5

Information online states that a 18/36-bit DECtape holds 576 blocks. People claim it may be possible to add a few more blocks, but no more. Where does this number come from? Is there a hard limit on the length of the tape?

Lars Brinkhoff
  • 3,111
  • 16
  • 35
  • 1
    There's certainly a limit on how much tape can fit on one spool... given that the spools have a standard size. I do not know if there was wiggle room to replace the tape with a slightly longer tape, and still have it fit on the spool. – dirkt Aug 28 '20 at 09:49
  • Can formatting be done "one the machine", and is that a hardware function? – dave Aug 28 '20 at 12:45
  • @another-dave not sure what you mean, but e.g. on the PDP-8, a special program was used to format the type, and the hardware interface is generic, it doesn't have an opaque "format" function. So in principle, you can change block count, block length, etc. – dirkt Aug 28 '20 at 18:58
  • 1
    A PDP-10 could format a DECtape. There was a manual switch labelled WRTM that had to be set to on before formatting could be done. WRTM stood for something like Write Timing and Mark track. Normally off. – Walter Mitty Aug 28 '20 at 20:56
  • Ah, I couldn't remember the details, and in particular I wasn't sure whether the timing marks could be written "on site". Apparently they can. PDP-8 info here – dave Aug 28 '20 at 21:00
  • OT: I searched in Bing for "PDP11 dectape" and it came up with "images of PDP11 Duct Tape". Sigh... – dave Aug 28 '20 at 22:33
  • @another-dave, low level access was possible from a PDP-10. The MIT people played tricks with the inter-block words to make a magic tape bootable from a PDP-6 (which didn't have hardware read-in). – Lars Brinkhoff Aug 29 '20 at 07:12
  • Maybe @WalterMitty may even remember some of that magic, since the Dynamic Modeling group relied heavily on DECtapes in the first years before getting an RP10 disk controller. – Lars Brinkhoff Aug 29 '20 at 07:13
  • Per this TC11 manual the block count was 578 on PDP-9, -10, -11, -15, and maybe -8 (the wording is unclear to me; maybe it was only the data-within-block that was different for the -8) – dave Aug 29 '20 at 15:37
  • By the time I joined Dynamic Modeling, they had disks. They were still loading the operating system from DECtape, or maybe loading the bootstrap from DECtape. My experience with the WRTM switch comes from earlier work at Computility, a timesharing retailer. – Walter Mitty Aug 29 '20 at 16:34
  • @another-dave http://www.bitsavers.org/pdf/dec/dectape/551/H-551_551dectapeCtlr.pdf says 1100₈, so apparently the number varies slightly. It also says there are two dummy blocks, one at the start and one at the end. That would make the total 578. – Lars Brinkhoff Aug 29 '20 at 19:11
  • Ah, right. The TC11 manual is a 'hardware' manual, and hardware guys always see things differently :-) – dave Aug 29 '20 at 19:23
  • Hey, using Duct Tape to store bits! Talk about persistent storage! – Walter Mitty Aug 29 '20 at 23:34
  • We used DECTape a lot at Computility in 1969. We only had a very small disk for all our users. Long term file storage was all on DECTape. One day, after an operator error we ended up where a DECtape had its directory block overwritten by the directory block of a different tape. Disaster! I took on the task of recovering the work on that tape. I wrote the equivalent of an "undelete" program in 24 hours, elapsed time. That's fast even by today's standards. I had been reading up on the on tape data structures for weeks, so I was primed for this challenge. – Walter Mitty Aug 29 '20 at 23:38
  • The PDP-6 didn't have hardware read in, but it did have 16 words of 36 bits each "hiding behind the accumulators" (assuming you had bought the fast accumulators option). These 16 words, in core memory, would survive even a power failure. A read in program for paper tape could be jammed into those 16 words. I think the same might have been done for DECTape. – Walter Mitty Aug 29 '20 at 23:43
  • @WalterMitty, at MIT there was a DECtape read-in for the PDP-6. It was called the "shadow loader" because it was shadowed behind the accumulators. https://github.com/PDP-10/its/pull/1030#issuecomment-400595814 – Lars Brinkhoff Aug 30 '20 at 07:04
  • Yes, that's what I was guessing. I should have said "may have been done". – Walter Mitty Aug 30 '20 at 10:17

1 Answers1

10

The standard tape reels had a capacity of 260 feet of tape. Standard tape thickness was 1.25 mil, so it would have theoretically been possible to put more tape on a reel if you made it thinner (which was hard to do because thinner tape was typically also stretchier). Making the reel larger wasn't a good option because the hubs on the drives were too close together to allow them to be much larger than they were already, and you would have needed two larger reels, one for the tape and a second one to use as the take-up reel.

The reason you could fit a few extra blocks on the tape was because both ends had a lot of leader tape so that you could securely wind it onto the take-up reel. You could therefore squeeze a few more blocks on the tape if you were willing to (a) reformat the tape, and (b) have a slightly increased risk of tape not winding properly because there wasn't enough on the take-up reel to hold securely.

I believe the 'hard' limit on the number of blocks would have been 4096 because block numbers were recorded as 12-bit values. This is assuming there's nothing in the electronics that would have limited it to something smaller, or which treated block numbers as signed values, either of which would have reduced the maximum number of blocks.

The operating system or software in use might also have imposed a limit, for example any system that treated the tape as a slow disk might assume the default number of blocks, and therefore be unable to use the extra blocks without modifications.

Ken Gober
  • 11,427
  • 1
  • 40
  • 56
  • I'm not sure you could have made the tape (much) thinner. Unlike vacuum-column drives DECtape drives directly pulled the tape over the head (in both directions), tugging on the other side. Had to be taut to keep the tape firmly on the heads, and there was no slack in vacuum columns, so the "rear" hub couldn't spin free - it had to provide tension. If you didn't want the tape to stretch - or even break - it needed to be pretty tough. (Plus, those of us who hung DECtapes on drives weren't really delicate about it ...) – davidbak Aug 28 '20 at 17:34
  • The two layers of Mylar which had the magnetic medium sandwiched between them also gave less opportunity to reduce the thickness. Did wonders for reliability however. – Brian Aug 28 '20 at 18:29