was still useable, but the stuff started getting muddled. 386DX (all but the earliest cacheless samples) - unusable for this.

Similarly, from Pentium 3 onwards the APIC has changed drastically so the interrupt controller handling is no longer the same. Granted, you can run it in backwards compatible mode, but it is not the same.

Similarly, IO on PCI devices is clearly nowhere near the original IO on x86. While there is some backward compatibility present, you have to go and do at least some bridge programming to get anywhere. That was not the case with any of the 8 and 16 bit IO on systems all the way up to the early 486-es. You could manipulate every device separately ignoring most bus issues.

Overall, nowdays if you want to teach anything low level you have to go to a simpler architecture like one of the 32 bit MIPS architectures. x86 in its current form is too complex to be useable even for an advanced college level architecture and drivers class.

Re:Best for learning programming

[TheMoog]

In the computer games industry it still pays to know your way around cycle counts, pipelines and caches. Just because your device has a cache, and you're coding mainly in an OO language, doesn't mean to say you've left the world of cycle-level optimisation behind. And particularly on Sony machine it's almost a requirement to fully understand the various hardware interactions in order to get a decent turn of speed out of it.

As an industry we're now finding it very hard to employ people who know this kind of stuff. Most graduates are taught Java or C++ and have no decent experience at the assembler or hardware level. Now I'm not saying that we spend all day hand-crafting assembly code - games are just far too big nowadays - but every now and then you'll get an unusual crash which can only be debugged using knowledge of the hardware. In my experience CS graduates just freak out when you show them a disassembly of their code!