I have one, but you're right it;s not common. It's a transputer, RAM and a special vector processor chip, not a lot more I don;t know the B012. I assume it has some kind of host interface circuity (look for C011 or C012 chipss) and maybe some link switches (C004 chips). I got started at an ever lower level back in 1989, startign with bare chips and wire-wrapping them. I found it a remarkably easy chip to use. On the otehr hand, I was lucky enough to have acess to a microvax with a link afapter card and all th software isntalled, so I have had to do battle with that part. Ah... Are you saying that you don't ahve any form of link adapter. That is a mahor problem. I would trty to find something -- IIRC a B011 is a VME card with a transputer nad link adapter on it, there's the B004 for the ISA bus, etc. You cna then link the RRAMs together and hang it off the host card. Nomrally a transputer runs a bootstrap in internal ROM which downloads a program from the host macvhine over one of the links. It is possible to cofigure a transputer to start running a program in external ROM, and I guess it would be possible to get it to communicate with a host using a more normal serial or parallel interface (in fact the very first INMOS trapsuptr cards used in the ITEM were developped before the link adapter chips and have normal RS232 ports on them (!)). But it's not a trivial project, particualrly if you've never worked with transptuers before. My sgugestions is that you look out for some kind of host card with a link adapter on it. That will get you started. -tony

Is the B012 jut a TRAM motherboard then, taking power form the VME bus and bringing the links out on (I asusme) the B connector? I've never used one. BAsaiclaly what I said... Hang on... Are you now saying the B012 is a host adapter? Thinking back, the B011 was a VME cars with a trnasputer on it, but I think it was normally used if you wanted to make a VME ssytem with the transputer as the main processor. The VME bus was interfaces to the transputer memory bus and you could drive normal VME cards with it. I've done a fair amount with the raw chips, but very little with normal commerical hardware like TRAMs and their motherboards. -tony

As you know, VME boards are standard double Eurocard form factor. it makes snese to use soemthing like that. IIRC, the B011 has a noraml VME interface on the 'A' (upper) connector but the 'B' connector carried the transputer links. The C004 is a bit like a 'telephone exchange' for trnasputer links :-). It lets you reconfigure how the trnasputers are interconencted. The interface on said chjip to load the conenciton data is ... a transputer link. Makea lot of sense, I guess... You _need_ a host adapter.... -tony

While Sun used triple eurocards on a VME bus, I didn't think that was officially part of the standard. Theyr'e rarer than transputers (and link adapters) ;-(. You didn't need them if you were making a fixed-configuration network after all. While the C004 did re-time the link signals, the links are not that fast (10MHz), and you can get away with using F TTL multiplexers, etc to swich links around on smaller systems. Yes, I have done it. -tony

ISTR someone seeking "inexpensive" 8" bernoulli cartridges a while back.... check EPay #150609014843. $5 starting bid. seller says he has several steve On 5/27/2011 11:16 AM, Dave McGuire wrote:

The C011 is very easy to interface to a stnadard microprocessor bus, all you really need to add is an address decoder. IIRC, the internal registers show some similarity with a DL11 (of all things!) Do you tink an original ITEM would have any financial value? (No, I do not have one for sale). -tony

Why? Transputers are not all that super rare, and like most early MPP architectures, really did not work all that well, as was originally intended. -- Will

In article <BANLkTik9qq5t-Y0z0ugs4dmfU3+1btabvQ at mail.gmail.com>, William Donzelli <wdonzelli at gmail.com> writes: For the same reason anyone implements any other CPU in an FPGA -- so you can use that CPU architecture in a modern design or as a hardware simulation of a vintage design. 6502s and 6800s are even more common and both of those are implemented as FPGA cores. I think there's a difference between commercial success and not working well. "Not working well" implies to me that it had some functional deficiency. Transputers ended up not being commercially successful because they were expensive and not many people wanted to program in Occam, despite whatever advantages CSP may have had. -- "The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download <http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/> Legalize Adulthood! <http://legalizeadulthood.wordpress.com>

On 5/29/11 10:20 AM, William Donzelli wrote: [reminds self to send Will a book about array processor design] -- Dave McGuire Port Charlotte, FL

In article <4DE209C6.24926.7919F at cclist.sydex.com>, "Chuck Guzis" <cclist at sydex.com> writes: Are you talking about the ES-1? Exactly the opposite case could be made; they sold off all their productive business units and kept the underperforming ones. -- "The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download <http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/> Legalize Adulthood! <http://legalizeadulthood.wordpress.com>

In article <4DE24CF0.13079.10DED41 at cclist.sydex.com>, "Chuck Guzis" <cclist at sydex.com> writes: ES-1 had a very narrow market window. The design functioned properly at lower clock rates, but one of the chips essentially melted when they ran it at the specified clock rate. The design was non-competitive at lower clock rates. E&S is just one of many companies that missed a transition to smaller devices. E&S never really adapted to the transition to workstations; when a workstation was proposed inside the company, it was said that David Evans pronounced workstations "a fad". When E&S finally did make a workstation in 1988/1989, it was too late to enter as a distinct competitor. They later tried to make graphics adapters for other workstations (Sun, HP, IBM) and had some success there, but it was again too late. Even with acquisition of PC accelerator companies, they were not able to compete in the PC world effectively, either. They were relegated to high-end simulation systems but even there SGI machines were stealing away market share from them. E&S had great engineers but amazingly bad marketing and management. IMO, they never should have embarked on ES-1 to begin with. While late with their workstation effort, they did manage to prove that PHIGS/PEX could be implemented with good performance. While not the only deciding factor, this probably redoubled the efforts of SGI to create OpenGL as a defacto standard instead of the industry settling on PHIGS/PHIGS-PLUS which is an ISO standard. -- "The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download <http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/> Legalize Adulthood! <http://legalizeadulthood.wordpress.com>

Massive systems. Remember when we were all supposed to have oodles of little processors we could throw at any problem? -- Will

On 29 May 2011 at 16:36, Dave McGuire wrote: I think (but can't remember for certain) that it was Neil Lincoln who compared massively parallel processors to using nine women to get a baby in one month. Some problems just don't parallelize well... --Chuck

On Sunday (05/29/2011 at 01:59PM -0700), Chuck Guzis wrote: I know I sure heard him say that a lot! But there are probably not enough of those kinds of problems to build a computer business around anymore. Not one that makes a profit anyway. Chris -- Chris Elmquist

On 5/29/11 6:00 PM, Richard wrote: There were plenty of array machines during those years; they've been around pretty much constantly since well before the Transputer era. It's not as if it's a new thing. -Dave -- Dave McGuire Port Charlotte, FL

The idea is old as dirt. I am talking about the mid-1980s, when microprocessors were cheap and fast enough to be assembled into a massive system and really do well in the price/performance area. Sure, you could have made an MPP system with some cheap minicomputers like PDP-8/Fs, but the beancounters would start screaming. And I bet it would have sucked. -- Will

Well, yes. A lot of early attempts at *anything* kind of sucked. We are now 20 odd years past the MPP "boom" of the mid 80s - I certainly hope technology would have marched on during those years. -- Will

In article <BANLkTimUUujMnXhvFqqq1S+-wriZDbATDA at mail.gmail.com>, William Donzelli <wdonzelli at gmail.com> writes: The architectures aren't substantially different; the component density and communication latencies/bandwidth are substantially different. -- "The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download <http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/> Legalize Adulthood! <http://legalizeadulthood.wordpress.com>

On 5/29/11 7:05 PM, William Donzelli wrote: Interesting. When in fact the Transputer's networking was significantly better than just about anything else that had been built by that time. -Dave -- Dave McGuire Port Charlotte, FL

