@Leo

Quote:

Of course. For me it shows either it's not possible or that it wouldn't worth the effort... and wouldn't work correctly.

But in real life it shows AROS needs work in other areas that are more importiant than SMP and MP. But now you have your chance to make a difference, step up with a plan on how to bring it to AROS. Or make a fork of it.

Quote:

Unless some head of AROS development clearly mentions it and it appears in official roadmaps it won't be "obvious"... And I don't wanna waste my time if that's not clearly the target. I want a new OS, UAE is fine for running 3.1 software.

Then understand the fact that AROS is not for you. That's not how AROS development work. There's no master controlling it. You have to waste you time on a fork of AROS and be the master yourself. You say what goes in and what goes out. The perfect world for you, right? AROS is alive because it's a fun community to participate in, devs work with the things they like and for the fun of it. You want to control them and you can't that's what pi***es you off.

Quote:

It's not about the progress, it's about the target, which isn't defined.

No it's seems that you've misunderstood why AROS even exist still after those years.

Quote:

Unless a new modern OS is the goal, and compatibility is dropped/gets less to no ressources assigned to it, I surely won't waste my time...

If not wasting time is so important to you then you should not comment on development of an OS you aren't one bit interested in that is IMHO wasting your time because no one really listens to you. And PS. don't even think of getting a hobby that only cost you money because it's fun, that must be the utter wasting of time.

_________________
AmigaOS or MorphOS on x86 would sell orders of magnitude more than the current,
hardware-intensive solutions. And they'd go faster.-- D.Haynie

Quote:

If Hyperion develop something in private with no consultation, then what if developers don't agree with the solution? It's no good for anyone if we end up with another ReAction - incompatible and inferior to the alternatives.

Since when do they care ?

_________________
http://www.warpdesign.fr/

Quote:

Then understand the fact that AROS is not for you. That's not how AROS development work. There's no master controlling it. You have to waste you time on a fork of AROS and be the master yourself. You say what goes in and what goes out. The perfect world for you, right?

Quote:

You want to control them and you can't that's what pi***es you off.

To control what ?! Of course not. Why would I want to control something I don't have enough skill for ?

I don't have interest to manage (nor... "control") a project, nor do I have enough knowledge to define some kind of API, nor to work at kernel level on anything. So I won't do it (unlike a lot of people who just go... into the wall).

I just don't understand the need for classic compatibility (besides UAE) right now, in 2010. So I don't understand why they spend so many resources on a such limited core (yes, it is limited: you want to live with it, fine.. but that's your choice... or rather, obligation, since you have no choice).

I can't do anything of the above, but I can do a lot of other things. I'd be ready (and happy) to spend some free time (yeah, no need for a bounty) to help, if only the targeted OS was something new that would leave behind all this useless legacy/limitations. Right now it's not clearly the case.

I'll be happy to contribute to Haiku for example (even though I don't control any single thing in the development process lol and fortunately :))

But I wish some true NG AmigaOS would emerge. Not something full of legacy and limitations... But it seems it requires a lot more time for people to understand it's not needed to shoot yourself in the foot... This doesn't mean I can't follow what's going on, and express my opinion, right ?

Last edited by Leo on 29-Mar-2010 at 08:07 PM.
Last edited by Leo on 29-Mar-2010 at 08:07 PM.
Last edited by Leo on 29-Mar-2010 at 08:06 PM.

_________________
http://www.warpdesign.fr/

I would suggest that AROS, of all the Amiga family of operating systems, is the one that should probably pursue SMP as a matter of some priority. The reason I say this is that it's the only one that currently runs on hardware where multiple core CPUs are commonplace.

_________________
Doing stupid things for fun...

@Leo

Quote:

But I wish some true NG AmigaOS would emerge. Not something full of legacy and limitations... But it seems it requires a lot more time for people to understand it's not needed to shoot yourself in the foot... This doesn't mean I can't follow what's going on, and express my opinion, right ?

No you have to understand that people do understand it. You aren't teaching us anything new. It's not that hard to point out what an OS like Amiga needs. Doing it and how to do it is where you should put your energy. Get in to details, join the AROS dev mailing list.

