|
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387086 is a reply to message #387084] |
Tue, 17 September 2019 09:00 |
|
Originally posted by: Bob Eager
On Tue, 17 Sep 2019 13:21:47 +0100, Ahem A Rivet's Shot wrote:
> On 17 Sep 2019 08:11:09 GMT Bob Eager <news0073@eager.cx> wrote:
>
>> There are others, more useful than just teaching (which is pretty
>> useful).
>>
>> It's been common (since the 1960s) to build compilers that initially
>> generate code for a fictitious machine that has a convenient
>> architecture. Then all one has to do is write a code generator that
>> converts that code into code for a real target machine - much less
>> work.
>>
>> https://en.wikipedia.org/wiki/BCPL
>
> See also Pascal P-code. I've never been quite sure which came
first
> P-code or Int-code or indeed whether there was a common conceptual
> ancestor somewhere.
BCPL 'OCODE' was around in the mid 1960s, pre-dating P-code by a long way
(Pascal didn't even appear until 1970).
BCPL Intcode was developed (as far as I recall) in the early 1980s, not
as an abstract machine for code generation but as an interpretive code.
It appeared most widely at that time on the BBC microcomputer.
The last time I looked at a BCPL compiler generating Intcode, it was
actually generating OCODE using the standard compiler, then generating
Intcode from that.
The conceptual ancestor may well have been OCODE, but neither P-code nor
Intcode were really meant as an intermediate step in compilation. They
were for imaginary machines, though!
These days we have LLVM, which is considerably more complex.
--
Using UNIX since v6 (1975)...
Use the BIG mirror service in the UK:
http://www.mirrorservice.org
|
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387095 is a reply to message #387086] |
Tue, 17 September 2019 10:59 |
Ahem A Rivet's Shot
Messages: 4843 Registered: January 2012
Karma: 0
|
Senior Member |
|
|
On 17 Sep 2019 13:00:26 GMT
Bob Eager <news0073@eager.cx> wrote:
> On Tue, 17 Sep 2019 13:21:47 +0100, Ahem A Rivet's Shot wrote:
>
>> On 17 Sep 2019 08:11:09 GMT Bob Eager <news0073@eager.cx> wrote:
>>
>>> There are others, more useful than just teaching (which is pretty
>>> useful).
>>>
>>> It's been common (since the 1960s) to build compilers that initially
>>> generate code for a fictitious machine that has a convenient
>>> architecture. Then all one has to do is write a code generator that
>>> converts that code into code for a real target machine - much less
>>> work.
>>>
>>> https://en.wikipedia.org/wiki/BCPL
>>
>> See also Pascal P-code. I've never been quite sure which came
> first
>> P-code or Int-code or indeed whether there was a common conceptual
>> ancestor somewhere.
>
> BCPL 'OCODE' was around in the mid 1960s, pre-dating P-code by a long way
> (Pascal didn't even appear until 1970).
>
> BCPL Intcode was developed (as far as I recall) in the early 1980s, not
> as an abstract machine for code generation but as an interpretive code.
> It appeared most widely at that time on the BBC microcomputer.
Hmmm I'm pretty sure it was Intcode that the BCPL porting kit I had
in from Martin Richards in 1980 (well before the BBC micro - the Proton
was on the shelf and we were still trying to get the Newbrain done) used for
an intermediate. The kit included the source for the compiler to Intcode
with a hook to add a native code translation pass and the Intcode compile
version. The idea was that you first wrote an Intcode interpreter for the
new platform and validated by using it to compile the compiler to Intcode -
then you wrote the native translator and ultimately tested it by using the
native compiler to compile itself.
