Mastodawn

Amin Girasol Aug 24, 2025

Sean Haas @adventofcomputing has a new podcast episode out - the first of three about the #DEC #PDP11. In the first episode, he covers the evolution of the design of the PDP-11, from its orthogonal instruction set to the innovation of #Unibus.

Part two will cover #LSI11 and the third will discuss applications and derivatives of these chips.

Subscriber-only for now: https://www.patreon.com/posts/episode-163-ever-137192301

...but it'll show up on the open RSS feed soon enough: https://adventofcomputing.com

Sean's not active on the Fediverse but his output is consistently fresh and interesting - you never get the old regurgitated stories. He ought to get more attention.

#vintagecomputing #retrocomputing #computerhistory

Mark J. Blair Jul 22, 2024

Is this a good name for a DEC Flip-Chip backplane breakout board?
#DEC #FlipChip #OMNIBUS #QBUS #UNIBUS

Tom Lyon ✅Feb 3, 2024

PDP-11 factory in Galway. Note the long ribbon running across the racks - it's the #UNIBUS - main system bus! Asynchronous design, so no timing limitations, just transmission ones. [From FB/PDP-11]

GripNews Oct 27, 2023

🌘 關於Unibus和QBUS的巴士仲裁
➤ Unibus和QBUS上的巴士仲裁問題及解決方案
✤ http://www.froghouse.org/~dab/papers/bus-arbitration/bus-arbitration.html
本文探討了在Unibus和QBUS上的巴士仲裁問題，並提供瞭解決方案。作者發現在Verilog代碼中存在競爭條件，這引起了他們對Unibus和QBUS設計中的巴士仲裁失敗的問題進行研究和解決的努力。本文旨在為後來者提供有關這個問題的結果。作者認為這個問題在DEC社區內口耳相傳，直到現在還沒有被書面記錄下來。
+ 這篇文章提供了有關Unibus和QBUS巴士仲裁問題的重要信息，對於研究這個領域的人來說非常有價值。
+ 很高興看到有人將這個問題寫下來，這將對未來的研究和開發工作有所幫助。
#巴士仲裁 #Unibus #QBUS

On Bus Arbitration on the Unibus and QBUS

Stephen Hoffman Apr 13, 2023

@industrialartifact everybody makes mistakes here. Had fun with memory-space bus games between a group of drivers of mine, and a driver that the DEC VMS group wrote.

Being the new drivers in the testing matrix mix, mine got blamed for an incompatibility found in testing. On investigation, my drivers were playing by the published bus memory allocation rules, and the VMS group driver… wasn’t.

Spent a lot of time writing doc for DEC module installs and troubleshooting, too. My earliest efforts there were, well, my effort there were less than entirely successful.

Bus grants and serpentine wiring tripped many. Writers, readers, field service, etc.

Carried a wire-wrap tool into the mid 1990s, just for BA11 boxes.

Much prefer PCI and PCIe, there.

PS: if you have enough “spare” UNIBUS grant modules (G7273?), use them, and remove all the grant jumpers on all the BA11 slots. Makes UNIBUS work easier.

BTW: if you don’t already have a copy of the field guide stashed somewhere:

http://www.ibiblio.org/pub/academic/computer-science/history/pdp-11/hardware/field-guide.txt

That may not be the “newest” of that field guide in existence, though.

#retrocomputing #retrocomputers #VAX #DigitalEquipmentCorporation #UNIBUS #DEC

Show thread

Stephen Hoffman Feb 10, 2023

@cvwise #UNIBUS, Q-Bus, and ilk were parallel, so less than fun to extend. Serial buses, of course, are somewhat easier.

Not that UNIBUS enclosures didn’t get daisy-chained more than they were probably spec’d for.

Had an #Adage #Ikonas graphics adaptor that would consume the whole of the VAX-11/750 UNIBUS for several minutes, which meant I kept having to politely prevent #DEC field service from swapping out what they thought was a failing DSA #UDA50 disk controller. The UDA50 was fine, it just lost access to the host #VAX for way too long, and the firmware triggered an error report and its recovery processing.

That Adage had other issues, too. The vendor driver had a subtle bug in it. The device driver setup routine didn’t initialize a VAX hardware register before setting some bits in the register (whoopsies) and loading it into the device control register on the Ikonas board.

On #VAX / #VMS V3.x, the low bit in the hardware register was accidentally always clear, due to some upstream code. On V4.x, the upstream code changed and was more variable, and the register was more often odd. Odd (low bit set) was the “go” bit in the device hardware register.

When the “go” bit was set, the board would then do a (effectively harmless) DMA write from VAX host memory out into the board, or a DMA read into host memory. A read which was less than harmless, depending on what part of host memory got clobbered by the read. Some part of DCL or the exec or who-knows-what-else would get clobbered. If it was free host memory or unused memory overwritten, nothing (visible) would happen. If not, well, *something* would tip over. Right away. Or eventually.

Found and fixed the driver bug after some rummaging, and more than a little colorful rich and language.

Now for the fun part: some years later, half a continent away, at a trade show, while now working at a completely different job fir a different employer, somebody showed up in the booth complaining to DEC about the Adage Ikonas board. Crashing.

It was their lucky day.

I explained to them the nature of the bug, exactly what the driver bug was located, and how to find and fix it.

They were… not expecting to get a hardware register-level answer to their third-party vendor complaint from an unrelated vendor.

Sometimes things work out.