And I am still saying that because the networking between the processors was not good, performance sucked for anything that involved a large number of processors with problems involving lots of message passing between nodes. This was the flaw of pretty much all of those early commercial MPP systems - regardless of the topology. You could have a fully interconnected crossbar for all the nodes of an MPP system, but if that crossbar is crap, then the whole system is going to be crap. "Wrong time/implementation" pretty much equals "did not work all that well", as I see it, regardless of how things have turned out today. -- Will

On 28 May 2011 22:13, Dave McGuire <mcguire at neurotica.com> wrote: Oh, cool, I did not know that. XMOS make the coprocessor in the new (not yet commercially-available) AmigaOne X1000. I've seen & played with the prototype - it's a very appealing machine and in the form of AmigaOS 4.1 with a pleasingly small, efficient and very fast OS, plus a respectable library of applications (OK, most running under emulation, but hey) /and/ a choice of decent modern Linux ports. http://www.a-eon.com/ I hadn't realised the XCore XS1-L1 sort-of-RISCy sort-of-DSP-like coprocessor of undefined functionality was in any way Transputer-related. It doesn't make it any more use, but it makes it cooler. :?) -- Liam Proven ? Info & profile: http://www.google.com/profiles/lproven Email: lproven at cix.co.uk ? GMail/GoogleTalk/Orkut: lproven at gmail.com Tel: +44 20-8685-0498 ? Cell: +44 7939-087884 ? Fax: + 44 870-9151419 AIM/Yahoo/Skype: liamproven ? MSN: lproven at hotmail.com ? ICQ: 73187508

On Thu, 26 May 2011 17:15:39 -0400 Dave McGuire <mcguire at neurotica.com> wrote: [...] Well. This could be arranged. If it is OK to you that those chips are soldered on said B012 board... There may be even some TRAMs sitting on top of that B012... -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

When I was at Bristol Universtiy, we did quite abit with Trasnputers. We had a microVac with a 3rd party trasputer link card (not Inmos) in it, the developement tools (Occam compiler, etc) were hosted on said VAX. I was building my own trasnputer-based hardware, so had little to do with any commercial boards. But I rememebr a B001 board being around, and I am pretty sure it ws used as the bus master in a VME crate with varios interfaces, etc, hung off it. Whether the B011 could also be used as a VME slave to hang trasnputers off anotehr VME processor I don't know. To the OP... If you feel up to it uou can get around the problem of not having a C011 or C012 link adapter chip. You need to take a 'spare' transputer chip and configure it to run from ezternal ROM (see the data sheets). And have some ROM whcih reads data from a host interfce board (also mapped into the transputer memory space) and sents it to the transputer's link hardare. You then use said links to talk to your other trasnputers. I am pretty sure some commercial host itnerfaces worked that way. I never saw a schematic of the Q-bus card we had at Bristol, and I wasn't left alone with it for long enough to reverse-engineer it :-). But IIRC it contaisn a T2, some EPROMs and some glue logic (and probably the normal Qbus interace chips, etc). I am pretty sure there was no C011 or C012 on there. But making one yourself may be more than you want to get involved with, at least until you've got some transptuer experience... -tony

We had at lest 2 of them, differnt makes, but I think fairly similar circuitry. I cna;t rememebr any more details, I used it about 20 years ago after all... Pity it didn't get to somebody who had a microVAX and some transputer stuff :-( I doubt the NoMC will make much of it... -tony

Yes :-( Probably the easiest thing to find -- note I did not say it was easy to find, but it's more common than many of the other boards -- is a B004. This is a full-length ISA crd with a link adapter chip (connected to the ISA bus), a T4 or T8 transputer (the T8 has a built-in FPU, they are pin compatible, you vcan pull one and stick in the otehr and some DRAM. The connecto no the rear bracked carries the trasnptuer links from that transputer and the host adapter chip. Normally you fit a jumper to connec the host adapter to one of the links on the on-board trnasputer and then connect the other 3 links to the rest of your hardware, but you don't have to do that. Do not ask me where to find one, though:-) -tony