--
Steve O'Hara-Smith | Directable Mirror Arrays
C:\>WIN | A better way to focus the sun
The computer obeys and wins. | licences available see
You lose and Bill collects. | http://www.sohara.org/
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387097 is a reply to message #387095] |
Tue, 17 September 2019 12:05 |
|
Originally posted by: Bob Eager
On Tue, 17 Sep 2019 15:59:22 +0100, Ahem A Rivet's Shot wrote:
> On 17 Sep 2019 13:00:26 GMT Bob Eager <news0073@eager.cx> wrote:
>
>> On Tue, 17 Sep 2019 13:21:47 +0100, Ahem A Rivet's Shot wrote:
>>
>>> On 17 Sep 2019 08:11:09 GMT Bob Eager <news0073@eager.cx> wrote:
>>>
>>>> There are others, more useful than just teaching (which is pretty
>>>> useful).
>>>>
>>>> It's been common (since the 1960s) to build compilers that initially
>>>> generate code for a fictitious machine that has a convenient
>>>> architecture. Then all one has to do is write a code generator that
>>>> converts that code into code for a real target machine - much less
>>>> work.
>>>>
>>>> https://en.wikipedia.org/wiki/BCPL
>>>
>>> See also Pascal P-code. I've never been quite sure which came
>> first
>>> P-code or Int-code or indeed whether there was a common conceptual
>>> ancestor somewhere.
>>
>> BCPL 'OCODE' was around in the mid 1960s, pre-dating P-code by a long
>> way (Pascal didn't even appear until 1970).
>>
>> BCPL Intcode was developed (as far as I recall) in the early 1980s, not
>> as an abstract machine for code generation but as an interpretive code.
>> It appeared most widely at that time on the BBC microcomputer.
>
> Hmmm I'm pretty sure it was Intcode that the BCPL porting kit I
had
> in from Martin Richards in 1980 (well before the BBC micro - the Proton
> was on the shelf and we were still trying to get the Newbrain done) used
> for an intermediate. The kit included the source for the compiler to
> Intcode with a hook to add a native code translation pass and the
> Intcode compile version. The idea was that you first wrote an Intcode
> interpreter for the new platform and validated by using it to compile
> the compiler to Intcode - then you wrote the native translator and
> ultimately tested it by using the native compiler to compile itself.
That's right. I may be a year or three out there. But the compiler still
generated OCODE first, I believe, then generated the Intcode. So the
translator the end user wrote was from OCODE to Intcode.
--
Using UNIX since v6 (1975)...
Use the BIG mirror service in the UK:
http://www.mirrorservice.org
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387098 is a reply to message #387095] |
Tue, 17 September 2019 12:06 |
|
Originally posted by: Bob Eager
On Tue, 17 Sep 2019 15:59:22 +0100, Ahem A Rivet's Shot wrote:
>> BCPL 'OCODE' was around in the mid 1960s, pre-dating P-code by a long
>> way (Pascal didn't even appear until 1970).
>>
>> BCPL Intcode was developed (as far as I recall) in the early 1980s, not
>> as an abstract machine for code generation but as an interpretive code.
>> It appeared most widely at that time on the BBC microcomputer.
>
> Hmmm I'm pretty sure it was Intcode that the BCPL porting kit I
had
> in from Martin Richards in 1980 (well before the BBC micro - the Proton
> was on the shelf and we were still trying to get the Newbrain done) used
> for an intermediate. The kit included the source for the compiler to
> Intcode with a hook to add a native code translation pass and the
> Intcode compile version. The idea was that you first wrote an Intcode
> interpreter for the new platform and validated by using it to compile
> the compiler to Intcode - then you wrote the native translator and
> ultimately tested it by using the native compiler to compile itself.
I should have said: I have an Intcode interpreter in C somewhere...
--
Using UNIX since v6 (1975)...
