On 2 Mar 2018 23:01:17 GMT, Bob Eager wrote:

On Fri, 02 Mar 2018 21:10:52 +0000, Tim Streater wrote:

In article , mechanic
wrote:

On Fri, 2 Mar 2018 04:00:02 -0800 (PST), whisky-dave wrote:

Shouldn't this be part of a students education in your establishment?

That depends on the course the studetns are doing the above as
computer science studetns how spend their time typing on keyboards
they want to do computer science.

Computers aren't assembled using soldering irons.

Depends. I bought a single board computer with a 68000 CPU and
peripheral chips. I had to solder on some jumpers to change the memory
range, to enable some interrupts, and to allow software to set and use
the timer in the PIO chip.

Indeed. And I soldered together a couple of PDP-8 clones.

Very heart-warming I'm sure, but not much to do with the assembly of
computers in factories.

On Sat, 03 Mar 2018 11:59:39 +0000, Tim Streater wrote:

If the student has to have a go at this, they might be minded to wonder
how circuit boards are actually made. A bit less magic, a bit more
reality.

Isn't that the usual complaint that schools teach people things that
were relevant a generation ago, not modern stuff? Same problem in
economics, physics and probably many other disciplines not just in
computer science.

On 03/03/2018 12:24, mechanic wrote:
On Sat, 03 Mar 2018 11:59:39 +0000, Tim Streater wrote:

If the student has to have a go at this, they might be minded to wonder
how circuit boards are actually made. A bit less magic, a bit more
reality.

Isn't that the usual complaint that schools teach people things that
were relevant a generation ago, not modern stuff? Same problem in
economics, physics and probably many other disciplines not just in
computer science.


They haven't changed the laws of physics for a while.
Computers haven't really changed much in the last few decades either.

Manufacturing has changed quite a bit.

On Sat, 03 Mar 2018 14:03:38 +0100, Martin wrote:

On 2 Mar 2018 23:01:17 GMT, Bob Eager wrote:

On Fri, 02 Mar 2018 21:10:52 +0000, Tim Streater wrote:

In article , mechanic
wrote:

On Fri, 2 Mar 2018 04:00:02 -0800 (PST), whisky-dave wrote:

Shouldn't this be part of a students education in your
establishment?

That depends on the course the studetns are doing the above as
computer science studetns how spend their time typing on keyboards
they want to do computer science.

Computers aren't assembled using soldering irons.

Depends. I bought a single board computer with a 68000 CPU and
peripheral chips. I had to solder on some jumpers to change the memory
range, to enable some interrupts, and to allow software to set and use
the timer in the PIO chip.

Indeed. And I soldered together a couple of PDP-8 clones.

Why not Mil Standard 1750 LSI 11 PDP 11 clones? :-)

I have a real LSI-11 already.

--
My posts are my copyright and if @diy_forums or Home Owners' Hub
wish to copy them they can pay me £1 a message.
Use the BIG mirror service in the UK: http://www.mirrorservice.org
*lightning surge protection* - a w_tom conductor

mechanic writes:

On Fri, 2 Mar 2018 04:00:02 -0800 (PST), whisky-dave wrote:

Computers aren't assembled using soldering irons.

Back in Summer 1978 I re-"assembled" the Ferranti F 100-M. Many hours
re-doing the backplane with wire-wrap. It was the medium-scale
integration (think 74xx series) prototype of the F 100-L.

--
Alan J. Wylie https://www.wylie.me.uk/

Dance like no-one's watching. / Encrypt like everyone is.
Security is inversely proportional to convenience

On Sat, 3 Mar 2018 13:03:09 +0000, dennis@home wrote:

They haven't changed the laws of physics for a while.

The whole understanding of the physics of elementary particles has
changed a few times in the last decades. Quantum mechanics and
cosmology are other active fields currently. Textbooks and teaching
ideas are way behind what students can see for themselves on Horizon
programmes, which must be a nightmare for teachers.

