AnyCPU http://anycpu.org/forum/ |
|
octal computing 2022 back to 1959 http://anycpu.org/forum/viewtopic.php?f=23&t=972 |
Page 2 of 2 |
Author: | BigEd [ Fri Dec 30, 2022 9:44 am ] |
Post subject: | Re: octal computing 2022 back to 1959 |
There's a link to Goldberg's 27 bits paper in 7 Bits Are Not Enough for 2-Digit Accuracy but it's not a great link, so here's one to the ACM version. |
Author: | oldben [ Mon Jan 09, 2023 5:01 pm ] |
Post subject: | Re: octal computing 2022 back to 1959 |
User programable roms seem to have come out around 1972 as well as most MSI needed for a 4 bit data path. The 20 bit cpu,looks to be the best design using the memory at the time: 1.5 us core. Floating point is 9 digits binary with +- 76 exponent. Spring 1973 looks resonable for the ball park FPGA emulation. Concept design is like this PDP11. Attachment: 45_pdp_11_45.jpg [ 1.04 MiB | Viewed 4092 times ] The 24 bit cpu,may be the best value for the money,as you get a 48 bit floating point number.(11 digits).Note sure if it will fit in same case as the PDP 11-45. Memory is still is the biggest cost of the machine.Floating point routines are slowly getting developed, as I have just the I/o to write. The cross assembler needs to have floating point input routines written. Ben. |
Author: | robfinch [ Tue Jan 10, 2023 5:24 am ] |
Post subject: | Re: octal computing 2022 back to 1959 |
24 bits is a nice size. Years ago, after reading about a homebrewed CPU in BYTE magazine, I sketched out a 24-bit CPU on graph paper. It was about 75 to 80 MSI chips, beyond what I could build at the time. I have been tempted several times to create it in an FPGA. It had four address modes, R, [R], [R]+, and –[R]. That PDP case looks good. I had the thought of using plastic shelving for a larger CPU. |
Author: | oldben [ Tue Jan 17, 2023 8:07 pm ] |
Post subject: | Re: octal computing 2022 back to 1959 |
I will stick with 20 bits as that makes a simple hex front panel. I only need to spend a arm and a leg and 3 toes for the PCB. I need some generic sub mini toggle switch foot prints for ki-cad. Know of any, anybody? Floating point is now done, but i need now to get some sort of boot strap programming lanquage using a revised meta II compiler compiler and a cross assembler ( in c) to support high level language constructs like floating point input, local variables and defines. Ben. |
Page 2 of 2 | All times are UTC |
Powered by phpBB® Forum Software © phpBB Group http://www.phpbb.com/ |