PPC 470 / A2 is also on it's way. Dunno about its availbility, but is quite interesting: - 1,6 GHz - Altivec + Fpu + DSP Mac - cache L2 (in multicore design only)
x86 CPUs are little-endian ... you couldn't have x86 executables running and using the AmigaOS API together with emulated 68k and PPC executables in the same environment.
Of course you can, you simply make the x86 act like a big-endian CPU, by modifying GCC to generate extra byte-swapping instructions on memory accesses.
And this has already been done for Amithlon years ago, proving it works. The overhead is also apparently quite small, especially when you consider x86 runs 3GHz compared to 0.5 to 1.0 GHz of our PPC machines.
Sadly there are valid reasons why OS4 won't go x86, but this isn't one of them.
Of course you can, you simply make the x86 act like a big-endian CPU, by modifying GCC to generate extra byte-swapping instructions on memory accesses.
And this has already been done for Amithlon years ago, proving it works. The overhead is also apparently quite small, especially when you consider x86 runs 3GHz compared to 0.5 to 1.0 GHz of our PPC machines.
There will be overhead and this means the same code will run slower than under other OS'es, thus AmigaOS could no longer be more efficient than other OS'es. To do constant byte-swapping for every memory operation for native code would be a pretty high price to pay for compatibility. If x86 were able to operate in both big and little-endian modes like PPCs, it would be a different story, but they can't. But this has been discussed to death in many threads, and is off-topic here, so please take it elsewhere.
the problem is that if ACube would start designing a board based on the 460EX after they finished the Sam-Flex, by the time they would have a product on the market, in my opinion the 460EX would be just as underpowered for the needs of the time as the Sam440 is now. I think the Freescale MPC8610 would be a better choice. Not only is it available at higher clock speeds (up to 1.33 GHz)
I think that you accidentally make the point that Amiga hardware developments are always going to be FAR behind the x86 world, so that people complaining about "lack" of GHz (compared to x86) are always going to be disappointed:
It is quite arguable that 1.33 GHz is "underpowered" for a modern system in a year or two (if not now already). How are we going to decode Blueray in real time? (Never mind whether this is a sensible thing to be want to do on an Amiga.) How are we going to do play Doom 3, after the source code is released? (Never mind our lack of a modern & efficient OpenGL implementation.) How are we going to do all the latest whizz bang things from the Windows world? (Never mind that those wanting that are always going to be disappointed, because we will always be playing catch-up.)
IMHO we should be enjoying the strengths of the AmigaOS & Amiga hardware, rather than trying to compete with x86 (which we can never do).
But this has been discussed to death in many threads, and is off-topic here, so please take it elsewhere.
Hmmm, is it off-topic? I saw quite a few x86 comments previously, and it seems related to how the conversation has gradually flowed from the original one.
Anyway, if you had just said that, then maybe I could have avoided discussing it any further. Unfortunately you also said the following:
Quote:
There will be overhead and this means the same code will run slower than under other OS'es, thus AmigaOS could no longer be more efficient than other OS'es.
If you only use the Amiga because it is more efficient "under the hood", rather than any actual experience, then I can understand your statement.
But for me, and I suspect many others, I like the AmigaOS because it FEELS fast & efficient, not to mention it is easy to understand & customise. Byte swapping would have no effect on that, and in fact by embracing byte swapping it can run FASTER (than sticking with PPC).
Summary: I would accept the compromise of byte swapping, to get OS4 on x86, while keeping backwards compatibility (for 68k & PPC software).
Quote:
To do constant byte-swapping for every memory operation for native code would be a pretty high price to pay for compatibility. If x86 were able to operate in both big and little-endian modes like PPCs, it would be a different story, but they can't.
Remember that register operations (such as byte swapping) take far FAR less time than memory operations (even when it hits the cache). Which explains why it apparently has little effect on performance. And even if it halved performance (!!!), we would still get a 1.5GHz performance (on a 3GHz x86) for a price far less than a 1.5GHz PPC mobo designed for a small market. But as I said, there are other reasons why this won't happen.
It is quite arguable that 1.33 GHz is "underpowered" for a modern system in a year or two (if not now already).
I don't think it would be underpowered. A 460EX-based board would be, if you want good multimedia performance.
Quote:
How are we going to decode Blueray in real time?
I can already play 720p content nicely on the 1Ghz Pegasos2, even though it has a very slow memory interface (only 133MHz FSB) and very slow video bandwidth (AGP x1) and only a few parts of avcodec.library are currently Altivec-optimized. Because the 8610 has a much faster memory interface (533MHz DDR2) and video bus (8-lane PCI-Express), I'm convinced that combining AltiVec with that high bandwidth would give exceptional multimedia performance, in fact the 8610 is advertised especially for things like real-time image processing. Besides as soon as BluRay is widely adopted, all gfx cards will have built-in decoder acceleration for them.
Quote:
How are we going to do play Doom 3, after the source code is released?
@COBRA If a 1GHz PPC machine can play Blueray, then great , although I personally don't care about it. I was just trying to think of unreasonable examples, to make my point - which was that comparing OS4 to Windows/x86 land will *always* leave people disappointed. (e.g. Javascript performance, since once x86 becomes fast enough, people will use it for things that were previously considered insane.)
Quote:
What makes you think you couldn't play Doom3?
IIRC (one of?) the Frieden brothers has said that our OpenGL implementation has serious bottlenecks, because it is implemented on-top of Warp3D.
No doubt you COULD play it if you turned off most of the shaders & other stuff that makes it look great. Or if you ran it at a really low resolution. But otherwise I would guess that an OpenGL-heavy game like Doom 3 would kill our current OpenGL implementation.
I also suspect that Doom 3 puts a heavy strain on the CPU - John Carmack likes to push everything he can. The "minimum requirements" for Doom 3 a "1.5 GHz Intel Pentium 4 chip or AMD Athlon 1500", which in reality means it would not perform very well on such a system. So unless we have a 2GHz PPC system, I doubt it would be too much fun, even assuming everything else was optimised as well as on x86.
If you want yet another one of those x86 discussions, then please open a new thread and try to keep on-topic here.
Talking of 8610 is equally off-topic as x86. You started saying it's taking too long to release Flex boards and started to discuss why they didn't replace it with something more powerful yourself in the first place ! I expressed my opinions only. I was PPC fanboy 3 months ago but I'm not anymore. AMCC's primary market is embedded and not desktop and Freescale might go bankrupt... But then again... (I don't want Intel's everywhere either because I'm personally against monopolies.) Raw CPU power is never bad. Anyway let's hope Titan will be released very soon. (Of course this is all only speculation and we can't know what Acube & Hyperion are planning and wanting to do.)
PS. And everything always depends on what are business targets of Hyperion and Acube. Are we happy staying very small fanboy niche forever. Or do we want to get more ex-Amigans back from other platforms.
Rock lobster bit me - so I'm here forever X1000 + AmigaOS 4.1 FE "Anyone can build a fast CPU. The trick is to build a fast system." - Seymour Cray
Sadly there are valid reasons why OS4 won't go x86
Slayer moonwalks backwards then drops down to the half splits
sorry couldn't resist, I find that comment rather comforting regardless of whether it'll stand the test of time.
I also support your general overview of a mentality of enjoying AmigaOS instead of worrying about its position in the cpu chain of performance, that has always been my way of looking at things. AmigaOS is a great cpu experience and I don't know how people use these crappy IBM or other environments. They get the job done and that's all most people want, and good on them for.
If you'd rather go rock climbing and down the river in the boat etc instead of burning a dvd or playing a 360 game then good for you, just don't expect me to admire you for it or consider your life is more fuller than mine.
not unlike those stupid cocacola ads which shows a sad looking person in a high rise apartment looking sadly down at a festival going on in the street and the logo is something like - where do you want to be?
I always think [censored] I like being up in the apartment away from all that BS!
Life is a damn perception and that is all it is
~Yes I am a Kiwi, No, I did not appear as an extra in 'Lord of the Rings'~ 1x AmigaOne X5000 2.0GHz 2gM RadeonR9280X AOS4.x 3x AmigaOne X1000 1.8GHz 2gM RadeonHD7970 AOS4.x
IIRC (one of?) the Frieden brothers has said that our OpenGL implementation has serious bottlenecks, because it is implemented on-top of Warp3D.
That is correct, but what does that have to do with the choice of CPU of future hardware? That is a matter of software and drivers, and it would not change even if it ran on a Quad-core intel beast.
Quote:
No doubt you COULD play it if you turned off most of the shaders & other stuff that makes it look great. Or if you ran it at a really low resolution.
The resolution or performance of shaders has nothing to do with the choice of CPU either.
Quote:
Sorry, but if you & others brings-up x86, without any complaints, then it seems unreasonable to tell me I can't reply to any of them.
I didn't say don't reply to comments, I said if you want to continue discussing the topic of x86 vs PPC, start another thread and continue there.
@TSK
Quote:
Talking of 8610 is equally off-topic as x86
I agree that discussing different PPC choices ACube could use for future boards is slightly off-topic in a thread about Sam-Flex, but I disagree that it's equally off-topic as x86. Discussing what other developments ACube could/should consider beside the Sam-Flex is closely related to the topic, but the topic of x86 is as far from the original topic of Sam-Flex/ACube as you could possibly get, besides it's been discussed to death in a dozen or so other threads already, and this is not the appropriate thread to have another one of those endless "why not x86" rants, so let's stop that and get back to the topic at hand, OK?
The flex is definitely an improvement in terms of value for money over the original Sam440, so obviously it's a good thing :) The point I was trying to make is that since it's apparently taking so long, they would have been better off making a 460-based (or 8610-based) board in the first place as they could already be prototyping it by now and it would have a much larger market (especially the 8610-based board as it'd be more powerful than any existing OS4 hardware).
I think that you, like many people tend to think that it's always possible to do what they want. I believe that if ACube does the Flex now there is a damn very good reason to do it this way.
One reason I can think of is that many months are needed to build a hardware. It means that when the decision to do the Flex was taken, maybe they didn't have the 460 yet. And there are probably many other reasons I can't think of
good day for me, my shipment of Amiga Classic gear arrived today (lots of stuff including 10 Amiga systems) and then this!
I'm going to get one and then use my exisiting SAM as a back up board!
Time to clear the credit card
~Yes I am a Kiwi, No, I did not appear as an extra in 'Lord of the Rings'~ 1x AmigaOne X5000 2.0GHz 2gM RadeonR9280X AOS4.x 3x AmigaOne X1000 1.8GHz 2gM RadeonHD7970 AOS4.x
[...] I just hope that the same thing that happened to eyetech doesn't happen to acube, ie the bulk of people waits for them to release something better but since the early products didn't sell enough, better never happens.
I've ordered my SamFlex right away..My A1 is thousands of KM away, and those 6 months without OS4x where sad.. I'm soo addicted...
News feed
OS4 Feedback forum
OS4Depot Feedback forum
Amigabounty forum
OpenAmiga forum
AmiCygnix forum
ABC shell forum
AmiKit forum
Cinnamon Writer forum
CodeBench forum
Digital Universe forum
Dopus 5 forum
E-UAE forum
Gnash forum
Ibrowse forum
JAmiga forum
Odyssey forum
OWB forum
Qt forum
SmartFileSystem forum
Timberwolf forum
TouchDevice forum
TuneNet forum
Unsatisfactory Software forum
Donate
, although I personally don't care about it. I was just trying to think of unreasonable examples, to make my point - which was that comparing OS4 to Windows/x86 land will *always* leave people disappointed. (e.g. Javascript performance, since once x86 becomes fast enough, people will use it for things that were previously considered insane.)