Computers haven't really changed much in the last few decades either.

Now that's just silly. Both for hardware and software.

On 03/03/2018 23:29, mechanic wrote:
On Sat, 3 Mar 2018 13:03:09 +0000, dennis@home wrote:

They haven't changed the laws of physics for a while.

The whole understanding of the physics of elementary particles has
changed a few times in the last decades. Quantum mechanics and
cosmology are other active fields currently. Textbooks and teaching
ideas are way behind what students can see for themselves on Horizon
programmes, which must be a nightmare for teachers.

Computers haven't really changed much in the last few decades either.

Now that's just silly. Both for hardware and software.


I used to design computers, first using TTL then bit slice then micros,
the're still the same.
Manufacturing has changed, computers haven't, they still work in the
same way.

Maybe quantum computers will mean a change in how computers work.

On Sun, 04 Mar 2018 13:58:32 +0000, Tim Streater wrote:

Computers haven't really changed much in the last few decades either.

Now that's just silly. Both for hardware and software.

No, all that's changed is that they've got faster and support more
memory. The basic architecture is still the same.

You mean apart from the changes in storage media (SSDs), video
cards, removable storage (USB), LCD monitors, UEFI booting and GPT
partitions,...
and for softwa
https://www.equinox.co.nz/blog/use-m...r-agile-or-not
- things have moved on from wading through manuals on assembly.

On 04/03/2018 13:58, Tim Streater wrote:

No, all that's changed is that they've got faster and support more
memory. The basic architecture is still the same.


And software has became bloatware so most of it works no faster than it
did 20 years ago on very slow machines.


--
mailto : news {at} admac {dot} myzen {dot} co {dot} uk

On Sun, 4 Mar 2018 12:21:36 +0000, dennis@home wrote:

I used to design computers, first using TTL then bit slice then micros,
the're still the same.

We were talking about training in up to date technology, as you just
illustrated this has changed vastly over the years. The processors
are so different you would have trouble in a current design shop.
The recent probs with Intel show how little people know about the
inner workings these days.

On 04/03/2018 18:47, mechanic wrote:
On Sun, 4 Mar 2018 12:21:36 +0000, dennis@home wrote:

I used to design computers, first using TTL then bit slice then micros,
the're still the same.

We were talking about training in up to date technology, as you just
illustrated this has changed vastly over the years. The processors
are so different you would have trouble in a current design shop.
The recent probs with Intel show how little people know about the
inner workings these days.


I wouldn't have trouble in a current design shop.
I pick things up far faster than almost anyone I have known.
Convincing them that someone my age can do the job would be more difficult.
Having three days treatment and 11 days compulsory rest every two weeks
doesn't leave much time for work anyway.

On 04/03/18 22:30, Tim Streater wrote:

Name me a CPU concept we have today and I'll point you at an old
machine with the same idea.


Advanced pipelines and branch prediction?

On 03/03/2018 13:03, dennis@home wrote:
They haven't changed the laws of physics for a while.

I think our understanding has changed.

Computers haven't really changed much in the last few decades either.

Well, when they were DILs with 0.1 inch pitch I was happy to mod them.
These days I can't even see the devices.

Manufacturing has changed quite a bit.

Apart from scale, less than you might think. SMT vs through-hole was a
jump of course, but since then only the scale has changed much.

Disclaimer: I didn't go in many manufacturing plants between ~1990-2010.

Andy

On 04/03/2018 22:30, Tim Streater wrote:


Those are peripheral devices, and memory is still accessed and
addressed as it was 50 years ago. Even the multiple core business is
nothing new, and neither is the notion of hyper-threading or whatever
it's called, with two sets of CPU registers sharing processor units
making it look like you've got 2 CPUs when there's actually only one.

System X has close coupled CPUs on a synchronous bus and they are
loosely coupled using high speed (well it was when I worked on them)
serial buses. Recent additions were to use microprocessors (running
Unix) on the serial links to boost processing power but I left before
any were deployed live so I don't know if they actually were.


You can look up the architecture of the CDC 6600 to see that it did all
of that back in 1965.

Machines back then were large because they didn't have integrated
circuits or single chip CPUs so it was all discrete logic. But that
didn't stop them having CPUs with main registers, index registers,
floating point, interrupts, protected memory, virtual memory, multiple
parallel access to memory, interleaved memory, it was all there.

Name me a CPU concept we have today and I'll point you at an old
machine with the same idea.

The idea of interrupting the CPU running the lowest priority task?
I held a patent on that for a while only to discover the patent
department had let it lapse when Intel used the same idea in its chip
sets. We could have made a few bucks from that.
That was quite a while ago so I suppose it depends on how old a machine
you want to talk about.

On 04/03/2018 22:37, Tim Watts wrote:
On 04/03/18 22:30, Tim Streater wrote:

Name me a CPU concept we have today and I'll point you at an old
machine with the same idea.


Advanced pipelines and branch prediction?

Early '80s.

On 04/03/18 23:23, dennis@home wrote:
On 04/03/2018 22:37, Tim Watts wrote:
On 04/03/18 22:30, Tim Streater wrote:

Name me a CPU concept we have today and I'll point you at an old
machine with the same idea.


Advanced pipelines and branch prediction?

Early '80s.


Which machine?

On 05/03/2018 08:09, Tim Watts wrote:
On 04/03/18 23:23, dennis@home wrote:
On 04/03/2018 22:37, Tim Watts wrote:
On 04/03/18 22:30, Tim Streater wrote:

Name me a CPU concept we have today and I'll point you at an old
machine with the same idea.


Advanced pipelines and branch prediction?

Early '80s.


Which machine?

SPARC and that was based on some experimental thing IIRC.

On 05/03/18 09:00, dennis@home wrote:
On 05/03/2018 08:09, Tim Watts wrote:
On 04/03/18 23:23, dennis@home wrote:
On 04/03/2018 22:37, Tim Watts wrote:
On 04/03/18 22:30, Tim Streater wrote:

Name me a CPU concept we have today and I'll point you at an old
machine with the same idea.


Advanced pipelines and branch prediction?

Early '80s.


Which machine?

SPARC and that was based on some experimental thing IIRC.


Ah - yes. I didn't peg SPARC as going back that far (I was wibbling with
VAXen back then)

On Friday, 2 March 2018 17:09:51 UTC, mechanic wrote:
On Fri, 2 Mar 2018 04:00:02 -0800 (PST), whisky-dave wrote:

Shouldn't this be part of a students education in your
establishment?

That depends on the course the studetns are doing the above as
computer science studetns how spend their time typing on
keyboards they want to do computer science.

Computers aren't assembled using soldering irons.

Very few products are.
But computerc science students do have to know about electricity and how things do get put togther in the real world that is the point.
They need to undersatand the problems bad connections make and such practical things.
I gave one student sandpaper to remove the enamal coating on wire he was using to build a speaker and he returned saying the wore just get disapearing he was not only removiong the enamal but continues until he had removed the copper too. This wire was 38 or 40 SWG so it didn't take much.
It;s a goofd idea that they get an idea of what soldering is too.

On Saturday, 3 March 2018 20:47:21 UTC, Alan J. Wylie wrote:
mechanic writes:

On Fri, 2 Mar 2018 04:00:02 -0800 (PST), whisky-dave wrote:

Computers aren't assembled using soldering irons.

Back in Summer 1978 I re-"assembled" the Ferranti F 100-M. Many hours
re-doing the backplane with wire-wrap. It was the medium-scale
integration (think 74xx series) prototype of the F 100-L.

This weeks practical was to construct a simple flip-flop for counting using sa 74S74N of course they all asked why haven;t I got any 74S74 so I said it;s because the academic got it wrong on the lab sheet and it should read 74HC74, which he asked me to purchase in an email sent at 7:45pm on the thursday night before the fridays lab which started at 2pm. How he thought I'd get 50+ of these chips from farnell in leeds in any weather let aling last weeks, but he did say don't worry he'd pick some up at Maplin on his way in Friday morning.

On Sunday, 4 March 2018 19:36:08 UTC, dennis@home wrote:
On 04/03/2018 18:47, mechanic wrote:
On Sun, 4 Mar 2018 12:21:36 +0000, dennis@home wrote:

I used to design computers, first using TTL then bit slice then micros,
the're still the same.

We were talking about training in up to date technology, as you just
illustrated this has changed vastly over the years. The processors
are so different you would have trouble in a current design shop.
The recent probs with Intel show how little people know about the
inner workings these days.


I wouldn't have trouble in a current design shop.
I pick things up far faster than almost anyone I have known.

That's called kleptomania ;-)

whisky-dave wrote:

he did say don't worry he'd pick some up at Maplin on his way in Friday morning.

They seem to have half the number you need spread between all their
london stores.

On Monday, 5 March 2018 16:33:34 UTC, Andy Burns wrote:
whisky-dave wrote:

he did say don't worry he'd pick some up at Maplin on his way in Friday morning.

They seem to have half the number you need spread between all their
london stores.

Well they arrived today from farnell 12p each we ordered 50,
Maplin might have had them but at 49p each.

whisky-dave writes:

On Saturday, 3 March 2018 20:47:21 UTC, Alan J. Wylie wrote:

Ferranti
(think 74xx series)

This weeks practical was to construct a simple flip-flop for counting
using sa 74S74N of course they all asked why haven;t I got any 74S74
so I said it;s because the academic got it wrong on the lab sheet and
it should read 74HC74,

Another job I did whilst at Ferranti was measuring the performance of
"staticisers" (on my first day I just nodded knowledgably until I found
out they were talking about flip flops).

If you toggled the D or JK inputs within a few nano-seconds of the clock
all sorts of strange things happened to the outputs. I had to test all
sorts of fabrication technologies at various temperatures. Low power
Schottky, TTL, CMOS, HC, etc.

When I came to plot the results, on log-linear graph paper, I got
some very impressive straight lines.

It was all to do with the first generation of automatic ticket reading
turnstiles on the London Underground.

--
Alan J. Wylie https://www.wylie.me.uk/

Dance like no-one's watching. / Encrypt like everyone is.
Security is inversely proportional to convenience

(Alan J. Wylie) writes:

P.S.

Getting back onto $Subject, I just made arrangements earlier today to
see, for the first time in a long time, a friend of 40 years who
originally came from Southend. His mother worked for Maplin, and her
picture, along with all the other employees was at the front of early
catalogues.

--
Alan J. Wylie https://www.wylie.me.uk/

Dance like no-one's watching. / Encrypt like everyone is.
Security is inversely proportional to convenience

On 01/03/2018 13:43, Martin Brown wrote:


The worst bit of the entire audio chain are the mechanical transducers
at the end of it (and the listening room acoustics).

Worse than the organic interface at the end?

On 04/03/2018 18:39, mechanic wrote:
things have moved on from wading through manuals on assembly.

Even when I was working full time in assembler that was never an issue.
In fact I think I look language things up more often now - C++ is a
_lot_ more complex than any assembler.

Andy

On Tue, 6 Mar 2018 22:56:12 +0000, Vir Campestris wrote:

On 04/03/2018 18:39, mechanic wrote:
things have moved on from wading through manuals on assembly.

Even when I was working full time in assembler that was never an issue.
In fact I think I look language things up more often now - C++ is a
_lot_ more complex than any assembler.


Didn't you have difficulty in creating OO structures in assembly?
That level of complexity is hidden in modern development languages,
which is part of their purpose of course. Things have moved on,and
C++is looking a bit ancient. This is progress, a concept denied by
the OP in this sub-thread. Stories about experiences in various
shops forty years ago refute this view.

On 07/03/18 12:09, mechanic wrote:
On Tue, 6 Mar 2018 22:56:12 +0000, Vir Campestris wrote:

On 04/03/2018 18:39, mechanic wrote:
things have moved on from wading through manuals on assembly.

Even when I was working full time in assembler that was never an issue.
In fact I think I look language things up more often now - C++ is a
_lot_ more complex than any assembler.


Didn't you have difficulty in creating OO structures in assembly?

Not at all.

The point about assembler and C is that you CAN create effective OO units.

But you are not forced to.


The easiest way is to use the lexical constructs (in C anyway) to
isolate OO functionality into one file. I.e use of static global
variables and static function names restricts their access to within
that file only. The file becomes the 'object'.

C is virtually macro assembler anyway, so there is little point going to
pure assembler unless you need hardware access or to do some weird stuff
with mode switches etc that are outside the scope of the C language



That level of complexity is hidden in modern development languages,
which is part of their purpose of course. Things have moved on,and
C++is looking a bit ancient. This is progress, a concept denied by
the OP in this sub-thread. Stories about experiences in various
shops forty years ago refute this view.


Actually since OOP code has got worse, slower and more buggy.

Becauses its written by idiots who think they dont need to understand
hardware, memory or instruction cycoles.

Bcak in et day peole like you wer touting the merits of Pascal. Today no
one uses it, or Modula II.

But C and assembler march on.


--
Climate Change: Socialism wearing a lab coat.

On Wed, 07 Mar 2018 12:09:31 +0000, mechanic wrote:

On Tue, 6 Mar 2018 22:56:12 +0000, Vir Campestris wrote:

On 04/03/2018 18:39, mechanic wrote:
things have moved on from wading through manuals on assembly.

Even when I was working full time in assembler that was never an issue.
In fact I think I look language things up more often now - C++ is a
_lot_ more complex than any assembler.


Didn't you have difficulty in creating OO structures in assembly? That
level of complexity is hidden in modern development languages,
which is part of their purpose of course. Things have moved on,and C++is
looking a bit ancient. This is progress, a concept denied by the OP in
this sub-thread. Stories about experiences in various shops forty years
ago refute this view.

That rather assumes that one believes OO is always useful, or progress!

--
My posts are my copyright and if @diy_forums or Home Owners' Hub
wish to copy them they can pay me £1 a message.
Use the BIG mirror service in the UK: http://www.mirrorservice.org
*lightning surge protection* - a w_tom conductor

On 07/03/18 13:26, Bob Eager wrote:
On Wed, 07 Mar 2018 12:09:31 +0000, mechanic wrote:

On Tue, 6 Mar 2018 22:56:12 +0000, Vir Campestris wrote:

On 04/03/2018 18:39, mechanic wrote:
things have moved on from wading through manuals on assembly.

Even when I was working full time in assembler that was never an issue.
In fact I think I look language things up more often now - C++ is a
_lot_ more complex than any assembler.


Didn't you have difficulty in creating OO structures in assembly? That
level of complexity is hidden in modern development languages,
which is part of their purpose of course. Things have moved on,and C++is
looking a bit ancient. This is progress, a concept denied by the OP in
this sub-thread. Stories about experiences in various shops forty years
ago refute this view.

That rather assumes that one believes OO is always useful, or progress!



Indeed. OO is actually a disaster. It complicates needlessely.

And it forces a stage of design forethought that is not necessary in so
many cases.

It really only has it place in huge companies employing crap coders to
produce bloatware.

The OP is te sort of person who probably thinks that memeory allocation
should be 'handled by the language' and then cant fix a situation where
his code stops for 5 seceonds whole memory garbage collection goes on.



--
"Nature does not give up the winter because people dislike the cold."

ۥ Confucius

On 07/03/18 12:09, mechanic wrote:
On Tue, 6 Mar 2018 22:56:12 +0000, Vir Campestris wrote:

On 04/03/2018 18:39, mechanic wrote:
things have moved on from wading through manuals on assembly.

