@NutsAboutAmiga

What I had in mind was to have each read/write mapped out using linear 1:1 translations where the virtual memory mapping provided by AmigaOS4 is exactly the same inside the Emulated CPU as outside within the AmigaOS native routines containing it and then deal with "exceptions" with any specific segment mappings as limited as possible,

I'm wanting to work on having a CPU Emulation as the only "hardware" Emulation component and running false drivers within the Emulation for display or other routines and causing an Internal Exception that is not an AmigaOS Exception but an Emulated CPU Exception to use Amiga OS native routines where possible to replace functionality required by the Application using the Emulated CPU.

The consideration here would be for a Windows or Mac OS program to be loaded and then use routines from the Emulation to replace parts of the OS within the target being Emulated/Simulated.

Windows XP 32bit for example, would have kernel material replaced by DLL files in a similar fashion to Wine and material running inside a CPU Emulation on top of the PPC but specific functionality such as the GDI KRNL and USER Exported syscalls to be replaced with shim calls where functionally possible...
these would then fake the presence of Windows without needing Windows the same as Wine does.

I consider this a "blended" setup in partially faking Hardware and partially providing the OS environment the Application came from, However I will find that the libraries I need to provide as shims will only plausibly function where the equivalent functionality can be provided by or connected properly with the Host OS syscalls...

One example would be faking a "commdlg.dll" and providing the User Interface Elements through use of Asl.Library... this is going to eat a lot of time and I will be making use of development in the Wine project to accelerate what I can here.

@opi

Quote:

the attackers

Well, I wasn't specifically referring to you. You've been somewhat negative in this thread. Others have been more so. Still, you engaged me on the point, so I'll respond in kind.

Quote:

When I said X1000 will be late and expensive, when I said Troika is a scam, when I said AI will not sponsor Kent's Sport Center nor they have deal with Sharp/whoever, when I questioned ACK or CUSA

In Amigaland, it's easy to sit back and say, "that is going to fail.", "that is going to fail too", "oh, and so is that", "that's going to fail miserably".. "oh, and only idiots will think that won't fail".

Quote:

Now, let me dig up this thread in a year or two, we'll have a good laugh.

The Amiga landscape is litered with carcasses. So, nobody needs the "clairvoyance" of a few, who are willing to state the obvious. We all "get" it... Most projects will either not succeed, or will take longer and be more expensive than expected. So, for anyone to feel 'vindicated' because they were right in being skeptical, is wanting to be congratulated for stating the obvious.

Quote:

Attacking is not asking question. Valid, technical questions. Why do people think that asking question is somehow rude?

Many of those questions have been of the form, "your history proves you are a con-artist, and can't deliver a thing. So why should I believe you now?".. So, yeah.. that's a question alright. It's also an attack.

Thing is.. I think some of us would love to be able to devote our lives to Amiga, and give 100% so our goals are always met. But, most of us have day jobs that interfere with our good intentions. Few of the developers involved in Amiga, make a living through that involvement. And it is much harder to produce stable, predictable schedules when Amiga is a sideline for you at best, that doesn't even pay one utility bill. And what is the point, when the community just throws stones, for not delivering on time and under budget, something you were doing mostly for the satisfaction of contributing. Somehow, the "community" expects a predictable, corporate, mass-market delivery and cost-effectiveness from this renegade band of "contributors", who are mostly donating spare time. Under these circumstances, ANYONE can foresee that failure (or at least tardiness) is a likely outcome. So, who needs it pointed out constantly? I just think the expectations are wrong.

Quote:

For me to buy his story there needs to be hard evidences. Production ready code, white papers he wrote, commits to well regarded FOSS project. Forum posts is not an evidence of anything.

It's so easy to be skeptical and dismissive. It takes a lot more courage to suspend disbelief and try to encourage people to have another go, dispite all the carcasses that litter the sidewalk.

On the other hand... Maybe we should ALL take the easy road; that is be highly critical, skeptical, dismissive and disbelieving of everything that is ever stated or proposed by anyone in conjunction with Amiga. Maybe a community of dismissive disbelievers is exactly the kind of welcoming environment any prospective Amiga user or developer would be thrilled to be a part of. Because then we can all feel so smart for being "proven" right, when to everyone's _surprise_ , yet another project is over time or over budget, or off the rails completely.

Against this backdrop.. I feel like any expectations other than low expectations are unwarranted. In other words, we are in no place to be picky or overly demanding.. It's just not the reality we are confronted with. And to have high expectations, and demand timely and cost-effective results (or else you'll be skeptical) just comes across as naive. By way of contrast, you don't see a dehydrated man in the desert complaining that the watering hole has dirt in it, do you? In other words, if you expect low-cost, timely, predictable results... well, then you're in the wrong place...

Quote:

opi wrote: The Emperor's New Clothes has a nice moral story to it.

The trouble with those who feel criticized for pointing out that "the king has no clothes"... is somehow they always assume everyone else is blind. They discount the other possibility... that other's see, but choose to deal with the _obvious_ in a way that doesn't destroy, defame, or deride.

_________________
Vic-20, C-64, C-128
Amiga 1000, 3000
AmigaOne X1000

@jingof

Quote:

Thing is.. I think some of us would love to be able to devote our lives to Amiga

:O ?!!!!?

@wawa

Quote:

Thing is.. I think some of us would love to be able to devote our lives to Amiga

Yeah.. see your point, that was a bit stronger than I intended. Just mean, many of us would love to devote our work week entirely to Amiga, and work on Amiga projects full time. Few have that luxury, but I for one would dig it! Unfortunately, I have a 6000 sqft house, a very expensive wife and 3 kids to pay for. So, Amiga for me is an infrequent, luxury of a diversion, where adherence to a "schedule" is a dream about a fantasy.

_________________
Vic-20, C-64, C-128
Amiga 1000, 3000
AmigaOne X1000

@jingof

You have bought yourself a wife? Arent they available for free?

Last edited by wawa on 02-May-2012 at 12:53 AM.

@belxjander

Here is a VM I wrote as an learning experiment quite some years ago. It's a virtual machine in the truest sense, that is, it emulates the behaviour of a load/store architecture CPU that exists nowhere other than on paper.

It presently compiles for OS3.x, WarpOS, OS4, 32 and 64-bit linux and produces an executable with an embedded "hello world" application. One that generates a greyscale 512x512 mandelbrot and dumps it to disk. The original intent was to make this into a library as an embeddable low-level scripting engine for some game/multimedia projects I was working on long ago.

I only just ported it to OS4 in the last 10 minutes. Anyway, some instructions are not implemented on all targets and I just discovered that loading addresses of string literals has a bug on PPC/68K that I don't recall happening previously. It's probably an endian issue.

There are 2 supported interpreter models (gigantic switch case versus function table), though I seem to have lost my hand-optimised 68K assembler version.

There's even some documentation: http://extropia.co.uk/projects/exvm-doc/

Sources here: http://extropia.co.uk/projects/exvm_multiplatform.tgz

As it is a very simple machine model, perhaps you could target it first to, erm, validate your idea without all the complexities of actual hardware emulation...

;)

_________________
Doing stupid things for fun...

@Karlos

I'm considering a blended CPU only Emulation with a fake OS API/ABI presentation to start and work up from there...

I'm not going to go for the complete fake hardware setup without a good long think about it along with discussion with any developers who may be interested.

The plans I have at the moment are based around completion of Perception-IME first and Polymorph-VM material over a longer time frame.

I didn't get to complete it the first time around along with loss of the sources so I have to deal with trying again... ho hum...

@others...

If anyone wants to point out where there is a better strategy or design-pattern to use... feel free...

@opi

Don't be so negative, some people here like to act funny. Just watch and enjoy ...

_________________
AmigaONE. Haha. Just because you can put label on it does not make it Amiga.

I borrowed this comments from here (#27 & #28):
http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=38873&forum=2&start=20&order=0

It's very easy to be negative. Much easier than being positive.
It's too easy to rember all bad things that have happened, it often makes one forget all the good stuff.
And in the end, we must not cling to past. We can learn from the past happenings, but not to hold them as evidence for not trying anything in the future.

@opi

U know, one does need to be "right" on all matters, no one is.
Life is easier if people accept failures, learn and try harder next time.
Most good things are worth to be done poorly/failing, just to eventually make it right one day.

"nor they have deal with Sharp/whoever,"

I think they had a deal with Sharp, like they had the deal with Nokia.
But AmigaInc sussed things up + Sharp and Nokia devices flopped, etc.

I'm not saying AmigaInc was not run by empty promising half-criminals (+idiots), just clearing some things.