Use the BIG mirror service in the UK:
http://www.mirrorservice.org
|
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387102 is a reply to message #387099] |
Tue, 17 September 2019 14:59 |
|
Originally posted by: JimP
On Tue, 17 Sep 2019 16:58:18 GMT, scott@slp53.sl.home (Scott Lurndal)
wrote:
> Peter Flass <peter_flass@yahoo.com> writes:
>> Andrew Swallow <am.swallow@btinternet.com> wrote:
>>> On 17/09/2019 07:08, Mike Spencer wrote:
>>>> Bob Eager <news0073@eager.cx> writes:
>>>>
>>>> > On Sat, 07 Sep 2019 02:42:40 -0300, Mike Spencer wrote:
>>>> >
>>>> >> Tell us more about these "fictitious machines".
>>>> >
>>>> > https://en.wikipedia.org/wiki/MMIX
>>>>
>>>> Ah, so. Tnx.
>>>>
>>>>
>>> An electronics company that wants its own range of chips could use the
>>> MMIX instruction set.
>>>
>>
>> Iâd be happy enough to see the 68K revived as a general-purpose chip.
>
> Why? It's so far behind the technology curve that it would never catch up.
>
> In any case, only compilers-writers care about the instruction set; nobody
> writes application code in assembler anymore.
Well, the Amiga computer could use a slightly improved 68k chip. Even
some new accelerators. I know some accelerators are being made in
England and Europe. Not sure how much of a bite I would have to pay on
import duties and shipping.
--
Jim
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387105 is a reply to message #387102] |
Tue, 17 September 2019 15:11 |
Peter Flass
Messages: 8375 Registered: December 2011
Karma: 0
|
Senior Member |
|
|
JimP <solosam90@gmail.com> wrote:
> On Tue, 17 Sep 2019 16:58:18 GMT, scott@slp53.sl.home (Scott Lurndal)
> wrote:
>> Peter Flass <peter_flass@yahoo.com> writes:
>>> Andrew Swallow <am.swallow@btinternet.com> wrote:
>>>> On 17/09/2019 07:08, Mike Spencer wrote:
>>>> > Bob Eager <news0073@eager.cx> writes:
>>>> >
>>>> >> On Sat, 07 Sep 2019 02:42:40 -0300, Mike Spencer wrote:
>>>> >>
>>>> >>> Tell us more about these "fictitious machines".
>>>> >>
>>>> >> https://en.wikipedia.org/wiki/MMIX
>>>> >
>>>> > Ah, so. Tnx.
>>>> >
>>>> >
>>>> An electronics company that wants its own range of chips could use the
>>>> MMIX instruction set.
>>>>
>>>
>>> Iâd be happy enough to see the 68K revived as a general-purpose chip.
>>
>> Why? It's so far behind the technology curve that it would never catch up.
>>
>> In any case, only compilers-writers care about the instruction set; nobody
>> writes application code in assembler anymore.
>
> Well, the Amiga computer could use a slightly improved 68k chip. Even
> some new accelerators. I know some accelerators are being made in
> England and Europe. Not sure how much of a bite I would have to pay on
> import duties and shipping.
>
ARM came from being an oddity to a mainstream chip. The 68K had an
orthogonal instruction set that could be easily expanded, and most of the
effort these days seems to go into the microcode under the hood.
--
Pete
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387106 is a reply to message #387097] |
Tue, 17 September 2019 15:09 |
Ahem A Rivet's Shot
Messages: 4843 Registered: January 2012
Karma: 0
|
Senior Member |
|
|
On 17 Sep 2019 16:05:38 GMT
Bob Eager <news0073@eager.cx> wrote:
> On Tue, 17 Sep 2019 15:59:22 +0100, Ahem A Rivet's Shot wrote:
>
>> On 17 Sep 2019 13:00:26 GMT Bob Eager <news0073@eager.cx> wrote:
>>
>>> On Tue, 17 Sep 2019 13:21:47 +0100, Ahem A Rivet's Shot wrote:
>>>
>>>> On 17 Sep 2019 08:11:09 GMT Bob Eager <news0073@eager.cx> wrote:
>>>>
>>>> > There are others, more useful than just teaching (which is pretty
>>>> > useful).
