@pavlor

Quote:

I used Rosetta as example of possible solution of compatibility problems connected to platform change. In this case MorphOS on "x86".

Unfortunately Amiga o.s. Design won't allow using Rosetta kind of solutions. Big endian software must run in an isolated environment. Thus UAE, or Janus UAE.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@pavlor

Quote:

If MorphOS Team decides port to another CPU architecture without maintaining compatibility (except UAE), it will be step to grave.

Perhaps. 68k binary compatibility is an asset you have or don't have. When you take it away you are giving away something what ties the operating system to Amiga.

But this risk need to be taken because PPC won't live forever.

Even if Rosetta was possible it would be too much work with limited resources.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Even Apple didn't wrote rosetta they just bought it from a company I don't remember the name so Amiga...

Kamelito

@kamelit0

Quote:

kamelit0 wrote: @mbrantley I think AmigaDave just want to be in the winning future but it doesn't exist. Quite frankly both side are enjoyable (68k vs PPC/X86/ARM). To me what is lacking on NG is full memory protection and ressource tracking, maybe Arix will bring those features before anyone else. While waiting I'll be playing with WinUAE and for a 1.0 core for my FPGAArcade. Kamelito

You are right, it does not exist, but you are wrong that I am looking for a winning side.

There is no "Winning Future" for the Amiga, other than the fact we are all winners because there is still an active user base and a few active developers who still care anything about the Amiga. And we are all losers because our platform has become so splintered and fragmented, that we fight with each other and the separation of so few resources really hurts the productivity and availability of new software, and improvements to the OS.

What is missing for you is full memory protection and resource tracking, but for me, what is missing is how the Amiga hardware, OS, and software, worked so well and intuitively together. It all made computing easy and fun, while also being well ahead of all the competition when it was first released, and maybe for a few years after that time.

Windows, MacOSX & Linux have all far surpassed the capabilities of the Amiga, even if they still lack the intuitive ease that was once almost unique to the Amiga. All mainstream OSes have progressed enough to make them "easy enough" to use, so why fight it by trying to use an Amiga-Like OS that can't compete for so many reasons. I choose to use the original AmigaOS for fun, programming simpliscity and nostalgia, and accept all of it's faults and limitations, but also support improvements, when and where they can be made without sacrificing the original workings that made the Amiga special.

_________________
Amiga! The computer that inspired so many, to accomplish so much, but has ended up in the hands of . . . . . . . . . .

@cdimauro

Quote:

It's clear that you don't or don't want to understand, but it's not my problem anymore.

So be it...

Quote:

I don't luck experience, but from what you reported it's clear that you have no idea of how an Amiga works and how it's emulated.

Quote:

No, you do NOT require double the CPU power every time that you double the resolution in one direction.

Bold words from someone without UAE experience...

Quote:

After that, you have to just scale the framebuffer, which is a very fast operation even for a low-end CPU, and certainly it do NOT require double the CPU power (for every resolution doubling).

In hires/double lines you render 640x512 pixels (or even more if you choose greater resolution eg. for overscan), whatever original resolution is (be it 320x256, 640x256 or 640x512).

Quote:

But honestly I prefer a technical explanation, since you seem to know many things about how an Amiga work, and emulating it. Thanks.

I wrote I have experience with emulators (and their limitations). Your previous posts about WinUAE (its RTG etc.) clearly show, how limited your knowledge is in this field. Still, you insist on standpoint, you even aren´t able verify.

Well, there is proof you may get even on your fast hardware (at least Core2) - emulation under emulation. Run WinUAE (with JIT) and then 68k E-UAE with configuration for different resolutions. Then your words will have some weight.

@itix

Quote:

When you take it away you are giving away something what ties the operating system to Amiga.

Exactly!

Quote:

But this risk need to be taken because PPC won't live forever.

Quote:

Even if Rosetta was possible it would be too much work with limited resources.

Looks like grim future for MorphOS.

@pavlor

when you want something "modern" it is the only option

There will be the new movie "AmigaOS alone at PPC"

@amigadave

The advantage of Aros 68k is that it is in development and there will be really new software for it that will not be available on 3.X. On the other hand there is also software that works on 3.X but not on Aros 68k. I personal prefer Aros to 3.X of course

@amigadave

Just to make this "real"!

In most cases... there are talkers and there are doers. (aka users and makers!)
Myself, I am just a talker and buyer.People like AEon's Trever is a doer/makers.

In most cases, the talkers flood the forums and talk. Now some talkers change into Doers by
paying up (usually a lot) and make a product/business!

So David, be happy, don't worry. There are Doers doing. They are just not on the forums talking.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: I don't luck experience, but from what you reported it's clear that you have no idea of how an Amiga works and how it's emulated. Quote: No, you do NOT require double the CPU power every time that you double the resolution in one direction. Bold words from someone without UAE experience...