Even when I was working full time in assembler that was never an issue.
In fact I think I look language things up more often now - C++ is a
_lot_ more complex than any assembler.


Didn't you have difficulty in creating OO structures in assembly?
That level of complexity is hidden in modern development languages,
which is part of their purpose of course. Things have moved on,and
C++is looking a bit ancient. This is progress, a concept denied by
the OP in this sub-thread. Stories about experiences in various
shops forty years ago refute this view.


You can create any structure in assembler, whether you would want to is
another matter. One of the fundamentals of programming is that once you
have built something you can use it as a building block for the next
stage of sophistication, complexity or obfuscation if you will.

I have always found it a great advantage to have started in the
computing business in the age of blinken lights. You can see through a
lot of nonsense if you understand the foundations.


--
djc

(–€Ì¿Ä¹Ì¯–€Ì¿ Ì¿)
No low-hanging fruit, just a lot of small berries up a tall tree.

In article , Huge
scribeth thus
On 2018-03-07, DJC wrote:

[23 lines snipped]

I have always found it a great advantage to have started in the
computing business in the age of blinken lights. You can see through a
lot of nonsense if you understand the foundations.

*applause*


Isn't that true of most anything?..
--
Tony Sayer

On 07/03/2018 12:09, mechanic wrote:
Didn't you have difficulty in creating OO structures in assembly?

None whatsoever. Because we didn't try.

Systems programming isn't like ordinary DP.

Andy

In article ,
Huge writes:
On 2018-03-09, Vir Campestris wrote:
On 07/03/2018 12:09, mechanic wrote:
Didn't you have difficulty in creating OO structures in assembly?

None whatsoever. Because we didn't try.

Systems programming isn't like ordinary DP.

I still remember adding up instruction cycle times to make sure we weren't
running too long with interrupts disabled (*). OO? WTF's that? )

(Rhetorical question. I know perfectly well what OO is.)

(* In device drivers under RSX11/M on a PDP11. Yes, it was a long time ago.)

Back in the days of X.25 packet switches, I wrote some carefully
crafted assembly code which could switch a packet in 127 instructions.
It allowed for the computer's pipeline, performing base register loads
well enough in advance that there was no stall in the indirect load,
generally interleaving two separate functions in alternate instructions.
It was part of a project which increased the packet switching
performance by over 10 times.

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]

On 10/03/2018 09:59, Huge wrote:
On 2018-03-10, Andrew Gabriel wrote:
In article ,
Huge writes:
On 2018-03-09, Vir Campestris wrote:
On 07/03/2018 12:09, mechanic wrote:
Didn't you have difficulty in creating OO structures in assembly?

None whatsoever. Because we didn't try.

Systems programming isn't like ordinary DP.

I still remember adding up instruction cycle times to make sure we weren't
running too long with interrupts disabled (*). OO? WTF's that? )

(Rhetorical question. I know perfectly well what OO is.)

(* In device drivers under RSX11/M on a PDP11. Yes, it was a long time ago.)

Back in the days of X.25 packet switches, I wrote some carefully
crafted assembly code which could switch a packet in 127 instructions.
It allowed for the computer's pipeline, performing base register loads
well enough in advance that there was no stall in the indirect load,
generally interleaving two separate functions in alternate instructions.

"You are not expected to understand this."

Why not?

Next you'll be saying you don't understand Meltdown and Spectre :P

Andy

On Sun, 11 Mar 2018 22:31:59 +0000, Huge wrote:

On 2018-03-11, Vir Campestris wrote:
On 10/03/2018 09:59, Huge wrote:
On 2018-03-10, Andrew Gabriel wrote:
In article ,
Huge writes:
On 2018-03-09, Vir Campestris
wrote:
On 07/03/2018 12:09, mechanic wrote:
Didn't you have difficulty in creating OO structures in assembly?

None whatsoever. Because we didn't try.

Systems programming isn't like ordinary DP.