>>>> >
>>>> > It's been common (since the 1960s) to build compilers that initially
>>>> > generate code for a fictitious machine that has a convenient
>>>> > architecture. Then all one has to do is write a code generator that
>>>> > converts that code into code for a real target machine - much less
>>>> > work.
>>>> >
>>>> > https://en.wikipedia.org/wiki/BCPL
>>>>
>>>> See also Pascal P-code. I've never been quite sure which came
>>> first
>>>> P-code or Int-code or indeed whether there was a common conceptual
>>>> ancestor somewhere.
>>>
>>> BCPL 'OCODE' was around in the mid 1960s, pre-dating P-code by a long
>>> way (Pascal didn't even appear until 1970).
>>>
>>> BCPL Intcode was developed (as far as I recall) in the early 1980s, not
>>> as an abstract machine for code generation but as an interpretive code.
>>> It appeared most widely at that time on the BBC microcomputer.
>>
>> Hmmm I'm pretty sure it was Intcode that the BCPL porting kit I
> had
>> in from Martin Richards in 1980 (well before the BBC micro - the Proton
>> was on the shelf and we were still trying to get the Newbrain done) used
>> for an intermediate. The kit included the source for the compiler to
>> Intcode with a hook to add a native code translation pass and the
>> Intcode compile version. The idea was that you first wrote an Intcode
>> interpreter for the new platform and validated by using it to compile
>> the compiler to Intcode - then you wrote the native translator and
>> ultimately tested it by using the native compiler to compile itself.
>
> That's right. I may be a year or three out there. But the compiler still
> generated OCODE first, I believe, then generated the Intcode.
Possibly, but the only output was Intcode, unless you added the
final stage translator.
> So the translator the end user wrote was from OCODE to Intcode.
No the translator was from Intcode to native, I wrote one once.
--
Steve O'Hara-Smith | Directable Mirror Arrays
C:\>WIN | A better way to focus the sun
The computer obeys and wins. | licences available see
You lose and Bill collects. | http://www.sohara.org/
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387108 is a reply to message #387105] |
Tue, 17 September 2019 15:36 |
scott
Messages: 4237 Registered: February 2012
Karma: 0
|
Senior Member |
|
|
Peter Flass <peter_flass@yahoo.com> writes:
> JimP <solosam90@gmail.com> wrote:
>> On Tue, 17 Sep 2019 16:58:18 GMT, scott@slp53.sl.home (Scott Lurndal)
>> wrote:
>>> Peter Flass <peter_flass@yahoo.com> writes:
>>>> Andrew Swallow <am.swallow@btinternet.com> wrote:
>>>> > On 17/09/2019 07:08, Mike Spencer wrote:
>>>> >> Bob Eager <news0073@eager.cx> writes:
>>>> >>
>>>> >>> On Sat, 07 Sep 2019 02:42:40 -0300, Mike Spencer wrote:
>>>> >>>
>>>> >>>> Tell us more about these "fictitious machines".
>>>> >>>
>>>> >>> https://en.wikipedia.org/wiki/MMIX
>>>> >>
>>>> >> Ah, so. Tnx.
>>>> >>
>>>> >>
>>>> > An electronics company that wants its own range of chips could use the
>>>> > MMIX instruction set.
>>>> >
>>>>
>>>> Iâd be happy enough to see the 68K revived as a general-purpose chip.
>>>
>>> Why? It's so far behind the technology curve that it would never catch up.
>>>
>>> In any case, only compilers-writers care about the instruction set; nobody
>>> writes application code in assembler anymore.
>>
>> Well, the Amiga computer could use a slightly improved 68k chip. Even
>> some new accelerators. I know some accelerators are being made in
>> England and Europe. Not sure how much of a bite I would have to pay on
>> import duties and shipping.
>>
>
> ARM came from being an oddity to a mainstream chip. The 68K had an
> orthogonal instruction set that could be easily expanded, and most of the
> effort these days seems to go into the microcode under the hood.