And in the end, those past events have nothing to do with some wannabe AmigaNG developer.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@Belxjander

Hmm, how to say this without being labelled another unduly negative naysayer. I've not read this thread in it's entirety but I gather you are planning on some sort of "generic" processor emulation to be used in conjunction with a Wine-like abstraction layer to allow executables for other platforms to be run on OS4.

In doing so, you've brushed aside a great deal of technical detail behind some flowery language and vague statements about the "muchness" of how processors retrieve and execute instructions, nebulous remarks about using some sort of trap/exception mechanism and simple 1:1 memory mappings and so on.

It all sounds very ambitious and grandiose, but speaking as someone that's actually written an emulator for a real processor (and not just a VM), I can tell you the devil is in the detail unless performance and compatibility is extremely unimportant. Emulating real processors with the intent of obtaining useful levels of performance is not a trivial task, particularly if you want to emulate things like memory management hardware.

Like several others here, I'm not really convinced of the feasibility of what you are suggesting. It's one thing to have crazy ideas you might not be able to complete (I've had plenty) but in making a bounty out of this project and accepting other people's money you must feel very self-assured of your ability to complete it. You'd have to be, because the alternative is that you've accepted other people's money to help fund a personal purchase and that's simply not cricket.

As it happens, I don't like to think the worst in people, which is why I suggested trying your idea on a system that has no hardware to emulate, no OS layers to abstract and has a trivially simple to understand and (re)implement execution model and can be pretty easily verified and debugged. And one which I happened to have just lying around doing nothing.

_________________
Doing stupid things for fun...

@Karlos

I accept your post as the questioning criticism it is... and yes I feel confident about parts of the project,
I also agree some of the material will be challenging, and I have tried (apparently unsuccessfully?) to explain how I am doing this.

Any errors or omissions in the descriptions I have put up are definitely my own,

I also invite you to keep an eye on the project and offer further critiques such as this one...as you see fit and have time to do so.

Polymorph is not the main project I will focus on exclusively, and I personally do see it as needing a very long-term development plan.

I do hope to see the code I produce speak for itself more as I express more of the way I want to do this,
as for coming across "flowery"... I'm not sure about that... maybe in how I am trying to be specific about things comes across in the abstract? is that the case?

So please feel free to ask and question me about this further...

To anyone else... I am *not* simply going to go away... even dishy_fis one of the more critical members who sees me as being full of hot air... has admitted I have been around for *years*.

I'm going to pursue rebuilding this project... and I am willing to trade some speed for accuracy.

I'm at best roughly estimating and at worst wildly optimistic about performance on the PPC based AOS4 systems... I do accept that as a valid criticism (hell I agree with the statement!), but that is not going to stop me trying anyway.

Karlos: I thank you for being honest in expressing your opinion and do hope to hear more critiques of the project from you in the future, good or bad.

The only disagreement of your post that I have is the statement
Quote:

In doing so, you've brushed aside a great deal of technical detail behind some flowery language and vague statements about the "muchness" of how processors retrieve and execute instructions, nebulous remarks about using some sort of trap/exception mechanism and simple 1:1 memory mappings and so on.

I apologize for being unclear in earlier posts but these are currently what I am considering... not actively using at this time. some of the material has been reviewed and I am actively changing the layout of the design due to a concept change from the original *lost* implimentation.

so I will be proceeding with the concept and building section by section, I'm not actively dismissing anything presented here so much as stating that my best explanations still have large gaps in what I am trying to explain.

I have run into this many many times before in that what I focus on (details and how they connect) seems to be perceived differently when I discuss these with others.

I can only hope that what I re-code in the way the library functions makes the entire concept clearer despite the possibly confusing nature of my own explanations.

I'll only ask people to have an open mind and from that question the project material itself without ad-hominem attacks.
I'm more than willing to try and explain detailed sections and how they inter-relate where this remains civilized.
I'm also *wanting* people like yourself and Opi to retain their skepticism about the project so that it receives some fair and differently biased review and consideration and critiques.

At any point where I seem dismissive of others in these threads is based on my own feeling of failing to explain how I perceive and understand the elements of this project.

Anyway... as you have stated, I *have* accepted payments for these projects (I have even documented as such in the project information directly) and will do my best to see both projects through.

If you see a part of the code and can suggest a better way to make what it does happen, please feel free to suggest that to me. I would hope however that you are open to my asking questions in the case where part of the changes suggested may be unclear to me.