I still remember adding up instruction cycle times to make sure we
weren't running too long with interrupts disabled (*). OO? WTF's
that? )

(Rhetorical question. I know perfectly well what OO is.)

(* In device drivers under RSX11/M on a PDP11. Yes, it was a long
time ago.)

Back in the days of X.25 packet switches, I wrote some carefully
crafted assembly code which could switch a packet in 127
instructions.
It allowed for the computer's pipeline, performing base register
loads well enough in advance that there was no stall in the indirect
load, generally interleaving two separate functions in alternate
instructions.

"You are not expected to understand this."

Why not?

[Whoosh]

https://thenewstack.io/not-expected-...and-explainer/

I even had a T-shirt made with it on, years ago. Gets worn for a certain
lecture.

--
My posts are my copyright and if @diy_forums or Home Owners' Hub
wish to copy them they can pay me £1 a message.
Use the BIG mirror service in the UK: http://www.mirrorservice.org
*lightning surge protection* - a w_tom conductor

Ah, I'd add Boots to that list of "I remember them". They've fallen into
the BHS trap of not really knowing what they are. Are they a pharmacist,
or a discount make-up store flogging the odd sandwich on the side ?

Those three bits seem to have survived the attempts to diversify

It's all the rest that has disappeared

How many remember they once had a library service?

My mother used to use that - presumably partly out of habit, since there was a big council library right opposite...

On 11/03/2018 21:31, Vir Campestris wrote:
On 10/03/2018 09:59, Huge wrote:
On 2018-03-10, Andrew Gabriel wrote:
In article ,
Â*Â*Â*Â*Huge writes:
On 2018-03-09, Vir Campestris wrote:
On 07/03/2018 12:09, mechanic wrote:
Didn't you have difficulty in creating OO structures in assembly?

None whatsoever. Because we didn't try.

Systems programming isn't like ordinary DP.

I still remember adding up instruction cycle times to make sure we
weren't
running too long with interrupts disabled (*). OO? WTF's that?Â*Â* )

(Rhetorical question. I know perfectly well what OO is.)

(* In device drivers under RSX11/M on a PDP11. Yes, it was a long
time ago.)

Back in the days of X.25 packet switches, I wrote some carefully
crafted assembly code which could switch a packet in 127 instructions.
It allowed for the computer's pipeline, performing base register loads
well enough in advance that there was no stall in the indirect load,
generally interleaving two separate functions in alternate instructions.

"You are not expected to understand this."

Why not?

I think I should having designed the hardware and firmware for the X25
card used for billing data on SystemX.


Next you'll be saying you don't understand Meltdown and Spectre :P

Easy to understand.
Much more difficult to workout what the data is.


Andy

In article ,
Tim Streater writes:
In article , Huge
wrote:

On 2018-03-11, Vir Campestris wrote:
On 10/03/2018 09:59, Huge wrote:
On 2018-03-10, Andrew Gabriel wrote:
Back in the days of X.25 packet switches, I wrote some carefully
crafted assembly code which could switch a packet in 127 instructions.
It allowed for the computer's pipeline, performing base register loads
well enough in advance that there was no stall in the indirect load,
generally interleaving two separate functions in alternate instructions.

"You are not expected to understand this."

Why not?

[Whoosh]

https://thenewstack.io/not-expected-...and-explainer/

Next you'll be saying you don't understand Meltdown and Spectre :P

Except I do.

The comment is ********, of course. And I assume the 127 instructions
for switching a packet will still work even if the internal workings of
the switch changes. E.g. if you are told to buy a more recent, cheaper
model of the switch that has no pipeline and a different memory
architecture.

It was optimised for what was then (1980's) our top range mini-
computer which had a 4-instruction pipeline. It also run on the
slower/cheaper 2- and 0-instruction pipeline systems, but they
were slower for many reasons besides just instruction pipeline
(things like lack of multi-ported memory, and less memory, and
narrower memory bus).
These all ran the same 127 instructions, but only the top range
system gained from specifically hand-crafting those instructions.

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]