ARM was never an "oddity". It was designed for the Acorn Risc Machine specifically
_because the 68k was not suitable_.
They introduced the 64-bit version circa 2012.
ARM processors do not generally have "microcode under the hood" (at least in the
processing core - they may have microcode in on-chip accelerators and
peripheral devices).
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387110 is a reply to message #387108] |
Tue, 17 September 2019 18:49 |
Ahem A Rivet's Shot
Messages: 4843 Registered: January 2012
Karma: 0
|
Senior Member |
|
|
On Tue, 17 Sep 2019 19:36:22 GMT
scott@slp53.sl.home (Scott Lurndal) wrote:
> Peter Flass <peter_flass@yahoo.com> writes:
>> ARM came from being an oddity to a mainstream chip. The 68K had an
>> orthogonal instruction set that could be easily expanded, and most of the
>> effort these days seems to go into the microcode under the hood.
>
> ARM was never an "oddity". It was designed for the Acorn Risc Machine
> specifically _because the 68k was not suitable_.
More like because someone thought there was a better way and was
convincing about it, and right. What few seem to remember is how impressive
the performance of the first ARM machines was.
--
Steve O'Hara-Smith | Directable Mirror Arrays
C:\>WIN | A better way to focus the sun
The computer obeys and wins. | licences available see
You lose and Bill collects. | http://www.sohara.org/
|
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387120 is a reply to message #387110] |
Wed, 18 September 2019 08:30 |
scott
Messages: 4237 Registered: February 2012
Karma: 0
|
Senior Member |
|
|
Ahem A Rivet's Shot <steveo@eircom.net> writes:
> On Tue, 17 Sep 2019 19:36:22 GMT
> scott@slp53.sl.home (Scott Lurndal) wrote:
>
>> Peter Flass <peter_flass@yahoo.com> writes:
>
>>> ARM came from being an oddity to a mainstream chip. The 68K had an
>>> orthogonal instruction set that could be easily expanded, and most of the
>>> effort these days seems to go into the microcode under the hood.
>>
>> ARM was never an "oddity". It was designed for the Acorn Risc Machine
>> specifically _because the 68k was not suitable_.
>
> More like because someone thought there was a better way and was
The problem, as I understand it, was because some 68k instructions could
take an arbitrary number of cycles, which would delay interrupts. Workloads
that worked on the 6502 wouldn't on the 68k due to excessive interrupt latency.
Acorn looked at all the existing microprocessors before deciding to roll their own.
> convincing about it, and right. What few seem to remember is how impressive
> the performance of the first ARM machines was.
|
|
|
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387125 is a reply to message #387121] |
Wed, 18 September 2019 09:48 |
scott
Messages: 4237 Registered: February 2012
Karma: 0
|
Senior Member |
|
|
Bob Eager <news0073@eager.cx> writes:
> git.gnome.org/pan2)
> X-Received-Bytes: 1805
> X-Received-Body-CRC: 2907341646
>
> On Wed, 18 Sep 2019 12:30:10 +0000, Scott Lurndal wrote:
>
>> The problem, as I understand it, was because some 68k instructions could
>> take an arbitrary number of cycles, which would delay interrupts.
>> Workloads that worked on the 6502 wouldn't on the 68k due to excessive
>> interrupt latency
>
> Interesting. The VAX seemed to solve that problem by having interruptable
> instructions.
With MOVC3/MOVC5, they could just update the registers and tell the
interrupt procedure to return to the MOVC instruction, if I recall correctly.
Not sure if the same could be true of the long instructions on 68K.
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387127 is a reply to message #387125] |
Wed, 18 September 2019 09:53 |
|
Originally posted by: Bob Eager
On Wed, 18 Sep 2019 13:48:03 +0000, Scott Lurndal wrote:
> Bob Eager <news0073@eager.cx> writes:
>> git.gnome.org/pan2)
>> X-Received-Bytes: 1805 X-Received-Body-CRC: 2907341646
>>
>> On Wed, 18 Sep 2019 12:30:10 +0000, Scott Lurndal wrote:
>>
>>> The problem, as I understand it, was because some 68k instructions
>>> could take an arbitrary number of cycles, which would delay
>>> interrupts. Workloads that worked on the 6502 wouldn't on the 68k due
>>> to excessive interrupt latency
>>
>> Interesting. The VAX seemed to solve that problem by having
>> interruptable instructions.
>
> With MOVC3/MOVC5, they could just update the registers and tell the
> interrupt procedure to return to the MOVC instruction, if I recall
> correctly.
That's right; the PSW contained a First Part Done bit (FPD) which said
that it was a partially completed instruction.
--
Using UNIX since v6 (1975)...
Use the BIG mirror service in the UK:
http://www.mirrorservice.org
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387150 is a reply to message #387039] |
Fri, 20 September 2019 13:42 |
|
Originally posted by: googlegroups jmfbahciv
On Saturday, September 14, 2019 at 1:56:02 PM UTC-4, Gerard Schildberger wrote:
> On Saturday, September 14, 2019 at 11:57:41 AM UTC-5, googlegroups jmfbahciv wrote:
>> On Friday, September 6, 2019 at 2:38:18 AM UTC-4, James Dow Allen wrote:
>>> On Sunday, November 4, 2018 at 1:56:06 AM UTC+7, françai s wrote:
>>>> All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died?
>>>
>>> I programmed the CDC 6400/6600 in the late 1960's. Mostly Fortran, but a fair amount of Compass Assembly code also.
>>>
>>> The earliest machine code for me was a very simple program for IBM 1620 in '65 or '66. I didn't write it in an assembly language -- I composed it in "absolute" machine code, using the Multiple-punch option on a card punch! (During the 1970's I developed machine code for about a dozen different machines.)
>>
>>
>> I did that (and then SPS) in college. Then I went to
>> PDP-10.
>
> Which SPS (Symbolic Programming System) was that?
The 1620.
I was responding to the mention of the 1620 as the OP's first
system.
/BAH
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387295 is a reply to message #387074] |
Mon, 23 September 2019 10:49 |
|
Originally posted by: pcanagnostopoulos
On Tuesday, September 17, 2019 at 2:27:30 AM UTC-4, Mike Spencer wrote:
> pcanagnostopoulos@gmail.com writes:
>
>> On Saturday, September 7, 2019 at 1:42:58 AM UTC-4, Mike Spencer wrote:
>>
>>> pcanagnostopoulos@gmail.com writes:
>>>
>>>> Now I create emulators for fictitious machines and then write
>>>> assembler code for them.=20
>>>
>>> Tell us more about these "fictitious machines".
>>
>> I invented the fictitious 1960s-era company International Computing
>> Machines (ICM). About once a year I get the yearning to design a new
>> machine that they would have built. Sometimes it's simple and other
>> times more complex. I write a Principle of Operations manual for
>> it. Then I implement a simplistic visual representation and an
>> emulator for the CPU and peripherals.
>>
>> Then I write software for the machine using my Assembler J and Linker
>> J, which are a cross-assembler and cross-linker. Sometimes I don't get
>> very far, other times I implement a piece of an operating system. Then
>> I get sidetracked by real work and development stops.
>
> You are a crazy person. Excellent. Much better answer than MMIX.
You are not the first person to think I'm crazy. However, you may be the first to think it's crazy in a good way!
~~ Paul
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #387308 is a reply to message #387295] |
Mon, 23 September 2019 22:26 |
|
Originally posted by: J. Clarke
On Mon, 23 Sep 2019 07:49:16 -0700 (PDT), pcanagnostopoulos@gmail.com
wrote:
> On Tuesday, September 17, 2019 at 2:27:30 AM UTC-4, Mike Spencer wrote:
>> pcanagnostopoulos@gmail.com writes:
>>
>>> On Saturday, September 7, 2019 at 1:42:58 AM UTC-4, Mike Spencer wrote:
>>>
>>>> pcanagnostopoulos@gmail.com writes:
>>>>
>>>> > Now I create emulators for fictitious machines and then write
>>>> > assembler code for them.=20
>>>>
>>>> Tell us more about these "fictitious machines".