But since I can't deny you having your opinion then that means I can have mine too, and my opinion is my gut feeling and that is that you like to come down hard on AROS only to upset people. But to be honest Leo, I'm not loosing any sleep because of it.

Last edited by Manu on 29-Mar-2010 at 08:41 PM.

_________________
AmigaOS or MorphOS on x86 would sell orders of magnitude more than the current,
hardware-intensive solutions. And they'd go faster.-- D.Haynie

@Leo

There is no future in the Amiga World, it's just shades of the past. Once you understand that you can appreciate and enjoy AROS, AOS 4 and Morphos....

@Manu

Quote:

But since I can't deny you having your opinion then that means I can have mine too, and my opinion is my gut feeling and that is that you like to come down hard on AROS only to upset people. But to be honest Leo, I'm not loosing any sleep because of it.

I have yet to see him take any cheap shots at AROS. Mean while, back in the real world, a early release for the upcoming AMD 12 core Opteron data proves to be an interesting read. For those OSs that support it, of course.

_________________
Dammy

@Manu Quote:

my opinion is my gut feeling and that is that you like to come down hard on AROS only to upset people

Naaah, he just comes across as one of those who is now burnt-out & disillusioned with the whole Amiga thing, and wants to ensure everyone else "sees the light" (or should that be darkness?)... Which is quite sad/unfortunate. Probably he'd fit right in on the MooBunny

That should stop us continuing to enjoy whatever flavour of AmigaOS we like. Why worry about the unknown future, if we can't control it? Better to enjoy the present, and hope for the best IMHO.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@damocles

Quote:

damocles wrote: @Manu Quote: But since I can't deny you having your opinion then that means I can have mine too, and my opinion is my gut feeling and that is that you like to come down hard on AROS only to upset people. But to be honest Leo, I'm not loosing any sleep because of it. I have yet to see him take any cheap shots at AROS. Mean while, back in the real world, a early release for the upcoming AMD 12 core Opteron data proves to be an interesting read. For those OSs that support it, of course.

Opterons are server CPUs, 12-core CPUs are not that amazing in the server world (16-core SPARC CPUs were announced back in 2007 for instance) . Amiga-like OS's are not suitable for servers for a number of reasons, and never were, even before the age of SMP, I hope you weren't suggesting it as a likely niche for AROS.

Oh and if you can't see Leo has been slagging off AROS and the direction AROS is taking since the beginning of this thread I suggest you open your eyes and look again.

I don't know how many times this has to be repeated before it sinks in but nobody is saying no to SMP. If you really want to see it happen you have to plan how you want to see it happen. Planning is not the same as listing your must-have features, it includes goals and details on how to achieve the goals.

You said before Anubis is practically dead, what do you think might help bring it back on track? A decent plan. It didn't have one before, it sure as hell could do with one now, wouldn't you agree?

How about this for restarting Anubis' direction. Go back to the Anubis mailing list and start a topic about what a modern implementation of datatypes would look like. That's something concrete to discuss, and is a key feature Anubis would need to be truly considered Amiga-like. At least it might get some constructive discussion going. If the project truly is deader than a dodo already it only shows there was a lack of direction and commitment from the get-go, which is sad.

Last edited by HenryCase on 29-Mar-2010 at 11:22 PM.
Last edited by HenryCase on 29-Mar-2010 at 11:22 PM.
Last edited by HenryCase on 29-Mar-2010 at 11:20 PM.

I give you my situation.
I want something amiga or amiga like.
Why ? cause it is what I grew up with. It's what I like to use. Sure I use Linux too. Still I enjoy my time with AROS most.
AROS does not feel old to me. Even if AROS would not be developed further I would use it for the rest of my life. Every little step forward make me happy and I'm sure it's the same for Amiga and morphos users.

What would AROS be without being amiga compatible. Just another OS project.
It would be dead long time ago.

To all people not supporting amiga world bug off. Find something else to do.

@damocles

Quote:

damocles wrote: Mean while, back in the real world, a early release for the upcoming AMD 12 core Opteron data proves to be an interesting read. For those OSs that support it, of course.

I have a 4-core cpu in my desktop box and I can probably count the number of times all four cores have been fully occupied for more than a few seconds on my fingers. Aside from "make -j 4" and scripts that deliberately fork 4 mencoder instances, it just doesn't happen in normal use.

To effectively utilize a larger number of cores, you need a situation in which you have a very large number of processes that are all ready to run and don't communicate with each other often. The only place you see that are in HPC and server applications.

In the few instances where parallelism can accelerate some desktop task, it's usually the case that it is some sort of stream processing application where using a GPU is far more sensible.

_________________
Doing stupid things for fun...

@HenryCase

Just because it's a server chip doesn't mean users don't use it... until I upgraded I was running an opteron 1216. I'm pretty sure the point is why get left behind when the 12 core x86 chips are on the way(since the athlon is mostly a cut down opteron after all.).


Come to think of it ... I might buy one when it's available :P

Last edited by terminills on 29-Mar-2010 at 11:49 PM.

_________________
Support AROS sponsor a developer.

"AROS is prolly illegal ~ Evert Carton" intentionally quoted out of context for dramatic effect

I like to see better GPU support too.
It sounds more Amiga like to me.
SMP,Memory protection,etc is fine too.
I am sorry that I can not get the time to program.
I got much to do.

_________________
John 3:16

@HenryCase

Quote:

The Barrelfish project is an example of a multikernel OS. I pointed it out as I thought you might be unfamiliar with the concept. I accept that these features will be important at some point (of course it would be nice to have now if we had more devs, even just to stop the whining), but what I take objection to is people who spread myths that AROS will never have these features without massive reorganisation, despite the evidence to the contrary.

Erm, the entire Amiga, and thus AROS approach is based on transferring data via a single shared memory. SMP and Memory protection is impossible without major API changes that will break this model and therefore all the software that relies on it. i.e. the entire Amiga software base.

Multikernel is a way of organising a kernel, it says nothing about memory. it solves neither of the problems.

If there was an easy way of adding SMP or memory protection, don't you think someone would have done it already?

Last edited by minator on 30-Mar-2010 at 01:39 AM.

_________________
Whyzzat?

@HenryCase

Quote:

Opterons are server CPUs, 12-core CPUs are not that amazing in the server world (16-core SPARC CPUs were announced back in 2007 for instance) . Amiga-like OS's are not suitable for servers for a number of reasons, and never were, even before the age of SMP, I hope you weren't suggesting it as a likely niche for AROS.

No, I was showing that the rate of multi core CPUs expansion with growth in number of cores being produced. 12 core server means we may see a reasonable priced 6 core Phenom in the near future. No one is going to do any server work beyond hobby server work with a OS lacking multi user, MP and SMP.

Quote:

You said before Anubis is practically dead, what do you think might help bring it back on track? A decent plan. It didn't have one before, it sure as hell could do with one now, wouldn't you agree?

I have little hopes of Anibus being revived. Perhaps something close to it with different management and some type of small budget for core Devs to work part time. Or go do something different with the same end goals. Such a pity.

_________________
Dammy

@minator

Break compatibility with the past and implement memory protect on an AROS fork. Let's join the 21st Century before it's the 22nd...

@Karlos
Quote:

Karlos wrote: I have a 4-core cpu in my desktop box and I can probably count the number of times all four cores have been fully occupied for more than a few seconds on my fingers. Aside from "make -j 4" and scripts that deliberately fork 4 mencoder instances, it just doesn't happen in normal use. To effectively utilize a larger number of cores, you need a situation in which you have a very large number of processes that are all ready to run and don't communicate with each other often. The only place you see that are in HPC and server applications. In the few instances where parallelism can accelerate some desktop task, it's usually the case that it is some sort of stream processing application where using a GPU is far more sensible.

Thank you Karlos, for reminding us that most users aren't seeing the true benefits of multi-core yet. Might even take a move away from the interrupt-driven nature of current CPUs before we can unleash the power of SMP.