Again... thank you for the critique

Last edited by Belxjander on 03-May-2012 at 01:41 AM.

@Belxjander

IIRC, you were also going to build some chinese/japanese language support for AOS, was it so?

IMHO, It would be dream come true if we had more than 0.01% chance to enter chinese market with AOS4. So to me that language "project" seems very important and worth to push forward as first priority.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK
The language project is "Perception-IME" and currently ahead of where "Polymorph VM" is.

I'm hoping to start and complete at least Japanese and make it open enough to be used for Chinese and Korean as well.

I've got some minimal Chinese and Korean material started but I have no idea how to deal with specifics for those two languages.

@KimmoK

Quote:

IMHO, It would be dream come true if we had more than 0.01% chance to enter chinese market with AOS4. So to me that language "project" seems very important and worth to push forward as first priority.

Hmm.. interesting point and a good argument for prioritize this higher. Especially with an apparently low-cost Hyperion/A-Eon laptop on the horizon, this could be a vital missing component. That convinces me... already donated to PolyVM, but I'll add another donation on Friday for the Perception IME part of Belxjander's project.

P.S. Given the diverse crowd of AOS users, are we sure Hyperion isn't already addressing this need in a near-term release?

_________________
Vic-20, C-64, C-128
Amiga 1000, 3000
AmigaOne X1000

@Belxjander

I do not remember if i ask before, but can you please point me what amiga apps you do in last 20 years ? I mean, can i download something from aminet done by you ? If you have nothing done, simple "i have nothing done before", will be enough. No big posts need it.

Last edited by kas1e on 03-May-2012 at 09:48 AM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@jingof

Existing donations have been towards a machine along with my own funds being put towards the same,
I'm only awaiting the postage information from the seller to my knowledge at this time.

Please be sure about your donations and Thank you for the support.

@kas1e

I have already repeatedly answered this, my prior work has been primarily helping others, I have not had any project with my name on it on aminet or os4depot.

@all

I'll be doing what I can with both projects and I've also published some code so far.

I've got a personal commitment to get Japanese at least presentable and a step up towards Chinese and Korean.

Last edited by Belxjander on 03-May-2012 at 10:12 AM.

So how's this Polymorph thing coming along? Because when I look at the "code" in the SVN all I see are a handful of stub functions, all of which do nothing and return NULL if they return anything at all. For example, I'd expect to see a bit more going on in the primarily initialization function then this:

void InitPolymorphDaemon(void)
{
return;
}


Is this the definition of algorythmic boundaries?

_________________
Here for the whimpering end

@Belxjander
Quote:

I have already repeatedly answered this, my prior work has been primarily helping others, I have not had any project with my name on it on aminet or os4depot.

Ok, so nothing was done before, its your first project.

@jorkany
Quote:

Is this the definition of algorythmic boundaries?

:))

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@Belxjander

Quote:

What I had in mind was to have each read/write mapped out using linear 1:1 translations where the virtual memory mapping provided by AmigaOS4 is exactly the same inside the Emulated CPU as outside within the AmigaOS native routines containing it and then deal with "exceptions" with any specific segment mappings as limited as possible.

Virtual memory in AmigaOS4 is nothing more then address space that can be swapped, the feature is only usable as long as you have less then 2Gb of RAM, Virtual Mem + Pysycal Mem can't be larger then maximum address space.

You can't use it for vitalization of memory aka Emulated memory address.
RM = AME + EA.

RM = Real Memory,
AME = allocated memory for emulation
EA = Emulated Address.

What your in fact saying her is that your going to use real address space whit in the Emulator, at the same time your going to have a emulated address space, that's going to be tricky because unless I'm mistaken your going to have to test each address if they are fake or real, do you see the problem, it will be just as costly as adding AME to EA, and also you might run in problems whit overlapping memory.

It will also require kernel hacking.

The only way you can deal whit real memory is your fake drivers or opcodes are reusing a address that you have recently been translated, so that do not need to translate again, the problem her is that this requires a complex JIT compiler.

Last edited by NutsAboutAmiga on 31-May-2012 at 05:43 PM.
Last edited by NutsAboutAmiga on 31-May-2012 at 05:40 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@kas1e

If this is his first project I think he might as well quit, this way more complicated then he thinks, he will not get anything done, because as soon he run in problem he need to redesign it.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS