Now that would be interesting.... How would we reproduce the 55dBa Noise output and the 1750 W heat dissipation? :-) http://www.inwap.com/pdp10/td-1b.html Doug

- Something running, say, a 6502, 8080, 8085, or Z80. Definitely not interested in anything post 8085. Yes, you would be interested (possibly, not sure!) in the Altair 8800 re-make. - Even if you did manage to get the money to build it, what software could take advantage of it? The whole reason for parallel processing isn't the "coolness" of it, it is because the person wants results faster. That problem is much more easily achieved with a modern CPU. Grandma doesn't need SSE3 (maybe you've seen a grandma with an IT degree, though). I'm willing to try and see if a low end from nowadays is more expensive to make than sixteen 286 or 32 Zilogs. - A 32 CPU Z80 system might make for a cool demo and web page, but not much more. The real sales won't come from that :) The company would still sell dual cores and such new things, but not exclusively. That's the idea. Also, these remanufactured machines would be made so that you can actually use them for a few tasks from today. You can decode MP3s with a 286, using a hardware MPEG-2 decoder for the ISA bus, for example. That's why I thought of putting more than one CPU in there. No other reason. Takes away the coolness ? Maybe, but it's healthy to have more CPU power. - and the drivers come from where? Coders. Same as everybody else. Besides, when taking the technologies from the original manufacturers, I would ask for the latest driver source files as well. Also, for the drivers that don't exist (like monochrome graphics on a Hercules video card for Windows NT family) I would need the specs and the coder would do the rest. - You think intel will give you the rights to manufacture old x86 designs? or nvidia will care to unearth and hand over the 3dfx source code (chips, drivers, test programs) because they think it would be a cool hack? No, but because they would get some money for that. I'm looking for ways to contact Intel Capital a bit more directly, to be honest. They have a lot of the chips I need - USB 1.1 controllers, ISA hard drive controllers, a lot of CPUs, etc. Nvidia simply managed to cut off the advancement of 3dfx cards and rendered them obsolete. I wouldn't interfere with the modern product line-up, just remake the old ones. - It sounds like a far-fetched pipe dream to me. See if you can pull off something on a smaller scale before you attempt to build grandiose systems. What other systems have you designed, built, and brought to completion? Quite! Yes, starting off with something that you can find the parts for (like a dual core system) is probably easier. Other systems designed by me ? I'm a guy with ideas, what you would call a producer. Not an engineer. I understand this list is populated by very technical people so you might have assumed I'm one of them, but I'm not. However, the single most interesting question was this: - Out of curiosity, are you talking about actually putting down 16 CPUs with 16 independent memory systems, or are you talking about some kind of large FPGA with 16 CPUs on-board? Oh God, thank you for my expert friend from Australia! Yes, Sir, he told me how. Basically the 286 would use hardware expanded memory as the link within a cluster of sixteen motherboards (out of which fifteen are identical, just the CPU, RAM and chipset as far as I understood it, the last one having all the other chips - graphics, sound, etc.) and the cluster would work as a single, multi-CPU computer. Now, when it comes to the 386 and onwards, these can address 4 GB of RAM. I wouldn't put 16 mobos in one cluster with each of them having 4 GB of RAM. Instead, the northbridge would probably be connected to a custom bus syncing all the CPUs. Something along the lines of HyperThreading, which he said is something similar to non uniform memory addressing. The point is, use the same amount of RAM for all the CPUs. Like the difference between regular multi-CPU systems and regular multi-core CPU systems. You probably know multi-CPU systems have their memory "per-CPU" while the cores inside a multi-core CPU all use the same system RAM. - I think what you're talking about is either a reproduction or emulation. I'm not natively English! :) I didn't use "refurbish" because I didn't mean refurbish, indeed. Reproduction is the most accurate term but what if the inscriptions on the chips would be changed ? I think AMD changed their logo since the 386 DX 40 MHz, didn't they ? Also, I would need two FPUs from Cyrix, which are now owned by VIA. I would probably have to stamp VIA on them, not Cyrix. I think someone asked about 16-bit or 8-bit operating systems able to take advantage of multi-CPU configurations. Yep, Linux can be coded, ReactOS can be coded, FreeDOS can be coded for those purposes. I would think these are enough. It's simple, not easy. There is already an 8-bit "sort-of" Linux called Lunix: http://hld.c64.org/poldi/lunix/lunix.html There is already a 16-bit Linux called ELKS: http://elks.sourceforge.net/ Extremely small market ? Again, there will be dual cores and quad cores and everything. Computers "with a purpose" the contents of which the average user won't know. He will simply know his needs are covered. I'm just here to take the pulse of the classic computer community. You apparently don't like the idea very much. The XKL Toad-1 is a huge beast for what I have in mind. Cloning that one would be a major pain and it would be very hard to ship to a buyer. 100 pounds including 40 pounds battery backup ? You gotta be kidding me. But thank you for all your opinions. Any other comments you might have, I'll be glad to read them. It can't hurt.