Easy way to do not prove your claiming about doubling the CPU power for every resolution change. The perfect way to "argument" in a (technical) discussion...
Quote:

Quote: After that, you have to just scale the framebuffer, which is a very fast operation even for a low-end CPU, and certainly it do NOT require double the CPU power (for every resolution doubling). In hires/double lines you render 640x512 pixels (or even more if you choose greater resolution eg. for overscan), whatever original resolution is (be it 320x256, 640x256 or 640x512).

Please, show me where I've negated that the final resolution is 640x480/512 or more.

I quote myself:

"whatever is the real resolution of your screen. After that, you have to just scale the framebuffer"

Now, can you show me how an emulator handles a 320x200 or 320x256 Amiga game, and where is the doubling of CPU for each resolution doubling? Thanks.
Quote:

Quote: But honestly I prefer a technical explanation, since you seem to know many things about how an Amiga work, and emulating it. Thanks. I wrote I have experience with emulators (and their limitations). Your previous posts about WinUAE (its RTG etc.) clearly show, how limited your knowledge is in this field. Still, you insist on standpoint, you even aren´t able verify.

The only point was about how the RTG driver is IMPLEMENTED. The API calls can be software rendered or offloaded, but this does NOT change how the emulator works (internally; the engine; whatever do you want to call the task of emulating the chipset and the CPU).
Quote:

Well, there is proof you may get even on your fast hardware (at least Core2) - emulation under emulation. Run WinUAE (with JIT) and then 68k E-UAE with configuration for different resolutions. Then your words will have some weight.

This "proof" doesn't show the technical reason why you need double CPU power when you double the resolution in each direction.

To help your task, I report your words about this: "Radeon HD cards don´t support low resolutions (320x256 in this case), their users are forced to use muuuch slower 640x480. Improving UAE to support compositing could be solution there."

I've underlined the key point here, which clearly refers to the scaling of the original framebuffer (320x200 or 320x256) to the output framebuffer (640x480).

Apart this, I remind you that the original discussion about this point was this:

Me: "Unfortunately, chipset emulation takes a big part of the core"

Since after that you've stated this: "chipset emulation is CPU intensive"

I think we have reached at least a point, which let to close another, previously pending, one.

Me: "It's not fast enough, because using one core the CPU has to interleave between chipset AND CPU emulation."
You: "I simply think performance gain of such big task is not worth the effort"

So, using a different core for the chipset and the CPU emulation provides a performance gain, since the chipset emulation is "CPU intensive" (your words here) and steals much CPU power.

If a core runs only the CPU emulation, it can run as fast as possible (and only needs to sync with the chipset in some precise scenarios, which aren't common using some tricks).

@cdimauro

Quote:

This "proof" doesn't show the technical reason why you need double CPU power when you double the resolution in each direction.

This "proof" shows how this problem looks in real world - that is important for users, no?

Quote:

So, using a different core for the chipset and the CPU emulation provides a performance gain, since the chipset emulation is "CPU intensive" (your words here) and steals much CPU power.

So, you are returning to your original idea of multiple CPU cores for multiple emulation tasks. As problem we discussed is obvious only on low/old CPUs, I don´t see benefit of such great change of UAE code (Toni Willen himself stated even he is not able to divide 68k CPU and chipset parts on different cores). However, your scenario works in case of PowerPC emulation in WinUAE: 680x0/chipset part runs on one core and PowerPC emulation on another.

@pavlor

Quote:

Looks like grim future for MorphOS.

I dont believe you care

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Quote:

I dont believe you care

On the opposite! MorphOS is great OS with many interesting features. Sure, I´m mainly AmigaOS (and then AROS) user, but I always considered MorphOS as valuable part of Amiga tradition and do my best to promote it where possible. In my country, Czech Republic, there is small and friendly MorphOS community, I´m even member of main czech MorphOS news/forum site.

Your hatred towards AmigaOS 4 blinds you. Current AmigaOS users are potential future MorphOS users, not enemies!

Love&Hate married. Healthy competition is the Best. Using Win 7 atm and waiting for the Net & Printer for the XE...

Edit: This feels like home. Just as old fashioned like PR. From the XE. Tune a little bit here and there.
Hopefully OWB or others are made for todays standars for all Amigalikes. Have Fun!
Had so much fun with the older C-times that kindly brainwashed. Quite a neutral like but we'll see..

Last edited by PR on 31-May-2015 at 09:51 AM.

@pavlor

I do not think Itix "hates" AmigaOS but all camps are in reality very separated with just a minority of user using more than one (the same for most devs). That is propably even true for Aros even though it is free of charge and runs on different platforms. You see that on the untrue comments you often read. So the typical AmigaOS user does not care about MorphOS and Aros, MorphOS user not about Aros and AmigaOS and so on. Perhaps this will change a little with MorphOS running on SAM and X5000.

@pavlor

Quote:

On the opposite! MorphOS is great OS with many interesting features. Sure, I´m mainly AmigaOS (and then AROS) user, but I always considered MorphOS as valuable part of Amiga tradition and do my best to promote it where possible. In my country, Czech Republic, there is small and friendly MorphOS community, I´m even member of main czech MorphOS news/forum site. Your hatred towards AmigaOS 4 blinds you. Current AmigaOS users are potential future MorphOS users, not enemies!

To me this sound like you contradict yourself. First you play nice guy and then you dont play nice guy. Tsk tsk.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: This "proof" doesn't show the technical reason why you need double CPU power when you double the resolution in each direction. This "proof" shows how this problem looks in real world - that is important for users, no?

Absolutely. I've nothing to say if you report a user point-of-view.

The problem comes when from this PoV you move to the technical one, claiming things which aren't true, and for which you systematically avoid to give even a simple, higher-level (technical) justification.

Specifically, I tried to show in several ways how an emulator works, and that the operation involved when you select a bigger resolution screen, which is a very simple, and very fast, scaling (just doubling the horizontal and vertical resolution).

Any coder with a minimum of skills about the argument can confirm that this's one of the fastest operations that a CPU, even a very low-end one, can handle without problems.

What I fail to understand is why you continue to discuss about technical things without having the right knowledge. Experiences as a user cannot cover it.
Quote:

Quote: So, using a different core for the chipset and the CPU emulation provides a performance gain, since the chipset emulation is "CPU intensive" (your words here) and steals much CPU power. So, you are returning to your original idea of multiple CPU cores for multiple emulation tasks.

I never left it.
Quote:

As problem we discussed is obvious only on low/old CPUs, I don´t see benefit of such great change of UAE code (Toni Willen himself stated even he is not able to divide 68k CPU and chipset parts on different cores).

I already explained why it will get benefits, and I don't want to copy & past it again.

Yes, I know that the current (Win)UAE implementation is very different and cannot be adapted for the multicore idea which I proposed.

I also know very well that writing a new emulator is challenge by itself, and will be MUCH harder for such multicore approach. It poses some non-trivial problems (see the comments which I and megol exchanged), the tricks that I've in mind are difficult to implement (very low-level stuff; and something non-portable which requires ad-hoc implementations for specific architectures), and obviously debugging such "monster" will be a nightmare.

Nevertheless, I strongly believe that the benefits will be very consistent and worth the effort in an Amiga + RTG scenario (which is the most interesting one; for playing videogames the current WinUAE approach is the best), because is the one where performance matters and users ask/need more processing power.
Quote:

However, your scenario works in case of PowerPC emulation in WinUAE: 680x0/chipset part runs on one core and PowerPC emulation on another.

I think that with the PowerPC emulation it's even faster and simpler, because you don't need the strict synchronization which is required between the 68K and the chipset. AFAIK, the communication between 68K and PowerPC worlds is easier to handle and implement, but I haven't investigated such low-level details.

@amigadave

Quote:

amigadave wrote: There is no "Winning Future" for the Amiga, other than the fact we are all winners because there is still an active user base and a few active developers who still care anything about the Amiga.

Our viewpoints are very well aligned on this. Rewind 10 years and it was 2005: I was still using an A1200 as my main computer (all be it, expanded right to the max) and I still wished for the return of AmigaOS to the frontline of computing.

But as the years have ticked by I've come to realise that the Amiga will never be at the frontline of computing again. It's just not going to happen. So I've abandoned these pipe-dreams.

That doesn't mean I've abandoned Amiga computing, though. I still use my Amiga machines and I enjoy them for what they are. I still purchase new software for them (most recently PPaint) and I still buy new hardware as well. So I'm an active and supporting member of the Amiga community, but my goals have changed. Now I just want to enjoy the platform for what it is. For serious, modern computing, I use Linux.

It's amazing that in 2015 we still have new software and hardware being developed for the Amiga plaftorm. This is to be celebrated, and long may it continue. I will support it for as long as it does. But my expectations are firmly tied to it as a hobbyist retro computing experience. And this means I am able to be happy about Amiga computing and platform developments.

@cdimauro

Quote:

What I fail to understand is why you continue to discuss about technical things without having the right knowledge. Experiences as a user cannot cover it.

Sure, I don´t know UAE internals or technical details of Amiga emulation. However, user experience is most important there. I suggested experimental verification, you ignored such proposal. Your choice.

Quote:

AFAIK, the communication between 68K and PowerPC worlds is easier to handle and implement, but I haven't investigated such low-level details.

You may read about WinUAE/PowerPC transition on EAB, there are few threads from last year dedicated to such topic.

@amigadave

Quote:

Every Amiga user is free to make their own choice. I have just changed directions and perspective.

Is is just matter of finding right mix of Amiga and other hobbies

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook