[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
Paul Koning wrote: > Suppose you had schematics of, say, a KA-10. You could turn those > gates into VHDL or Verilog, and that should deliver an exact replica > of the original machine, bug for bug compatible. That assumes the > timing quirks are manageable The mapping from asynchronous pulses, delay lines, etc, to VHDL or Verilog isn't entirely straight forward. Still, it can be done, and in the case of a KA10, it has. See https://github.com/aap/fpdpga
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
CAREY SCHUG wrote: > What I wish somebody would create is an S-100 card (probably with a > raspberry pie daughter running simulation for future upgradeability) > that, initially emulates a complete Byte-8 or Imsai computer including > memory and disk images on sdc cards, 24x40 display on an HDMI display > and USB keyboard. serial and parallel ports emulated. I believe this tick some of your boxes: https://thehighnibble.com/imsai8080/
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
our 1620 model 2 still did multiplication by table lookup. --Carey > On 02/27/2024 9:53 PM CST Chuck Guzis via cctalk > wrote: > > > On 2/27/24 18:34, CAREY SCHUG via cctalk wrote: > > > And the 1620 does addition and multiplication by table lookup. > > That was only the CADET; the Model II had the math hardcoded. There was > an octal arithmetic option for the Model II, so it could do binary math > of a sort. Spent lots of fun hours on a CADET with a 1311 disk drive. > > --Chuck
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
On 2/27/24 18:34, CAREY SCHUG via cctalk wrote: > And the 1620 does addition and multiplication by table lookup. That was only the CADET; the Model II had the math hardcoded. There was an octal arithmetic option for the Model II, so it could do binary math of a sort. Spent lots of fun hours on a CADET with a 1311 disk drive. --Chuck
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
On 2/27/24 20:34, CAREY SCHUG via cctalk wrote: Again, even if somebody offered me a complete IBM model 30 with disk and tape drives, I could not afford the shipping. would A 360/30 could be a real problem. It used air bags to push the microcode cards against the bit line boards. Those air bags looked suspiciously like IV bags from the hospital. I can't imagine they would still hold air after 50 years. The 360/40 had mylar cards with flex-print to make the microcode word lines, these were punched to break the lines so they went either through, or around the sense transformers to select a 1 or 0 for that bit. Old mylar tends to crack, and I assume that would break the conductive printing. The 360/50 and /65 had capacitive microcode read only storage which consisted of bit line boards with copper squares and word line boards that had zig-zag word line traces that had either the drive line or balance line widened to cover the bit line square, to select 1 or 0. These boards were separated by a 1 mil mylar sheet, and squeezed by a spring-loaded metal plate and a foam pressure spreader. I think these have at least a SLIGHT chance of still working after 5 decades. The power was not real high on these machines, although the peripherals could draw a lot. The 360/50 and /65 CPUs drew just a couple KW each. Of course, the models /50 and /65 were a LOT bigger than a /30, and much heavier. There's a reason people used to talk about "big iron". As for the logic, the 360's used SLT, a 1/2" square ceramic hybrid with, generally two transistors, 4 diodes and 4 resistors per package. The complete schematics for all of them are in the FEMM, and I'll bet one could replace them with a tiny PCB with SOT23 transistors, etc. to replace any bad modules. Jon
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
I'd say the real cost is the second or third system to get spare parts. that is why I want to replace the WD chip. the microprocessor talks to it at bus speed. the os knows it has to wait, though some waits are for the wd chip to say it is done. a SIMPLE mod to the legacy OS can eliminate those waits, thus IO as fast as a real hard disk, maybe as fast as an SSD. I love the 1130, the only I/O instructions given to the printer are input. And the 1620 does addition and multiplication by table lookup. Again, even if somebody offered me a complete IBM model 30 with disk and tape drives, I could not afford the shipping. would probably have to take out the a window to get it inside (big expense), and it might instantly become a basement installation. I'd have to have much more power and A/C installed. And no spare parts. Now, any front panels... Even recreations