Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > sci.electronics.design > #452905 > unrolled thread
| Started by | John Larkin <jjlarkin@highlandtechnology.com> |
|---|---|
| First post | 2017-03-21 10:21 -0700 |
| Last post | 2017-03-24 08:54 -0400 |
| Articles | 20 on this page of 113 — 24 participants |
Back to article view | Back to sci.electronics.design
big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-21 10:21 -0700
Re: big drives bitrex <bitrex@de.lete.earthlink.net> - 2017-03-21 13:47 -0400
Re: big drives George Herold <gherold@teachspin.com> - 2017-03-21 12:42 -0700
Re: big drives dagmargoodboat@yahoo.com - 2017-03-21 12:55 -0700
Re: big drives George Herold <gherold@teachspin.com> - 2017-03-21 13:37 -0700
Re: big drives bill.sloman@ieee.org - 2017-03-21 21:51 -0700
Re: big drives George Herold <gherold@teachspin.com> - 2017-03-22 07:26 -0700
Re: big drives Winfield Hill <hill@rowland.harvard.edu> - 2017-03-22 03:38 -0700
Re: big drives Don Kuenz <g@crcomp.net> - 2017-03-22 14:23 +0000
Re: big drives David Brown <david.brown@hesbynett.no> - 2017-03-22 16:43 +0100
Re: big drives Don Kuenz <g@crcomp.net> - 2017-03-22 18:35 +0000
Re: big drives David Brown <david.brown@hesbynett.no> - 2017-03-23 10:24 +0100
Re: big drives Don Kuenz <g@crcomp.net> - 2017-03-23 16:11 +0000
Re: big drives krw@notreal.com - 2017-03-23 13:05 -0400
Re: big drives Don Kuenz <g@crcomp.net> - 2017-03-23 17:39 +0000
Re: big drives krw@notreal.com - 2017-03-23 21:05 -0400
Re: big drives Don Kuenz <g@crcomp.net> - 2017-03-24 03:01 +0000
Re: big drives krw@notreal.com - 2017-03-23 23:09 -0400
Re: big drives Don Kuenz <g@crcomp.net> - 2017-03-24 04:15 +0000
Re: big drives krw@notreal.com - 2017-03-24 22:34 -0400
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 12:25 -0700
Re: big drives dplatt@coop.radagast.org (Dave Platt) - 2017-03-22 13:57 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 16:03 -0700
Re: big drives krw@notreal.com - 2017-03-22 20:29 -0400
Re: big drives tabbypurr@gmail.com - 2017-03-22 04:36 -0700
Re: big drives Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> - 2017-03-22 08:57 -0400
Re: big drives Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> - 2017-03-22 11:59 -0400
Re: big drives krw@notreal.com - 2017-03-22 13:23 -0400
Re: big drives Jon Elson <jmelson@wustl.edu> - 2017-03-22 16:38 -0500
Re: big drives krw@notreal.com - 2017-03-22 20:31 -0400
Re: big drives gray_wolf <g_wolf@nospam.com> - 2017-03-23 20:59 -0500
Re: big drives gray_wolf <g_wolf@nospam.com> - 2017-03-23 21:00 -0500
Re: big drives Robert Baer <robertbaer@localnet.com> - 2017-03-23 22:21 -0800
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-22 07:51 -0700
Re: big drives krw@notreal.com - 2017-03-22 13:23 -0400
Re: big drives Robert Baer <robertbaer@localnet.com> - 2017-03-23 22:19 -0800
Re: big drives tabbypurr@gmail.com - 2017-03-24 04:26 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 08:38 -0700
Re: big drives Jon Elson <jmelson@wustl.edu> - 2017-03-22 16:37 -0500
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-22 16:07 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 16:21 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-22 16:45 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 23:21 -0700
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 13:06 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 16:56 -0700
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 17:35 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 21:47 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 16:19 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-22 16:50 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 17:58 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-22 21:00 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-22 22:50 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-23 08:27 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-23 12:25 -0700
Re: big drives tabbypurr@gmail.com - 2017-03-24 04:31 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 06:38 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 08:41 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 11:38 -0700
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 13:55 -0700
Re: big drives tabbypurr@gmail.com - 2017-03-24 17:14 -0700
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 17:41 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 18:00 -0700
Re: big drives Joerg <news@analogconsultants.com> - 2017-03-23 09:35 -0700
Re: big drives "Michael A. Terrell" <mike.terrell@earthlink.net> - 2017-03-23 18:38 -0400
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-23 16:18 -0700
Re: big drives Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> - 2017-03-23 18:29 -0400
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-23 18:38 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-23 18:54 -0700
Re: big drives pcdhobbs@gmail.com - 2017-03-23 19:16 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-23 19:28 -0700
Re: big drives pcdhobbs@gmail.com - 2017-03-23 19:55 -0700
Re: big drives krw@notreal.com - 2017-03-23 23:04 -0400
Re: big drives pcdhobbs@gmail.com - 2017-03-23 22:00 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 08:46 -0700
Re: big drives krw@notreal.com - 2017-03-24 22:38 -0400
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-23 22:22 -0700
Re: big drives pcdhobbs@gmail.com - 2017-03-23 23:08 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 00:21 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 09:00 -0700
Re: big drives David Brown <david.brown@hesbynett.no> - 2017-03-24 17:14 +0100
Re: big drives tabbypurr@gmail.com - 2017-03-24 09:19 -0700
Re: big drives David Brown <david.brown@hesbynett.no> - 2017-03-24 17:47 +0100
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 12:01 -0700
Re: big drives Jasen Betts <jasen@xnet.co.nz> - 2017-03-24 20:30 +0000
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 14:06 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 11:55 -0700
Re: big drives "Michael A. Terrell" <mike.terrell@earthlink.net> - 2017-03-25 04:00 -0400
Re: big drives upsidedown@downunder.com - 2017-03-24 22:18 +0200
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 14:15 -0700
Re: big drives Tauno Voipio <tauno.voipio@notused.fi.invalid> - 2017-03-25 13:21 +0200
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-25 07:54 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-25 10:53 -0700
Re: big drives upsidedown@downunder.com - 2017-03-25 23:11 +0200
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-25 15:14 -0700
Re: big drives upsidedown@downunder.com - 2017-03-26 12:18 +0300
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-26 03:33 -0700
Re: big drives Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> - 2017-03-26 16:36 -0400
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-26 14:13 -0700
Re: big drives Tauno Voipio <tauno.voipio@notused.fi.invalid> - 2017-03-25 21:04 +0200
Re: big drives pcdhobbs@gmail.com - 2017-03-24 19:03 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 08:53 -0700
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 12:12 -0700
Re: big drives John Larkin <jjlarkinxyxy@highlandtechnology.com> - 2017-03-24 14:20 -0700
Re: big drives krw@notreal.com - 2017-03-24 22:43 -0400
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 20:10 -0700
Re: big drives krw@notreal.com - 2017-03-24 23:23 -0400
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-24 22:25 -0700
Re: big drives Tauno Voipio <tauno.voipio@notused.fi.invalid> - 2017-03-25 13:23 +0200
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-23 19:36 -0700
Re: big drives upsidedown@downunder.com - 2017-03-24 10:14 +0200
Re: big drives Don Y <blockedofcourse@foo.invalid> - 2017-03-24 01:31 -0700
Re: big drives John Larkin <jjlarkin@highlandtechnology.com> - 2017-03-23 16:21 -0700
Re: big drives "Michael A. Terrell" <mike.terrell@earthlink.net> - 2017-03-24 08:54 -0400
Page 5 of 6 — ← Prev page 1 2 3 4 [5] 6 Next page →
| From | tabbypurr@gmail.com |
|---|---|
| Date | 2017-03-24 09:19 -0700 |
| Message-ID | <2adcd4f3-021c-488b-b1fc-ffffb33a2283@googlegroups.com> |
| In reply to | #453407 |
On Friday, 24 March 2017 16:01:03 UTC, John Larkin wrote: > The first (and only) computer that I actually designed was two boards > full of TTL. It had three instructions and a 20 KHz 4-phase clock. > > Everybody should design a CPU just once, even as a mental activity. > > My career has been blessed by working for companies (including my own) > that had no coherent marketing. What can you do with 3 instructions? Even + - x / requires 4. I've tried to mentally imagine a variety of oddball CPU ideas, including one with just one 'operateall' instruction. The idea is it has a great number of bits width, and the one & only operation adds one section of the input number, multiplies another section, etc etc. A bizarre case of parallel processing that removes the need to handle differing instructions entirely. NT
[toc] | [prev] | [next] | [standalone]
| From | David Brown <david.brown@hesbynett.no> |
|---|---|
| Date | 2017-03-24 17:47 +0100 |
| Message-ID | <ob3ie5$7aq$1@dont-email.me> |
| In reply to | #453412 |
On 24/03/17 17:19, tabbypurr@gmail.com wrote: > On Friday, 24 March 2017 16:01:03 UTC, John Larkin wrote: > >> The first (and only) computer that I actually designed was two >> boards full of TTL. It had three instructions and a 20 KHz 4-phase >> clock. >> >> Everybody should design a CPU just once, even as a mental >> activity. >> >> My career has been blessed by working for companies (including my >> own) that had no coherent marketing. > > What can you do with 3 instructions? Even + - x / requires 4. > > I've tried to mentally imagine a variety of oddball CPU ideas, > including one with just one 'operateall' instruction. The idea is it > has a great number of bits width, and the one & only operation adds > one section of the input number, multiplies another section, etc etc. > A bizarre case of parallel processing that removes the need to handle > differing instructions entirely. > I have seen a real life microcontroller that had only one instruction - a "move" instruction (with a small number of addressing modes). ALU operations were done by writing the two operands to fixed addresses, and reading the result from another address (different operations had different result addresses). Of course, you could argue that this is cheating since it had lots of ALU functions - it's just that the ALU was a memory-mapped peripheral. Then there are these: <https://en.wikipedia.org/wiki/One_instruction_set_computer>
[toc] | [prev] | [next] | [standalone]
| From | Don Y <blockedofcourse@foo.invalid> |
|---|---|
| Date | 2017-03-24 12:01 -0700 |
| Message-ID | <ob3q9g$3ke$1@dont-email.me> |
| In reply to | #453412 |
On 3/24/2017 9:19 AM, tabbypurr@gmail.com wrote: > On Friday, 24 March 2017 16:01:03 UTC, John Larkin wrote: > >> The first (and only) computer that I actually designed was two boards full >> of TTL. It had three instructions and a 20 KHz 4-phase clock. > What can you do with 3 instructions? Even + - x / requires 4. You can multiply as repeated addition; divide as repeated subtraction. So, if you have more than one accumulator/memory, you really only need + and "change sign" Historically, there have been many machines with *truly* "reduced" instruction sets. When you sit down with one, you immediately wonder how you can ever get anything done with so scant capabilities! But, once you start using it, you realize that much is possible with very little -- its just "tedious". > I've tried to mentally imagine a variety of oddball CPU ideas, including one > with just one 'operateall' instruction. The idea is it has a great number of > bits width, and the one & only operation adds one section of the input > number, multiplies another section, etc etc. A bizarre case of parallel > processing that removes the need to handle differing instructions entirely. Some of the more interesting ideas (esp legacy systems) come from how you treat dependencies in the data/operations "in live logic". I.e., letting the hardware sort out that the result of this "execution unit" relies on the result from some other "execution unit" that is currently "busy" -- and automatically harvesting that result when it becomes available, etc. Or, opportunistic computations to anticipate potential program flow before it is encountered, etc.
