On Sunday, January 15, 2023 at 5:30:18 PM UTC+1, hughag...@gmail.com wrote:
> The MiniForth processor, for which I wrote the MFX assemble, simulator and Forth cross-compiler,
> was Harvard Architecture. The code-memory and data-memory each had their own address bus
> and data bus. This is why each opcode executed in one clock cycle. Also, the MiniForth was VLIW.
> Each opcode contained up to 5 instructions, all of which executed concurrently in one clock cycle.
> My assembler allowed the programmer to write his instructions as if they executed sequentially
> but my assembler would rearrange them to pack them into the opcodes with a minimal number
> of NOP instructions being inserted, but still guaranteeing that the program did the same thing
> that it would have done if the instructions were assembled one per opcode in the same order
> that they were written in the source-code. This is called: "out-of-ordering."

The accepted definition of processor architectures is: "Von Neumann Architecture is a digital computer
architecture whose design is based on the concept of stored program computers where program
data and instruction data are stored in the same memory. This architecture was designed by the
famous mathematician and physicist John Von Neumann in 1945. 

Harvard Architecture is the digital computer architecture whose design is based on the concept
where there are separate storage and separate buses (signal path) for instruction and data. It was
basically developed to overcome the bottleneck of Von Neumann Architecture".. 

Now, 4tH isn't a processor - it's a VM - so if you want to explain how I can make "busses" in high
level C, I'd like to hear from you. But the 4tH VM *does* have separate (virtual) memory areas for
different kinds of data. If you don't agree with the definition - fine. You're a one person minority
and then all discussion stops.

It's as simple as that.

Hans Bezemer

In article <47d55a21-3102-4341-9d5f-7687397e4bf3n@googlegroups.com>,
Hans Bezemer <the.beez.speaks@gmail.com> wrote:
>On Sunday, January 15, 2023 at 5:30:18 PM UTC+1, hughag...@gmail.com wrote:
>> The MiniForth processor, for which I wrote the MFX assemble, simulator
>and Forth cross-compiler,
>> was Harvard Architecture. The code-memory and data-memory each had
>their own address bus
>> and data bus. This is why each opcode executed in one clock cycle.
>Also, the MiniForth was VLIW.
>> Each opcode contained up to 5 instructions, all of which executed
>concurrently in one clock cycle.
>> My assembler allowed the programmer to write his instructions as if
>they executed sequentially
>> but my assembler would rearrange them to pack them into the opcodes
>with a minimal number
>> of NOP instructions being inserted, but still guaranteeing that the
>program did the same thing
>> that it would have done if the instructions were assembled one per
>opcode in the same order
>> that they were written in the source-code. This is called: "out-of-ordering."
>
>The accepted definition of processor architectures is: "Von Neumann
>Architecture is a digital computer
>architecture whose design is based on the concept of stored program
>computers where program
>data and instruction data are stored in the same memory. This
>architecture was designed by the
>famous mathematician and physicist John Von Neumann in 1945. 

That was the basic brilliant idea.

>
>Harvard Architecture is the digital computer architecture whose design
>is based on the concept
>where there are separate storage and separate buses (signal path) for
>instruction and data. It was
>basically developed to overcome the bottleneck of Von Neumann Architecture". 

That is throttle and a cludge for technical reasons. It doesn't replace
the von Neumann structure in principle. You have a signal path to
instructions? How can a desktop computer have instructions to that
memory? Underlying von Neumann structure, baby!
You can have a washing machine with all code in rom, and even that
is promoted with the idea of easy change.
Von Neumann's idea is the reason why computers are powerful as it is.

Actually his idea was not as original as it may seem. The universal
Turing machine came first as a theoretical construct.

>
>Now, 4tH isn't a processor - it's a VM - so if you want to explain how I
>can make "busses" in high
>level C, I'd like to hear from you. But the 4tH VM *does* have separate
>(virtual) memory areas for
>different kinds of data. If you don't agree with the definition - fine.
>You're a one person minority
>and then all discussion stops.

There are no end of variations, the 386 sports connecting to 4 segments
each of a choice of 16 of 32 bit sizes separate for code and data.
This shouldn't detract from the principal idea that computers can
change their own programs.

>
>It's as simple as that.
>
>Hans Bezemer
--
Don't praise the day before the evening. One swallow doesn't make spring.
You must not say "hey" before you have crossed the bridge. Don't sell the
hide of the bear until you shot it. Better one bird in the hand than ten in
the air. First gain is a cat spinning. - the Wise from Antrim -

On Friday, January 13, 2023 at 10:33:41 PM UTC+1, hughag...@gmail.com wrote:
> Hans Bezemer claims to be a big expert on Forth and he tries to educate the world
> with his Back & Forth videos.
I claim nothing of the kind. Where in these videos do I explicitly claim to be an expert?
I quote:

"Hello, I'm Hans Bezemer" - I have proof for that.
"I've been a programmer for 40 years" - I can prove that too. I got a resume.
"and I'm using Forth" - which may be quite evident - taken where I'm posting.
"Forth is a deceivingly simple language that is notoriously hard to master" - true. You may
think otherwise. I'd like to see your arguments to the contrary. But it's my honest impression.
"But it enables me to make more complex programs - and faster than I ever could in C" - Again,
my impression. I do have some data on that where applications are involved. Making reusable
libs may be a little harder than C, admitted.
"In this series I'm breaking down some of the most intriguing projects I took upon me" - *I* find
them intriguing. My audience is free to think otherwise. And I'd be happy if they let me know
so I can try and find something even more intriguing.

> The problem is that nobody on comp.lang.forth tells
> him that he is wrong about what Harvard Architecture is.
Maybe they agree with me. I'm always happy to discuss definitions.

> That would be politically incorrect!
That's an assumption. You have to provide proof for that.

> This is cult behavior. So long as he continues to be part of the cult (mostly by attacking me),
> then everybody in the cult carefully avoids pointing out his gross ignorance of basic
> concepts such as Harvard Architecture.
Your problem is that you're completely unable to distinguish between not agreeing with you
and a personal attack. If Albert (and others) claim that this or that word comprises a lousy
definition - they may be right, and I even may agree with them. I don't take it personal.

I am not my work - and my work is not of consistent quality in every detail or aspect
(I wonder whose is). But a significant difference between you and me is I don't feel or have
the need to be in a competition. I mainly do it because I love doing it. There are also things
I mainly do to provide for my family.

If you want to do a quarter mile to prove who's the fastest you'll find yourself lonely at the
finish line, because I frankly don't have the desperate need to know to keep up my fragile
confidence.

I'm here because I want to learn and contribute - since I love and value collaboration. I
don't have to be the top monkey. I don't see the value in that.

Hans Bezemer

On Monday, January 16, 2023 at 11:21:04 AM UTC+1, none albert wrote:
> This shouldn't detract from the principal idea that computers can
> change their own programs.
Within the 4tH VM - you can't.

HB

On Tuesday, September 13, 2022 at 9:13:26 AM UTC+2, S wrote:
> Now, Hans, please go away and chuck your domineering fit somewhere else. You have had and
> continue to have, negliable benefit for people here. It's your fault you are in the wrong place
> and can't be bothered to correctly interpret things, and regard even absolute facts as
> relativistic. Which is not credible or scientific.
You go away! ;-) - and remember: science is based on the strength of your arguments - which
some people are completely lacking. Hence, they never climb very far on the ladder of
"Grahams triangle of disagreement" - and lose themselves in tactics which most grown men
haven't used since they left the schoolyard.

Hans Bezemer

On Monday, January 16, 2023 at 6:21:04 AM UTC-4, none albert wrote:
> In article <47d55a21-3102-4341...@googlegroups.com>,
> Hans Bezemer <the.bee...@gmail.com> wrote:
> >On Sunday, January 15, 2023 at 5:30:18 PM UTC+1, hughag...@gmail.com wrote:
> >> The MiniForth processor, for which I wrote the MFX assemble, simulator
> >and Forth cross-compiler,
> >> was Harvard Architecture. The code-memory and data-memory each had
> >their own address bus
> >> and data bus. This is why each opcode executed in one clock cycle.
> >Also, the MiniForth was VLIW.
> >> Each opcode contained up to 5 instructions, all of which executed
> >concurrently in one clock cycle.
> >> My assembler allowed the programmer to write his instructions as if
> >they executed sequentially
> >> but my assembler would rearrange them to pack them into the opcodes
> >with a minimal number
> >> of NOP instructions being inserted, but still guaranteeing that the
> >program did the same thing
> >> that it would have done if the instructions were assembled one per
> >opcode in the same order
> >> that they were written in the source-code. This is called: "out-of-ordering."
> >
> >The accepted definition of processor architectures is: "Von Neumann
> >Architecture is a digital computer

First fallacy. Von Neumann architecture is one way to design computers. It is very popular today. It is not the "accepted definition of processor architectures".

> >architecture whose design is based on the concept of stored program
> >computers where program
> >data and instruction data are stored in the same memory. This
> >architecture was designed by the
> >famous mathematician and physicist John Von Neumann in 1945.
> That was the basic brilliant idea.
> >
> >Harvard Architecture is the digital computer architecture whose design
> >is based on the concept
> >where there are separate storage and separate buses (signal path) for
> >instruction and data. It was
> >basically developed to overcome the bottleneck of Von Neumann Architecture".
> That is throttle and a cludge for technical reasons. It doesn't replace
> the von Neumann structure in principle.

I don't know what is meant by "replace". Von Neumann architecture is a common memory interface for data and instructions and the Harvard architecture has a separate memory interface for data and instructions. The two can't exist at the same time, although there are hybrid designs, where data and instruction memory are separate, but data can be moved between the two with specific instructions.

> You have a signal path to
> instructions? How can a desktop computer have instructions to that
> memory?

Desktop computers are not the only computers in the world. The AVR processors are essentially Harvard architecture (modified) internally. Data and program storage are separate and accessible simultaneously. Many DSP processors are Harvard, with two or even three address spaces, accessible simultaneously.

> Underlying von Neumann structure, baby!
> You can have a washing machine with all code in rom, and even that
> is promoted with the idea of easy change.
> Von Neumann's idea is the reason why computers are powerful as it is.
>
> Actually his idea was not as original as it may seem. The universal
> Turing machine came first as a theoretical construct.
> >
> >Now, 4tH isn't a processor - it's a VM - so if you want to explain how I
> >can make "busses" in high
> >level C, I'd like to hear from you. But the 4tH VM *does* have separate
> >(virtual) memory areas for
> >different kinds of data. If you don't agree with the definition - fine.
> >You're a one person minority
> >and then all discussion stops.
> There are no end of variations, the 386 sports connecting to 4 segments
> each of a choice of 16 of 32 bit sizes separate for code and data.
> This shouldn't detract from the principal idea that computers can
> change their own programs.

Segments have nothing to do with the memory architecture of a processor, in the sense of von Neumann vs Harvard. They are simply a manner of generating address ranges that are perfectly capable of overlapping.

--

Rick C.

--- Get 1,000 miles of free Supercharging
--- Tesla referral code - https://ts.la/richard11209