>>>
>>> I invented the fictitious 1960s-era company International Computing
>>> Machines (ICM). About once a year I get the yearning to design a new
>>> machine that they would have built. Sometimes it's simple and other
>>> times more complex. I write a Principle of Operations manual for
>>> it. Then I implement a simplistic visual representation and an
>>> emulator for the CPU and peripherals.
>>>
>>> Then I write software for the machine using my Assembler J and Linker
>>> J, which are a cross-assembler and cross-linker. Sometimes I don't get
>>> very far, other times I implement a piece of an operating system. Then
>>> I get sidetracked by real work and development stops.
>>
>> You are a crazy person. Excellent. Much better answer than MMIX.
>
> You are not the first person to think I'm crazy. However, you may be the first to think it's crazy in a good way!
Hardly. That's definitely the right kind of crazy.
>
> ~~ Paul
|
|
|
Re: All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #392540 is a reply to message #387124] |
Sun, 29 March 2020 12:39 |
Robin Vowels
Messages: 426 Registered: July 2012
Karma: 0
|
Senior Member |
|
|
On Wednesday, September 18, 2019 at 11:16:14 PM UTC+10, Peter Flass wrote:
> Bob Eager <n.....@eager.cx> wrote:
>> On Wed, 18 Sep 2019 12:30:10 +0000, Scott Lurndal wrote:
>>
>>> The problem, as I understand it, was because some 68k instructions could
>>> take an arbitrary number of cycles, which would delay interrupts.
>>> Workloads that worked on the 6502 wouldn't on the 68k due to excessive
>>> interrupt latency
>>
>> Interesting. The VAX seemed to solve that problem by having interruptable
>> instructions.
>>
>
> Same for IBM 360 line.
I think that instructions on the S/360 had to be
executed to completion.
Interruptible instructions did not appear until the S/370.
These included the MVCL and CLCL that could manipulate large
numbers of bytes.
|
|
|
Re: long running instructions, All programmers that developed in machine code and Assembly in the 1940s, 1950s and 1960s died? [message #392550 is a reply to message #392540] |
Sun, 29 March 2020 14:30 |
John Levine
Messages: 1405 Registered: December 2011
Karma: 0
|
Senior Member |
|
|
In article <e0c7662f-a3c3-4196-acd2-d46cbd50da48@googlegroups.com>,
<robin.vowels@gmail.com> wrote:
>>> Interesting. The VAX seemed to solve that problem by having interruptable
>>> instructions.
>>
>> Same for IBM 360 line.
>
> I think that instructions on the S/360 had to be
> executed to completion.
>
> Interruptible instructions did not appear until the S/370.
> These included the MVCL and CLCL that could manipulate large
> numbers of bytes.
That's right.
The PDP-6 and PDP-10, designed around the same time as the 360,
allowed arbitrarily deep indirect addressing, but took interrupts
during the address calculation. The address calculation didn't modify
anything so they'd just restart. There were two instructions ILDB and
IDPB which incremented a byte pointer and then loaded or stored the
byte it pointed to, both of which involved an address calculation.
There was a "byte interrupt" status bit set if the interrupt was
during the latter step so the CPU would know to skip the increment
when the interrupt returned and restarted the instruction.
There was also an interruptible BLT block transfer instruction which
was sort of like MVCL, on interrupt it stored the updated pointers in
the original register so it could restart normally.
--
Regards,
John Levine, johnl@taugh.com, Primary Perpetrator of "The Internet for Dummies",
Please consider the environment before reading this e-mail. https://jl.ly
|
|
|