@terminills

Quote:

terminills wrote: Just because it's a server chip doesn't mean users don't use it... until I upgraded I was running an opteron 1216. I'm pretty sure the point is why get left behind when the 12 core x86 chips are on the way(since the athlon is mostly a cut down opteron after all.). Come to think of it ... I might buy one when it's available :P

Users might want it because it's the latest and greatest, but whether they'll see much benefit from those extra cores, even on a mainstream OS, is another matter (see Karlos' last post).

@minator

Quote:

minator wrote: Erm, the entire Amiga, and thus AROS approach is based on transferring data via a single shared memory. SMP and Memory protection is impossible without major API changes that will break this model and therefore all the software that relies on it. i.e. the entire Amiga software base. Multikernel is a way of organising a kernel, it says nothing about memory. it solves neither of the problems. If there was an easy way of adding SMP or memory protection, don't you think someone would have done it already?

Arrrgggghhhh, it's like people don't know how to read... minator, in the very same post you quoted I pointed to the memory protection solution that would be suitable for AROS, AmigaOS, MorphOS, etc... because it details how to take a single shared memory OS and give it full memory protection. Here you go:
http://130.203.133.121:8080/viewdoc/summary;jsessionid=103DA74DEBD7CC3130A374F517E75851?doi=10.1.1.16.9704

If this memory model is different enough to warrant a paper written about it, we're looking at something that's different enough from most coders minds to not be obvious, so possibly this is why it wasn't considered before (I had thought of something similar, but I lacked the ability to describe it in this detail).

Quite frankly, no I don't believe it's inevitable it would have been done already. First of all, I never said it was easy, just that it was possible. Secondly, when did the first multi-core CPUs appear in the desktop market? As far as I can tell it was the Athlon 64 X2 back from mid 2005, so we're talking 5 years ago.

In case you hadn't noticed Amiga isn't a mainstream platform anymore, so there's bound to be some lag in getting features like this. Furthermore, the only Amiga platform capable of running on these commercial multi-core machines was AROS, so considering AROS has only in the last year or so got a proper USB stack you can see the sort of lag we're talking about.

I hope you can see that saying 'why hasn't it happened yet' is a straw man argument.

Quote:

damocles wrote: I have little hopes of Anibus being revived. Perhaps something close to it with different management and some type of small budget for core Devs to work part time. Or go do something different with the same end goals. Such a pity.

You don't need money, you need committment.

@minator: at long last... I agree, thank you.

What's the use of keeping compatibility with 20 years old software anyway ? It's not like Amiga had some office/visicalc suite being used by thousands of people, in which case, compatibility could be needed...

Yeah, you all people want to run DOpus probably ? But if the system's manager hadn't been so limited in the first place I'm pretty sure DOpus wouldn't have been created... So better implement a new powerful filemanager than working on compatibility.

_________________
http://www.warpdesign.fr/

@HenryCase

the average user is the one who uses smp lately. decompressing files, while playing games, while browsing the web for cheats, while encoding videos. Yes I do this quite often ps3 and xbox 360 video servers typically encode videos in the background. Yes I don't need 12 cores but we do need smp(gpgpu won't help in this type of scenerio)

This is another place memory protection is needed.

Also hyperthreading came out in 2002 which was a virtual multicore.

@persia
I would like to see the 64 bit Aros forked for that purpose.


Tim

Last edited by terminills on 30-Mar-2010 at 09:51 AM.
Last edited by terminills on 30-Mar-2010 at 09:15 AM.

_________________
Support AROS sponsor a developer.

"AROS is prolly illegal ~ Evert Carton" intentionally quoted out of context for dramatic effect

@Karlos

"the only one that currently runs on hardware where multiple core CPUs are commonplace."

Yes, but it seems that there are so many other basic things that in AROS are behind other OSs like AOS4.
And since AOS4 developers have had multiple PPC cores on their hand for years, I hope something is already prepared behind the scenes & AOS4 delivers SMP this year.

Last edited by KimmoK on 30-Mar-2010 at 09:57 AM.

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