Talk:PDP-8/S

From Computer History Wiki
Jump to: navigation, search

Clock speed

"Computer Structures: Readings and Examples", page 120: "A PDP-S/S is one-fifteenth of a PDP-8 at one-half the cost. ...even though the same 2-megahertz logic clock is used in both cases." Nczempin (talk) 17:32, 18 August 2019

It's probably best not to rely on secondary sources like that, when real manuals are available - because in this case, they got it wrong. From the "PDP-8/S Maintenance Manual" (F-875, August 1969), "Word Times" (Section 2.2, pg. 2.7):
Each word time is made up of 14 bit times .. during which the clock generates a string of fourteen bit pulses 750 ns apart.
The basic bit clock is therefor 1.333 MHz, whereas the word clock is 95kHz.
There can be up to 6 'word times' during the execution of an instruction, or as little as 3 ("fetch", "execute", "end"), depending on the type of instruction, and what else is going on in the machine ("break" word time, for interrupts).
There could be an interesting discussion about what "clock speed" should refer to; Template:Infobox Machine says it is "CPU cycle time", which for the -8/S is probably the 'word time'. Since it's a bit-serial machine, probably best to give both. Jnc (talk) 15:36, 19 August 2019 (CEST)
Intriguing... I assume you are aware that the book I referred to is not by some random chumps; but of course they could have been wrong. I wonder then if the info on the straight 8 is also wrong, given that they're both supposed to have the same clock --Nczempin (talk) 21:27, 19 August 2019 (CEST)
Well, Bell didn't do the PDP-8, he was the PDP-6 and PDP-11; the PDP-8 was Edson DeCastro, later of DG. So he's probably not familiar with the details, right off the top of his head.
Yes, the straight-8 is also wrong, per its Maint Manual. I'll write up what I found later (once I fully understand it). Jnc (talk) 22:34, 19 August 2019 (CEST)
Okay, I've found something: This clock board can go up to 2 MHz, which doesn't mean that it does: [1]. And I'm assuming they were referring to the board by "the same 2-MHz clock". So it's basically a potentially 2 MHz clock that's actually run at 1.33 MHz, as far as I understand. I would argue that talking about 2 MHz is a bit misleading then, though. [2] [3]--Nczempin (talk) 23:19, 19 August 2019 (CEST)