[toc] | [prev] | [next] | [standalone]
| From | Jasen Betts <jasen@xnet.co.nz> |
|---|---|
| Date | 2017-03-24 20:30 +0000 |
| Message-ID | <ob3vks$pjf$1@gonzo.alcatraz> |
| In reply to | #453427 |
On 2017-03-24, Don Y <blockedofcourse@foo.invalid> wrote: > On 3/24/2017 9:19 AM, tabbypurr@gmail.com wrote: >> On Friday, 24 March 2017 16:01:03 UTC, John Larkin wrote: >> >>> The first (and only) computer that I actually designed was two boards full >>> of TTL. It had three instructions and a 20 KHz 4-phase clock. > >> What can you do with 3 instructions? Even + - x / requires 4. > > You can multiply as repeated addition; divide as repeated subtraction. > So, if you have more than one accumulator/memory, you really only need > + and "change sign" add and change sign can both be done using subtraction a - (0 - b) so the only arithmetic operation needed is subtraction. -- This email has not been checked by half-arsed antivirus software
[toc] | [prev] | [next] | [standalone]
| From | John Larkin <jjlarkinxyxy@highlandtechnology.com> |
|---|---|
| Date | 2017-03-24 14:06 -0700 |
| Message-ID | <mv1bdc93uei1sb3hbl5sif1crvo2c7ico5@4ax.com> |
| In reply to | #453412 |
On Fri, 24 Mar 2017 09:19:19 -0700 (PDT), tabbypurr@gmail.com wrote: >On Friday, 24 March 2017 16:01:03 UTC, John Larkin wrote: > >> The first (and only) computer that I actually designed was two boards >> full of TTL. It had three instructions and a 20 KHz 4-phase clock. >> >> Everybody should design a CPU just once, even as a mental activity. >> >> My career has been blessed by working for companies (including my own) >> that had no coherent marketing. > >What can you do with 3 instructions? Even + - x / requires 4. Move, conditional jump, extended jump. It was for a shipboard data logger, which printed bell logs. It didn't do any math; the ADC generated ASCII. I figured that I could add an ALU as a peripherial, but we never needed it. Not long after I did this, the MC6800 became affordable. I actually wrote a pre-emptive RTOS for the 6800 (in longhand, in Juneau, Alaska. That's a whole nother story.) > >I've tried to mentally imagine a variety of oddball CPU ideas, including one with just one 'operateall' instruction. The idea is it has a great number of bits width, and the one & only operation adds one section of the input number, multiplies another section, etc etc. A bizarre case of parallel processing that removes the need to handle differing instructions entirely. > ARM does some sort-of-combined instructions. A VLIW RISC machine can have a gigantic instruction word with a zillion bit fields, and can do multiple things on each clock. Like microcode. As I said, it's fun to design a CPU and instruction set, once. -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
[toc] | [prev] | [next] | [standalone]
| From | Don Y <blockedofcourse@foo.invalid> |
|---|---|
| Date | 2017-03-24 11:55 -0700 |
| Message-ID | <ob3ptb$1m7$1@dont-email.me> |
| In reply to | #453407 |
On 3/24/2017 9:00 AM, John Larkin wrote: > On Fri, 24 Mar 2017 00:21:50 -0700, Don Y > <blockedofcourse@foo.invalid> wrote: > >> On 3/23/2017 11:08 PM, pcdhobbs@gmail.com wrote: >>>> How much did you part with for the kit YOU were using in the mid 70's? >>> >>> Your reference was from 1970, >> >> No, the MANUAL that I referenced was from 1970. My reference to the >> *equipment* was: >> "My first *hard* disk had a capacity of 256K words..." >> John Larkin referenced 1970 (feel free to check back up-thread) >> >>> when I was 10 years old and had just got a >>> soldering gun and some plumber's solder for my birthday. (My parents were >>> very supportive but didn't know much about radio. The debt I owe them has >>> hopefully been paid to the next generation.) >>> >>> Soon after that point I had to save up a month's pocket money plus what I >>> earned from extra chores to buy parts to build projects with. >> >> I built my first computers from "samples" begged from sales reps (while >> working my way through school). And, from bits of kit gifted to me by friends >> who worked at firms like DEC (I could never afford to buy even USED kit of this >> sort AND arrange to pay my tuition, rent, board, etc.) > > The first (and only) computer that I actually designed was two boards > full of TTL. It had three instructions and a 20 KHz 4-phase clock. My first "computing device" was a combination door lock built out of relays, latching relays, time delay relays and mechanical sequencing elements. It was the size of a large microwave oven and housed in a plywood box. My first *analog* "computer" design was applied as a "football" game (offensive and defensive "play" selectors, result logic to indicate net gain/loss per play, accumulators to reflect position on the field, yards to first down, etc). It filled a 4'x8' sheet of plywood (and the back of a stationwagon to get it to school). My first *digital* computer was tonka-toy logic with bipolar ROMs as the control store (easier and more versatile than instruction decoding *logic*), 16b address/data bus, 20MHz system clock (though the bus ran at 10MHz) fabbed on wire-wrapped perf-board. I also built one using 29xx devices -- but that was far more "traditional" in design and capability (and more expensive to implement). Since then, many "specialty" processors (application specific). E.g., one had an opcode that operated on 2D vectors and would do 4 (16b) multiplies, a division and two adds in a single 1.6us instruction. > Everybody should design a CPU just once, even as a mental activity. I think the best part of such an "exercise" is the consideration towards what "operators" would be appropriate to the intended application. If you just sit down and design a generic instruction set, it's just an exercise in state machine design. OTOH, if you think about how it will be *applied* to a particular class of problems, then you have more freedom to *imagine* the opcodes that will be of most use to the "programmer". > My career has been blessed by working for companies (including my own) > that had no coherent marketing. More deliberate effort is required as a contractor; you have to make a point of looking for different application domains requiring different skill sets, etc. Otherwise, you get sucked into a rut where you're just repeating/refining (your!) designs in perpetuity and never really learninig (much) new. [OTOH, I know many folks who have been financially very successful with this strategy. But, they tend to be intellectually bored/stagnant/unimaginative.]
[toc] | [prev] | [next] | [standalone]
| From | "Michael A. Terrell" <mike.terrell@earthlink.net> |
|---|---|
| Date | 2017-03-25 04:00 -0400 |
| Message-ID | <-oKdnZ7jGIA1vkvFnZ2dnUU7-Y3NnZ2d@earthlink.com> |
| In reply to | #453426 |
Don Y wrote:
>
> My first "computing device" was a combination door lock built out of
> relays, latching relays, time delay relays and mechanical sequencing
> elements. It was the size of a large microwave oven and housed in a
> plywood box.
My first CMOS project was to design an electronic lock with a single
4017 counter IC. Some RC networks, a pass transistor and a diode
controlled the output. The only limits were that you couldn't reuse a
number, and any number out of sequence cause a rest with a five second
delay to prevent someone from running through the possible codes in a
hurry. You could configure it to use from two, to ten digit access codes.
--
Never piss off an Engineer!
They don't get mad.
They don't get even.
They go for over unity! ;-)
[toc] | [prev] | [next] | [standalone]
| From | upsidedown@downunder.com |
|---|---|
| Date | 2017-03-24 22:18 +0200 |
| Message-ID | <hvuadc9h9tbsdg49v4pqu43v6l8u9gcbbj@4ax.com> |
| In reply to | #453407 |
On Fri, 24 Mar 2017 09:00:57 -0700, John Larkin <jjlarkin@highlandtechnology.com> wrote: >On Fri, 24 Mar 2017 00:21:50 -0700, Don Y ><blockedofcourse@foo.invalid> wrote: > >>On 3/23/2017 11:08 PM, pcdhobbs@gmail.com wrote: >>>> How much did you part with for the kit YOU were using in the mid 70's? >>> >>> Your reference was from 1970, >> >>No, the MANUAL that I referenced was from 1970. My reference to the >>*equipment* was: >> "My first *hard* disk had a capacity of 256K words..." >>John Larkin referenced 1970 (feel free to check back up-thread) >> >>> when I was 10 years old and had just got a >>> soldering gun and some plumber's solder for my birthday. (My parents were >>> very supportive but didn't know much about radio. The debt I owe them has >>> hopefully been paid to the next generation.) >>> >>> Soon after that point I had to save up a month's pocket money plus what I >>> earned from extra chores to buy parts to build projects with. >> >>I built my first computers from "samples" begged from sales reps (while >>working my way through school). And, from bits of kit gifted to me by friends >>who worked at firms like DEC (I could never afford to buy even USED kit of this >>sort AND arrange to pay my tuition, rent, board, etc.) > >The first (and only) computer that I actually designed was two boards >full of TTL. It had three instructions and a 20 KHz 4-phase clock. > >Everybody should design a CPU just once, even as a mental activity. Look at the original Data General Nova computer, which was designed around the 74181 ALU chip :-). I once bought some 74181 (not 74S or 74LS) chips to be much simpler than a own design, but unfortunately I didn't build an real computer.
[toc] | [prev] | [next] | [standalone]
| From | John Larkin <jjlarkinxyxy@highlandtechnology.com> |
|---|---|
| Date | 2017-03-24 14:15 -0700 |
| Message-ID | <kl2bdc9jid0ukj8ptvc5m53lg5cmd396ih@4ax.com> |
| In reply to | #453438 |
On Fri, 24 Mar 2017 22:18:48 +0200, upsidedown@downunder.com wrote: >On Fri, 24 Mar 2017 09:00:57 -0700, John Larkin ><jjlarkin@highlandtechnology.com> wrote: > >>On Fri, 24 Mar 2017 00:21:50 -0700, Don Y >><blockedofcourse@foo.invalid> wrote: >> >>>On 3/23/2017 11:08 PM, pcdhobbs@gmail.com wrote: >>>>> How much did you part with for the kit YOU were using in the mid 70's? >>>> >>>> Your reference was from 1970, >>> >>>No, the MANUAL that I referenced was from 1970. My reference to the >>>*equipment* was: >>> "My first *hard* disk had a capacity of 256K words..." >>>John Larkin referenced 1970 (feel free to check back up-thread) >>> >>>> when I was 10 years old and had just got a >>>> soldering gun and some plumber's solder for my birthday. (My parents were >>>> very supportive but didn't know much about radio. The debt I owe them has >>>> hopefully been paid to the next generation.) >>>> >>>> Soon after that point I had to save up a month's pocket money plus what I >>>> earned from extra chores to buy parts to build projects with. >>> >>>I built my first computers from "samples" begged from sales reps (while >>>working my way through school). And, from bits of kit gifted to me by friends >>>who worked at firms like DEC (I could never afford to buy even USED kit of this >>>sort AND arrange to pay my tuition, rent, board, etc.) >> >>The first (and only) computer that I actually designed was two boards >>full of TTL. It had three instructions and a 20 KHz 4-phase clock. >> >>Everybody should design a CPU just once, even as a mental activity. > >Look at the original Data General Nova computer, which was designed >around the 74181 ALU chip :-). It was a 4-bit architecture, multiple clocking one ALU to get to 16 bit data and instructions. Steve Wozniak said that he learned digital design by studying the Nova schematics. The Nova, and the HP2114, were basically 16-bit PDP8s; an accumulator or two, paged memory, conditional skips. DEC was going to do that, but one late beer session invented the PDP-11 architecture, which was far superior. I think the PDP-11 inspired the 68K. -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
[toc] | [prev] | [next] | [standalone]
| From | Tauno Voipio <tauno.voipio@notused.fi.invalid> |
|---|---|
| Date | 2017-03-25 13:21 +0200 |
| Message-ID | <ob5jlh$s0q$1@dont-email.me> |
| In reply to | #453448 |
On 24.3.17 23:15, John Larkin wrote: > On Fri, 24 Mar 2017 22:18:48 +0200, upsidedown@downunder.com wrote: > >> On Fri, 24 Mar 2017 09:00:57 -0700, John Larkin >> <jjlarkin@highlandtechnology.com> wrote: >> >>> On Fri, 24 Mar 2017 00:21:50 -0700, Don Y >>> <blockedofcourse@foo.invalid> wrote: >>> >>>> On 3/23/2017 11:08 PM, pcdhobbs@gmail.com wrote: >>>>>> How much did you part with for the kit YOU were using in the mid 70's? >>>>> >>>>> Your reference was from 1970, >>>> >>>> No, the MANUAL that I referenced was from 1970. My reference to the >>>> *equipment* was: >>>> "My first *hard* disk had a capacity of 256K words..." >>>> John Larkin referenced 1970 (feel free to check back up-thread) >>>> >>>>> when I was 10 years old and had just got a >>>>> soldering gun and some plumber's solder for my birthday. (My parents were >>>>> very supportive but didn't know much about radio. The debt I owe them has >>>>> hopefully been paid to the next generation.) >>>>> >>>>> Soon after that point I had to save up a month's pocket money plus what I >>>>> earned from extra chores to buy parts to build projects with. >>>> >>>> I built my first computers from "samples" begged from sales reps (while >>>> working my way through school). And, from bits of kit gifted to me by friends >>>> who worked at firms like DEC (I could never afford to buy even USED kit of this >>>> sort AND arrange to pay my tuition, rent, board, etc.) >>> >>> The first (and only) computer that I actually designed was two boards >>> full of TTL. It had three instructions and a 20 KHz 4-phase clock. >>> >>> Everybody should design a CPU just once, even as a mental activity. >> >> Look at the original Data General Nova computer, which was designed >> around the 74181 ALU chip :-). > > It was a 4-bit architecture, multiple clocking one ALU to get to 16 > bit data and instructions. Steve Wozniak said that he learned digital > design by studying the Nova schematics. > > The Nova, and the HP2114, were basically 16-bit PDP8s; an accumulator > or two, paged memory, conditional skips. DEC was going to do that, but > one late beer session invented the PDP-11 architecture, which was far > superior. I think the PDP-11 inspired the 68K. There were 4 accumulators on the Nova. There are rumors that the Nova was what should be PDP-11, but DEC delayed it to prevent problems to PDP-8 sales, and the guys spun off to found Data General. The PDP-11 did inspire the Motorola chip designers, but it went via the 6800 and 6809 until the 68K. There were 8 general registers on the PDP-11, but the idea was nothing new at that time: already the IBM S/360 had general registers. -- -TV
[toc] | [prev] | [next] | [standalone]
| From | John Larkin <jjlarkin@highlandtechnology.com> |
|---|---|
| Date | 2017-03-25 07:54 -0700 |
| Message-ID | <0i0ddc1dokesujr1qfen6ojbrvb72b2ui6@4ax.com> |
| In reply to | #453543 |
On Sat, 25 Mar 2017 13:21:05 +0200, Tauno Voipio <tauno.voipio@notused.fi.invalid> wrote: >On 24.3.17 23:15, John Larkin wrote: >> On Fri, 24 Mar 2017 22:18:48 +0200, upsidedown@downunder.com wrote: >> >>> On Fri, 24 Mar 2017 09:00:57 -0700, John Larkin >>> <jjlarkin@highlandtechnology.com> wrote: >>> >>>> On Fri, 24 Mar 2017 00:21:50 -0700, Don Y >>>> <blockedofcourse@foo.invalid> wrote: >>>> >>>>> On 3/23/2017 11:08 PM, pcdhobbs@gmail.com wrote: >>>>>>> How much did you part with for the kit YOU were using in the mid 70's? >>>>>> >>>>>> Your reference was from 1970, >>>>> >>>>> No, the MANUAL that I referenced was from 1970. My reference to the >>>>> *equipment* was: >>>>> "My first *hard* disk had a capacity of 256K words..." >>>>> John Larkin referenced 1970 (feel free to check back up-thread) >>>>> >>>>>> when I was 10 years old and had just got a >>>>>> soldering gun and some plumber's solder for my birthday. (My parents were >>>>>> very supportive but didn't know much about radio. The debt I owe them has >>>>>> hopefully been paid to the next generation.) >>>>>> >>>>>> Soon after that point I had to save up a month's pocket money plus what I >>>>>> earned from extra chores to buy parts to build projects with. >>>>> >>>>> I built my first computers from "samples" begged from sales reps (while >>>>> working my way through school). And, from bits of kit gifted to me by friends >>>>> who worked at firms like DEC (I could never afford to buy even USED kit of this >>>>> sort AND arrange to pay my tuition, rent, board, etc.) >>>> >>>> The first (and only) computer that I actually designed was two boards >>>> full of TTL. It had three instructions and a 20 KHz 4-phase clock. >>>> >>>> Everybody should design a CPU just once, even as a mental activity. >>> >>> Look at the original Data General Nova computer, which was designed >>> around the 74181 ALU chip :-). >> >> It was a 4-bit architecture, multiple clocking one ALU to get to 16 >> bit data and instructions. Steve Wozniak said that he learned digital >> design by studying the Nova schematics. >> >> The Nova, and the HP2114, were basically 16-bit PDP8s; an accumulator >> or two, paged memory, conditional skips. DEC was going to do that, but >> one late beer session invented the PDP-11 architecture, which was far >> superior. I think the PDP-11 inspired the 68K. > > >There were 4 accumulators on the Nova. There are rumors that the Nova >was what should be PDP-11, but DEC delayed it to prevent problems to >PDP-8 sales, and the guys spun off to found Data General. That's the story, except that Gordon Bell and Rick Merrill and people then had time (rumored to be one late-night session at Bell's house) to think it over and come up with a new architecture. > >The PDP-11 did inspire the Motorola chip designers, but it went via >the 6800 and 6809 until the 68K. > >There were 8 general registers on the PDP-11, but the idea was nothing >new at that time: already the IBM S/360 had general registers. I think s/360 probably inspired parts of the PDP-11 architecture. The PDP11/20 (I got a very early one) was all TTL, about 510 MSI chips on two big boards. It was all hard logic, no microcode, and was done in the classic DEC hairball logic design style: a tangle of async clocks, one-shots, RCs, tapped delay lines. There were some short instruction sequences that would lock it up hard. Later on somebody taught them about synchronous state machines and microcoding. The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 registers, octal worked. I could once compile instruction to octal by sight; I probably still can do some. It fit beautifully. -- John Larkin Highland Technology, Inc lunatic fringe electronics
[toc] | [prev] | [next] | [standalone]
| From | Don Y <blockedofcourse@foo.invalid> |
|---|---|
| Date | 2017-03-25 10:53 -0700 |
| Message-ID | <ob6alq$74v$1@dont-email.me> |
| In reply to | #453554 |
On 3/25/2017 7:54 AM, John Larkin wrote: > The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 > registers, octal worked. I could once compile instruction to octal by > sight; I probably still can do some. It fit beautifully. This was true of a lot of machines. I used to bit-switch diagnostics into Novas by just writing the code in my head and "assembling" it as the nerve impulses traveled down to my fingertips before hitting the DEP-NEXT toggle. Worked at a place that had *heated* arguments as to whether to "hand-assemble" code for the Z80 in hex or octal (a large swath of the opcode space -- e.g., all the load/store group -- could be easily mapped onto an octal representation). I just shook my head and asked: "Why not let the *machine* remember all those details CORRECTLY and reference things SYMBOLICALLY??" (sigh) People invest in learning <something>, then cling to that OBSOLETE skillset as technology advances (refusing to acknowledge that what they KNEW is no longer worth KNOWING!)
[toc] | [prev] | [next] | [standalone]
| From | upsidedown@downunder.com |
|---|---|
| Date | 2017-03-25 23:11 +0200 |
| Message-ID | <fimddclg0mijti8fqhe4jh6q9acju8cml2@4ax.com> |
| In reply to | #453579 |
On Sat, 25 Mar 2017 10:53:42 -0700, Don Y <blockedofcourse@foo.invalid> wrote: >On 3/25/2017 7:54 AM, John Larkin wrote: >> The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 >> registers, octal worked. I could once compile instruction to octal by >> sight; I probably still can do some. It fit beautifully. I programmed a lot of diagnostic code into the PDP-11 using front panel (binary) switches. That was quite easy. The problem was entering the branch instruction offsets, especially backwards, when you had to calculate the negative offset and had to consider the actual value of the PC, when the branch was taken. > >This was true of a lot of machines. I used to bit-switch diagnostics >into Novas by just writing the code in my head and "assembling" it >as the nerve impulses traveled down to my fingertips before hitting the >DEP-NEXT toggle. The Nova was a 74181 ALU chip with steroids :-) >Worked at a place that had *heated* arguments as to whether to >"hand-assemble" code for the Z80 in hex or octal (a large swath >of the opcode space -- e.g., all the load/store group -- could be >easily mapped onto an octal representation). I just shook >my head and asked: "Why not let the *machine* remember all >those details CORRECTLY and reference things SYMBOLICALLY??" The 8080 (and hence Z80) are basically octal machines. I do not understand why Intel switched to hex after initially using octals. > >(sigh) People invest in learning <something>, then cling to that >OBSOLETE skillset as technology advances (refusing to acknowledge >that what they KNEW is no longer worth KNOWING!)
[toc] | [prev] | [next] | [standalone]
| From | Don Y <blockedofcourse@foo.invalid> |
|---|---|
| Date | 2017-03-25 15:14 -0700 |
| Message-ID | <ob6pue$om4$1@dont-email.me> |
| In reply to | #453608 |
On 3/25/2017 2:11 PM, upsidedown@downunder.com wrote: > On Sat, 25 Mar 2017 10:53:42 -0700, Don Y > <blockedofcourse@foo.invalid> wrote: > >> On 3/25/2017 7:54 AM, John Larkin wrote: >>> The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 >>> registers, octal worked. I could once compile instruction to octal by >>> sight; I probably still can do some. It fit beautifully. > > I programmed a lot of diagnostic code into the PDP-11 using front > panel (binary) switches. That was quite easy. The problem was entering > the branch instruction offsets, especially backwards, when you had to > calculate the negative offset and had to consider the actual value of > the PC, when the branch was taken. And, remembering "specialty" opcodes (e.g., INTDS) that didn't fit a nice "decoding pattern". >> This was true of a lot of machines. I used to bit-switch diagnostics >> into Novas by just writing the code in my head and "assembling" it >> as the nerve impulses traveled down to my fingertips before hitting the >> DEP-NEXT toggle. > > The Nova was a 74181 ALU chip with steroids :-) > >> Worked at a place that had *heated* arguments as to whether to >> "hand-assemble" code for the Z80 in hex or octal (a large swath >> of the opcode space -- e.g., all the load/store group -- could be >> easily mapped onto an octal representation). I just shook >> my head and asked: "Why not let the *machine* remember all >> those details CORRECTLY and reference things SYMBOLICALLY??" > > The 8080 (and hence Z80) are basically octal machines. I do not > understand why Intel switched to hex after initially using octals. How is it -- or any machine -- an "octal" machine vs. a hexadecimal machine vs. a binary machine vs. ...? Does it process data in groups of three bits -- instead of four? or 8? or 1? Just because the opcodes were chosen to organize certain "selectors" (operations, registers, conditions, etc.) in particular groupings doesn't really mean anything. The bits in the opcode could have been arbitrarily scrambled -- the instruction decoder doesn't care where the bits are "physically" located. By the time you could purchase a "computer on a chip", the technology (including toolchains) were affordable enough that the idea of "hand assembling" code was just silly. Like refusing to use a calculator because you spent so much time learning how to do "long division" as a child... :< [You see the same sort of thing with HLL programmers trying to hand-optimize code -- that the compiler can usually do *better* and more accurately]
[toc] | [prev] | [next] | [standalone]
| From | upsidedown@downunder.com |
|---|---|
| Date | 2017-03-26 12:18 +0300 |
| Message-ID | <0d1fdchip2tev0pjl2gfjp8bd1qnqkko4e@4ax.com> |
| In reply to | #453615 |
On Sat, 25 Mar 2017 15:14:19 -0700, Don Y <blockedofcourse@foo.invalid> wrote: >On 3/25/2017 2:11 PM, upsidedown@downunder.com wrote: >> On Sat, 25 Mar 2017 10:53:42 -0700, Don Y >> <blockedofcourse@foo.invalid> wrote: >> >>> On 3/25/2017 7:54 AM, John Larkin wrote: >>>> The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 >>>> registers, octal worked. I could once compile instruction to octal by >>>> sight; I probably still can do some. It fit beautifully. >> >> I programmed a lot of diagnostic code into the PDP-11 using front >> panel (binary) switches. That was quite easy. The problem was entering >> the branch instruction offsets, especially backwards, when you had to >> calculate the negative offset and had to consider the actual value of >> the PC, when the branch was taken. > >And, remembering "specialty" opcodes (e.g., INTDS) that didn't fit a nice >"decoding pattern". > >>> This was true of a lot of machines. I used to bit-switch diagnostics >>> into Novas by just writing the code