On Mon, 23 May 2011 21:25:26 +0100 James Wilson <james at machineroom.info> wrote: Yes. I know what Transputers are and that they are very, _very_, _*very*_ rare. Really rare, not in the ePay sense "rare". The bigger was my surprise to find this stuff in a dumpster. ... and a few other pages as well. The problem is: Datasheets don't tell me _what_ I can do with this stuff and how to do it. Datasheets cover a lot of small details. I am missing the "big picture" that gives me an overview. I am no friend of ePay. Ideally I would like to trade this stuff for something I can make a use out of. If I sell it, I would like to prefere people who actually use this stuff. I can't make this selection on ePay. I am not interrested in squeezeing as much EUR out of this stuff as I can. I got it for free. If I know it is going to someone who actually uses this stuff I may sell it for a more symbolic price. Ahh, other hardware of interrest would be a SGI Fuel or a VPro grafics and R14k CPU for my Octane. Or an Onyx 3500. ;-) -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Wed, May 25, 2011 at 7:42 AM, Richard <legalize at xmission.com> wrote: and CSP is making a comeback in the form of Google's Go language: http://golang.org/doc/go_faq.html#csp

Are they? I;ve got anumber of them, both as TRAMs and bare chips. And yet you just said that you know waht a transputer is... Basically (and this is what the data sheets don't tell you), it's a microcontroller. It's a CPU, RAM, ROM (a bootstrap program) and I/O (4 transputer links) on one chip. It's designed for parallel porcesing, you can either have several processings running on one chip (timesliced, of course), or on separate chips commuinicating via the links. The software changes in Occam to go from one to the other are very minor. The chip has an external memory bus, and generates the refresh address for DRAMs. The meory control signals are configruable. Adding external memory (even DRAM) or I/O to these devices is very easy, at lerast on the common T4s and T8s. There's one member of the T8 family (I forget the number, it's either T801 or T805) that was optimised to use SRAM only and has a somewaht faster memeory cycle time than the other members of the family, but I would be surprised if you'd got that one. The TRAM (TRAnsputer Module, BTW) communcated with the outside world via the links only. You don't have access to the memory bus. If you use the bare chip you do. As for what you might use them for, I guess it's more like 'what do you use any computer for' :-) -tony

On Wed, 25 May 2011, Tony Duell wrote: I don't know of any transputers with on-chip ROM, which model would that be? The 801 is the SRAM optimised one (separate address and data buses); the 805 is the 800 plus the break instruction and the Event Waiting pin. Alexey

On Wed, 25 May 2011 18:25:13 +0100 (BST) ard at p850ug1.demon.co.uk (Tony Duell) wrote: I know what a Transputer is. But I don't know how they where used, i.e. what a typical setup of a complete Transputer system with software developement facilities looked like, feelt like, how the workflow was and what software was used in what way. I.e. I don't know anything around the Transputers. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Thu, 26 May 2011 07:40:08 +0200 "e.stiebler" <emu at e-bbes.com> wrote: If you want some TRAMs for it... As said: PeeCee boards usually imply WinDOS. I am not going to touch this @!*$(^#. Not even for Transputers. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Fri, 27 May 2011 03:15:02 +0100 (BST) ard at p850ug1.demon.co.uk (Tony Duell) wrote: Available time. :-( If I really get a considerable amount of time to spend on a classic computing related project, it will be a gate-level reimplementation of a Zuse Z23 in a FPGA... -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Wed, 25 May 2011 18:25:13 +0100 (BST) ard at p850ug1.demon.co.uk (Tony Duell) wrote: I am a dumpser diver for more then 20 years. I visited museums and attend the Vintage Computing Festival Europa most of the time. I have seen a lot of wired things. Yet, this was the first time I came across Transputers. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}