16

I spend a lot of time looking at hex dumps of memory, which have a format that should be familiar to most:

00010000 00 FF 48 65 6c 6c 6f 20-77 6f 72 6c 64 21 00 00 ..Hello world!..

But on 36-bit machines like the PDP-10, the natural unit of presentation is a 36-bit word. Since 36 is evenly divisible by 3 and an octal digit is 3 bits, the natural base for rendering such words was octal. What did memory dumps look like on those machines?

4
  • 5
    Raffzahn's answer on the dumper and DDT is correct. PDP-10 programmers thought in octal. For the -10 a common dump format was two octal fields representing half-words - as the address was 18-bits, which was also the size of the field in an instruction that held both addresses and constants. Two other common formats were used for text data: one that understood DEC SIXBIT, and one that understood the commonly used format of 5 7-bit characters + high bit.
    – davidbak
    Jan 13 at 18:26
  • ISTR, PDP-11 programmers also were trained to think in octal, even despite the fact that the machine used 16-bit words. Jan 14 at 14:12
  • If you look at all the old minicomputer and mainframe manuals, they were predominantly in octal.
    – cup
    Jan 14 at 15:49
  • @SolomonSlow - correct - as many of the fields in a PDP11 instruction were 3 bits - 3 bits for 8 registers, 3 bits for address mode - the instruction itself was 4 bits but at the high end of the word thereby absorbing the "extra" bit ... the PDP11 front panel (which, admittedly, many users never saw) was grouped by 3s ...
    – davidbak
    Jan 14 at 17:26
16

A PDP-10 dump is primary a plain memory dump stored in a file typed 'DMP'. To get a printout a utility like DUMP or DUMPR must be used. both offer a huge variety of options to configure what to be displayed and in which format.

Not to mention dumps being processed by DDT to get a high level view. Also, while there are similarities across DEC machines it depends a lot on the OS as well. Simply umping some values is technology of 1960. At times of PDP-10 usage the classic world was way more advanced than that, as a peek inside the TOPS10 Crash Analysis Guide may show (*1).


Beside modern tools like Lars Brinkhoffs ITS Disassembler that is :)


A typical printout of a minimal processed dump might have looked like this example from page 8-31 of the PDP-10 Timesharing Handbook:

enter image description here

(Helpful to keep in mind that real printers do at least 132 characters per line :))

Please note also Davidbak's comment about half word formated dumps.


*1 - The fact that the modern PC wold has fallen way back behind what already was standard is a complete different story.

8
  • I'm looking for a "common" or "default" format for these dumps. What would a debugger write when ask to display the contents of an area of memory? Jan 13 at 17:47
  • 1
    Under TOPS-10, "the debugger" would be DDT or one of its variants. And you'd get the dump format you asked for, which as far as I recall, depended on the command you typed. There might have been a command to set the default radix. Jan 13 at 17:56
  • 3
    DDT-10 manual Jan 13 at 18:08
  • 2
    Under RSX-11M, I rarely needed 'live debugging', but crash dump analysis was another thing entirely. It was common to extend dump analyzers to interpret data structures in some useful form, not just get swathes of octal. Jan 13 at 18:12
  • 3
    RSX-11M Crash Dump Analyzer. Though I was more prone to use NDA, the (DECnet) network dump analyzer Jan 13 at 18:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.