in my head and "assembling" it >>> as the nerve impulses traveled down to my fingertips before hitting the >>> DEP-NEXT toggle. >> >> The Nova was a 74181 ALU chip with steroids :-) >> >>> Worked at a place that had *heated* arguments as to whether to >>> "hand-assemble" code for the Z80 in hex or octal (a large swath >>> of the opcode space -- e.g., all the load/store group -- could be >>> easily mapped onto an octal representation). I just shook >>> my head and asked: "Why not let the *machine* remember all >>> those details CORRECTLY and reference things SYMBOLICALLY??" >> >> The 8080 (and hence Z80) are basically octal machines. I do not >> understand why Intel switched to hex after initially using octals. > >How is it -- or any machine -- an "octal" machine vs. a hexadecimal >machine vs. a binary machine vs. ...? Does it process data in groups >of three bits -- instead of four? or 8? or 1? Then there were also decimal computers used for accounting etc. I once wrote a cross assembler for a Philips machine with exactly 1000 memory words, each of which IIRC was 10 decimal digits
[toc] | [prev] | [next] | [standalone]
| From | Don Y <blockedofcourse@foo.invalid> |
|---|---|
| Date | 2017-03-26 03:33 -0700 |
| Message-ID | <ob8598$raa$1@dont-email.me> |
| In reply to | #453718 |
On 3/26/2017 2:18 AM, upsidedown@downunder.com wrote: > On Sat, 25 Mar 2017 15:14:19 -0700, Don Y > <blockedofcourse@foo.invalid> wrote: > >> On 3/25/2017 2:11 PM, upsidedown@downunder.com wrote: >>> On Sat, 25 Mar 2017 10:53:42 -0700, Don Y >>> <blockedofcourse@foo.invalid> wrote: >>> >>>> On 3/25/2017 7:54 AM, John Larkin wrote: >>>>> The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 >>>>> registers, octal worked. I could once compile instruction to octal by >>>>> sight; I probably still can do some. It fit beautifully. >>> >>> I programmed a lot of diagnostic code into the PDP-11 using front >>> panel (binary) switches. That was quite easy. The problem was entering >>> the branch instruction offsets, especially backwards, when you had to >>> calculate the negative offset and had to consider the actual value of >>> the PC, when the branch was taken. >> >> And, remembering "specialty" opcodes (e.g., INTDS) that didn't fit a nice >> "decoding pattern". >> >>>> This was true of a lot of machines. I used to bit-switch diagnostics >>>> into Novas by just writing the code in my head and "assembling" it >>>> as the nerve impulses traveled down to my fingertips before hitting the >>>> DEP-NEXT toggle. >>> >>> The Nova was a 74181 ALU chip with steroids :-) >>> >>>> Worked at a place that had *heated* arguments as to whether to >>>> "hand-assemble" code for the Z80 in hex or octal (a large swath >>>> of the opcode space -- e.g., all the load/store group -- could be >>>> easily mapped onto an octal representation). I just shook >>>> my head and asked: "Why not let the *machine* remember all >>>> those details CORRECTLY and reference things SYMBOLICALLY??" >>> >>> The 8080 (and hence Z80) are basically octal machines. I do not >>> understand why Intel switched to hex after initially using octals. >> >> How is it -- or any machine -- an "octal" machine vs. a hexadecimal >> machine vs. a binary machine vs. ...? Does it process data in groups >> of three bits -- instead of four? or 8? or 1? > > Then there were also decimal computers used for accounting etc. > I once wrote a cross assembler for a Philips machine with exactly > 1000 memory words, each of which IIRC was 10 decimal digits But you don't need a "decimal computer" to do decimal arithmetic. And, an "octal computer" doesn't do "octal arithmetic" -- any more than it does "hexadecimal arithmetic".
[toc] | [prev] | [next] | [standalone]
| From | Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> |
|---|---|
| Date | 2017-03-26 16:36 -0400 |
| Message-ID | <fK6dnRuHc85iuEXFnZ2dnUU7-XXNnZ2d@supernews.com> |
| In reply to | #453615 |
On 03/25/2017 06:14 PM, Don Y wrote: > On 3/25/2017 2:11 PM, upsidedown@downunder.com wrote: >> On Sat, 25 Mar 2017 10:53:42 -0700, Don Y >> <blockedofcourse@foo.invalid> wrote: >> >>> On 3/25/2017 7:54 AM, John Larkin wrote: >>>> The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 >>>> registers, octal worked. I could once compile instruction to octal by >>>> sight; I probably still can do some. It fit beautifully. >> >> I programmed a lot of diagnostic code into the PDP-11 using front >> panel (binary) switches. That was quite easy. The problem was entering >> the branch instruction offsets, especially backwards, when you had to >> calculate the negative offset and had to consider the actual value of >> the PC, when the branch was taken. > > And, remembering "specialty" opcodes (e.g., INTDS) that didn't fit a nice > "decoding pattern". > >>> This was true of a lot of machines. I used to bit-switch diagnostics >>> into Novas by just writing the code in my head and "assembling" it >>> as the nerve impulses traveled down to my fingertips before hitting the >>> DEP-NEXT toggle. >> >> The Nova was a 74181 ALU chip with steroids :-) >> >>> Worked at a place that had *heated* arguments as to whether to >>> "hand-assemble" code for the Z80 in hex or octal (a large swath >>> of the opcode space -- e.g., all the load/store group -- could be >>> easily mapped onto an octal representation). I just shook >>> my head and asked: "Why not let the *machine* remember all >>> those details CORRECTLY and reference things SYMBOLICALLY??" >> >> The 8080 (and hence Z80) are basically octal machines. I do not >> understand why Intel switched to hex after initially using octals. > > How is it -- or any machine -- an "octal" machine vs. a hexadecimal > machine vs. a binary machine vs. ...? Does it process data in groups > of three bits -- instead of four? or 8? or 1? A machine with 12-bit words, such as a PDP-8, is naturally amenable to octal. Some early MPU development systems, such as one for the 6800 that I remember from 1975ish, used octal, but it's a poor fit for word lengths that are powers of 2. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