________________________________________ From: cctalk-bounces at classiccmp.org [cctalk-bounces at classiccmp.org] On Behalf Of Alexandru Lovin [thypope at gmail.com] Sent: Wednesday, May 27, 2009 6:25 PM To: General Discussion: On-Topic and Off-Topic Posts Subject: Re: Making vintage computers [snip] Oh God, thank you for my expert friend from Australia! Yes, Sir, he told me how. Basically the 286 would use hardware expanded memory as the link within a cluster of sixteen motherboards (out of which fifteen are identical, just the CPU, RAM and chipset as far as I understood it, the last one having all the other chips - graphics, sound, etc.) and the cluster would work as a single, multi-CPU computer. Now, when it comes to the 386 and onwards, these can address 4 GB of RAM. I wouldn't put 16 mobos in one cluster with each of them having 4 GB of RAM. Instead, the northbridge would probably be connected to a custom bus syncing all the CPUs. Something along the lines of HyperThreading, which he said is something similar to non uniform memory addressing. The point is, use the same amount of RAM for all the CPUs. Like the difference between regular multi-CPU systems and regular multi-core CPU systems. You probably know multi-CPU systems have their memory "per-CPU" while the cores inside a multi-core CPU all use the same system RAM. Sorry, hyperthreading (aka SMT) is NOTHING like non-uniform memory addressing (NUMA). Hyperthreading is an architecture where additional program counters and registers can access the same functional units (e.g. ALU) to exploit thread-level parallelism, while the traditional superscalar architecture exploits instruction-level parallelism. With the architecture you describe, it seems like a cache-coherent NUMA. BTW, multi-CPU vs. multicore isn't that much different, except that it's less hardware for multicore systems to share the memory interface - each one has its own cache, with traditional protocols to avoid sequential incoherency (i.e. a different result from multiple CPUs processing an instruction stream as opposed to one CPU). -- Ian

On Wed, May 27, 2009 at 6:53 PM, Doug Jackson <doug at stillhq.com> wrote: Build the CPU into a working clothes dryer? -ethan

Doug Jackson wrote: There ARE multi-CPU vintage systems. Some of the CPUs in question were many, many boards full of chips. Peace... Sridhar

On 28 May 2009 at 17:43, Patrick Finnegan wrote: CDC 6500? (Purdue used to own one). There's also the 6700 (6600+6400 in the same box). --Chuck

On 28 May 2009 at 20:09, Patrick Finnegan wrote: For certain, ILLIAC, but the 6500 was 2 6400 CPUs sharing a common memory and PPU set. Doubtless there were other 60's era machines to do the same, but no others readily come to mind. Each CPU in the 6500 had no hardware-defined way to tell what the other was doing. The PPUs could exchange jump them selectively and I seem to recall that when CEJ/MEJ came along, one or the other, but not both, could be in monitor mode. --Chuck

On 28 May 2009 at 21:30, William Donzelli wrote: I think you're right about Burroughs, but not about the B5500. I believe the military D825 had dual CPUs. --Chuck

On 28 May 2009 at 22:53, William Donzelli wrote: I'm calling the D825 as being earlier as it was a vacuum-tube box, while the B5500 was solid-state. In any case, the design was probably much earlier than the B5500. The Fact Sheet for the BUIC claims: "Data exchange occurs simultaneously between any memory and any computer or input/output module." That could imply that although SAGE used them as backup, they weren't necessarily limited in that way. Anyone know for certain? Could both CPUs be running at once? --Chuck

On Thu, 28 May 2009 15:31:41 -0700, Rich Alderson wrote: I thought the 370/158-MP was IBMs first true tightly bound multiprocessor it was introduced in Feb 1973. The 370/158-AP & MP was introduced in 1976. The AP did in fact have 2 cpus but they were not so tightly bound as the MP, it was more of a co- processor as I recal. The 3033MP came along in 1978. The rest is history. The other Bob

I'm surprised that no one had yet mentioned BINAC. Of course, in it, the two CPUs weren't for performance so much as for fault detection. BLS

I thnik you need to distinguish between : Multi-chip CPUs, like many older minicomputers, where the CPU may well be hundreds of chips, but where there is really only one CPU running one program at a time And Multiple CPUs, where the CPUs may be single chips or boards of many chips, but where there are several programs running at the same time,. one on each CPU. And then you have the case where there are several microporcessors in the cabinet, but some of then run programs from ROM only (you can't run user programs on them) for things like I/O, disk control, etc. There are 'classsic' examples of both IMHO. Or at least I assume things like older transputers, DAPs, etc are now 'classic' -tony

On 29 May 2009 at 15:12, Sridhar Ayengar wrote: I was trying to restrict myself to multiple processing units sharing a common memory and peripherals. Would we now call that 'multi- core?" --Chuck