of front panels would be treasured. In the 1980s the Bradford exchange had a 360/40 front panel mounted in their space with lights artificially flashing. I might even attempt to modify an emulator program to flash the lights appropriately. --Carey > On 02/27/2024 7:16 PM CST ben via cctalk wrote: > > > On 2024-02-27 3:09 p.m., Paul Koning via cctalk wrote: > > > > > >> On Feb 27, 2024, at 4:49 PM, CAREY SCHUG via cctalk > >> wrote: > >> > >> Religion warning: I was a mainframer. Since at any practical budget, > >> they can only be emulated, > > > > Dumpster diving is a 0 dollar budget. > People could afford the APPLE II, 8080 S-100 bus, SWTPC 6909. I assume > with careful shopping one can rebuild them for about the same price, in > small quanities. > Power supplies require harder to find parts. > > Main frame rebuilding is costly, but I suspect the real cost is I/O > that can't be duplicated. A hardware emulation using microcode to me > is real computer, a windows fly by night emulation is not, as the base > platform is too unstable. > > > > Depends on your definition of emulated. Is an FPGA version merely an > > "emulation"? You might say yes if it's a functional model. Arguably no, > > if it's a gate level model. > > > I have bad luck with FPGA's, too many timing issues with routing. > I have better luck with a 2901 4 bit alu and some support logic mounted > on a small pcb. > > > Suppose you had schematics of, say, a KA-10. You could turn those gates > > into VHDL or Verilog, and that should deliver an exact replica of the > > original machine, bug for bug compatible. That assumes the timing quirks > > are manageable, which for most machines should be true. (It isn't for a > > CDC 6600.) > > > > paul > > The IBM 1130 is also a pretty scary machine inside. > The blog is here. > https://rescue1130.blogspot.com/ > > Ben.
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
On 2024-02-27 3:09 p.m., Paul Koning via cctalk wrote: On Feb 27, 2024, at 4:49 PM, CAREY SCHUG via cctalk wrote: Religion warning: I was a mainframer. Since at any practical budget, they can only be emulated, Dumpster diving is a 0 dollar budget. People could afford the APPLE II, 8080 S-100 bus, SWTPC 6909. I assume with careful shopping one can rebuild them for about the same price, in small quanities. Power supplies require harder to find parts. Main frame rebuilding is costly, but I suspect the real cost is I/O that can't be duplicated. A hardware emulation using microcode to me is real computer, a windows fly by night emulation is not, as the base platform is too unstable. Depends on your definition of emulated. Is an FPGA version merely an "emulation"? You might say yes if it's a functional model. Arguably no, if it's a gate level model. I have bad luck with FPGA's, too many timing issues with routing. I have better luck with a 2901 4 bit alu and some support logic mounted on a small pcb. Suppose you had schematics of, say, a KA-10. You could turn those gates into VHDL or Verilog, and that should deliver an exact replica of the original machine, bug for bug compatible. That assumes the timing quirks are manageable, which for most machines should be true. (It isn't for a CDC 6600.) paul The IBM 1130 is also a pretty scary machine inside. The blog is here. https://rescue1130.blogspot.com/ Ben.
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
On 2/27/24 14:09, Paul Koning via cctalk wrote: > Suppose you had schematics of, say, a KA-10. You could turn those gates into > VHDL or Verilog, and that should deliver an exact replica of the original > machine, bug for bug compatible. That assumes the timing quirks are > manageable, which for most machines should be true. (It isn't for a CDC > 6600.) A section manager from years ago mentioned to me that, as a fresh EE out of school, his first job at CDC was measuring the loops of (taper pin) wire on the 6600 to which Seymour had attached tags that said "tune". --Chuck
[cctalk] Re: recreating old computers [was: Paper tape in casettes...]
> On Feb 27, 2024, at 4:49 PM, CAREY SCHUG via cctalk > wrote: > > Religion warning: I was a mainframer. Since at any practical budget, they > can only be emulated, Depends on your definition of emulated. Is an FPGA version merely an "emulation"? You might say yes if it's a functional model. Arguably no, if it's a gate level model. Suppose you had schematics of, say, a KA-10. You could turn those gates into VHDL or Verilog, and that should deliver an exact replica of the original machine, bug for bug compatible. That assumes the timing quirks are manageable, which for most machines should be true. (It isn't for a CDC 6600.) paul