[toc] | [prev] | [next] | [standalone]
| From | John Larkin <jjlarkin@highlandtechnology.com> |
|---|---|
| Date | 2017-03-26 14:13 -0700 |
| Message-ID | <3magdcln9gkcd0hrm712ris7638klfk7ua@4ax.com> |
| In reply to | #453801 |
On Sun, 26 Mar 2017 16:36:53 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote: >On 03/25/2017 06:14 PM, Don Y wrote: >> On 3/25/2017 2:11 PM, upsidedown@downunder.com wrote: >>> On Sat, 25 Mar 2017 10:53:42 -0700, Don Y >>> <blockedofcourse@foo.invalid> wrote: >>> >>>> On 3/25/2017 7:54 AM, John Larkin wrote: >>>>> The PDP-11 was a 16-bit machine programmed in octal. Because it had 8 >>>>> registers, octal worked. I could once compile instruction to octal by >>>>> sight; I probably still can do some. It fit beautifully. >>> >>> I programmed a lot of diagnostic code into the PDP-11 using front >>> panel (binary) switches. That was quite easy. The problem was entering >>> the branch instruction offsets, especially backwards, when you had to >>> calculate the negative offset and had to consider the actual value of >>> the PC, when the branch was taken. >> >> And, remembering "specialty" opcodes (e.g., INTDS) that didn't fit a nice >> "decoding pattern". >> >>>> This was true of a lot of machines. I used to bit-switch diagnostics >>>> into Novas by just writing the code in my head and "assembling" it >>>> as the nerve impulses traveled down to my fingertips before hitting the >>>> DEP-NEXT toggle. >>> >>> The Nova was a 74181 ALU chip with steroids :-) >>> >>>> Worked at a place that had *heated* arguments as to whether to >>>> "hand-assemble" code for the Z80 in hex or octal (a large swath >>>> of the opcode space -- e.g., all the load/store group -- could be >>>> easily mapped onto an octal representation). I just shook >>>> my head and asked: "Why not let the *machine* remember all >>>> those details CORRECTLY and reference things SYMBOLICALLY??" >>> >>> The 8080 (and hence Z80) are basically octal machines. I do not >>> understand why Intel switched to hex after initially using octals. >> >> How is it -- or any machine -- an "octal" machine vs. a hexadecimal >> machine vs. a binary machine vs. ...? Does it process data in groups >> of three bits -- instead of four? or 8? or 1? > > >A machine with 12-bit words, such as a PDP-8, is naturally amenable to >octal. Some early MPU development systems, such as one for the 6800 >that I remember from 1975ish, used octal, but it's a poor fit for word >lengths that are powers of 2. > >Cheers > >Phil Hobbs The PDP-11 double operand 16-bit instruction bits were B OOO SSS MMM DDD MMM B = byte / word operation OOO = opcode SSS = source register DDD = dest register MMM = mode. which breaks up beautifully into octal. There were eight registers, R6 being the stack pointer and R7 the program counter. Indexing off registers (including PC and SP), autoinc/autodec, absolute/relative addressing, position-independent code... all were obvious. 012737 became "move immediate xxxxxx to absolute destination" xxxxxx namely MOV #xxxxx, @#ddddd or MOV (pc)+, @(pc)+ dddddd 014747 is "copy this instruction one word below itself and execute." You could do fun stuff like XOR R2, PC or MOV wombat(R3), PC And then we wound up with the i86 instruction set. -- John Larkin Highland Technology, Inc lunatic fringe electronics
[toc] | [prev] | [next] | [standalone]
| From | Tauno Voipio <tauno.voipio@notused.fi.invalid> |
|---|---|
| Date | 2017-03-25 21:04 +0200 |
| Message-ID | <ob6eq1$l0f$1@dont-email.me> |
| In reply to | #453554 |
On 25.3.17 16:54, John Larkin wrote: > The PDP11/20 (I got a very early one) was all TTL, about 510 MSI chips > on two big boards. It was all hard logic, no microcode, and was done > in the classic DEC hairball logic design style: a tangle of async > clocks, one-shots, RCs, tapped delay lines. There were some short > instruction sequences that would lock it up hard. Later on somebody > taught them about synchronous state machines and microcoding. I was a dozen years ago called to repair one, as it seems that there were no other people left who could tackle the task. The DEC logic made it quite a challenge, and the sick part was a 7400, which was replaced with a 74LS00 and the paper machine continued running. -- -TV
[toc] | [prev] | [next] | [standalone]
| From | pcdhobbs@gmail.com |
|---|---|
| Date | 2017-03-24 19:03 -0700 |
| Message-ID | <0993ac70-3d79-484e-9818-ac48295d36ef@googlegroups.com> |
| In reply to | #453438 |
>Look at the original Data General Nova computer, which was designed >around the 74181 ALU chip :-). I had to write a simple Laplace's equation simulator for one of those in a physics lab class back in about 1978-9. It had a 256K split-core multiprogramming facility, so that it could run a couple of students' code at once. The next year I had an undergraduate research assistantship doing Monte Carlo simulations of radiative transfer in interstellar giant molecular clouds, running Fortran-77 on an Amdahl 470-V8 under the Michigan Terminal System. (Of course that's resume-speak--I was scrambling to learn Fortran and debug this Danish astronomer's code at the same time. It did eventually work, and provided some interesting results if radio astronomy is your thing.) Cheers Phil Hobbs
[toc] | [prev] | [next] | [standalone]
Page 5 of 6 — ← Prev page 1 2 3 4 [5] 6 Next page →
Back to top | Article view | sci.electronics.design
csiph-web