Quote:

To put things a bit into perspective, I ran Nbench on both my 2GHz G5 machine and on my Chromebook, sporting a 1.7GHz Exynos 5 CPU. GCC 4.4.5 was used on the G5 and GCC 4.6.3 was used on the Chromebook.

G5 2.0GHz: Memory index: 9.299 - integer index: 11.381 - floating-point index: 13.520
Exynos 5 1.7GHz: Memory index: 11.070 - integer index: 12.591 - floating-point index: 12.016

Their performance is about equal on average, although I suspect that Altivec-optimisations in Mplayer, the G5 will perform quite a bit better for something like video playback. The Chromebook retailed in USA for $250 a year ago, BTW.

>> I wouldn't be surprised if in about 1,5 year from today, cheap smartphones will
>> be reaching the territories of the most advanced (single core) PPC raw power or
>> even outperformed them.

> I think that happened some time ago...

Even if we pretend G5 not going beyond 2.0 GHz, the decision if this statement holds true or not even for the present time will depend on the definition of "cheap".

I'm in for changing architectures. Whatever the next one will be, I'm sure it'll be thought out well and will provide for other software to be ported to MorphOS NG* easier.


* See what I did there

Let's appriciate what we do have at the moment. For the first time in over 15 YEARS I feel I have an Amiga that I can use for my everyday needs. The next system will come, we all know that, but before that we can enjoy a working system TODAY. Let's make that better, evaluate our options and when the day comes; buy ourself a new MorphOS system and go on yet another adventure

Just my 2 cents here: maybe something like this would make sense for MorphOS:

http://arstechnica.com/information-technology/2013/07/99-arm-based-pc-runs-either-ubuntu-or-android/

Linux is dead slow on these kind of systems but MOS sounds like it would fly despite the hardware's limitations. That's a really good argument for non-Amiga users to switch to MorphOS IMHO.

BUT, I trust the developers to know 1000% better in these matters than I do. The above statement just makes sense to me personally

Quote:

Interesting, thanks!

A one year old Cortex-A15 ARM device performing about on par with one of the G5 PPC machines that "we" consider to be top of the line (albeit not the fastest, but OTOH there are a bit more powerful Cortex-A15's as well). And the PPC machine did cost almost ten times as much when sold almost ten years ago, not to mention power consumption, etc. Really puts things into perspective indeed. And the new 64-bit ARMv8 processors are about to enter the scene on a broad scale, and these will absolutely run the PPC's we have into the ground.

These old PPC desktop/laptop machines are truly in the backwater. And if we were to compare to the 4th generation Intel Core i3 / Core i5 / Core i7, then things get *truly* depressing.

Anyway, I like the Mac Mini PPC I have today and MorphOS really makes the most out of it. The biggest thing I miss is enough power for decoding/viewing x264 1080p streams etc. And I am happy with the development pace of MorphOS, and the fact that there is a "roadmap" for a couple of more versions at least, so I will happily continue using it until something better comes along!

Quote:

I like that idea. And I'm convinced we will see similar devices powered with 64-bit Cortex-A53/Cortex-A57 CPU's within a year!


Quote:

I disagree there! I have an Efika Smartbook and an Efika Smarttop, they are using a single core Cortex-A8 (that even had some bottlenecks removed in a later version). It's not a speed daemon, but Linux with the appropriate drivers and optimizations does not feel dead slow even on that. Rather the opposite, it's very usable, and I know Konstantinos Margaritis of libfreevec fame claimed he used one as his main system at some time.

A Cortex-A9, especially with more than one cores (the device you pointed at seems to be available in up to quad core), is a great deal faster. So I can't imagine how Linux could be "dead slow", something must be terribly wrong there if that's the case.

I think you could generally say that Cortex-A9 is well into "G4" territory, and a Cortex-A15 is in "G5" territory. In 2014, the A53/A57 will leave these territories *far* behind. And it will still be highly integrated, cheap, cool running and power efficient.

I wouldn't mind MorphOS on that kind of a device at all!

Quote:

No, I would not either.
But then, you guys were the people that convinced me to take a look at ARM in the first place.
And we are not going to have to wait a year for the A57/A53 cores.

On the other hand, what Frank said is reassuring.
I've grown quite fond of the PPC and if it were not for the pricing issue...
Well this year's cheap dual and quad core A9 products kind of completed that argument.

And while ARM has plans to try to keep the 32 bit cores attractive, if the Chinese start pumping out cheap A53 and A57 products, this whole line may get pushed forward purely because of the economics.

The one thing I am waiting for is an ARM device that has decent support for a discreet gpu. Not that the core should not have one, but an upgrade path would be nice.

And, whatever happens, I'm already planning for future revisions of our current OS.

While it would not run MorphOS on more than one core, I am beginning to wonder if ASMP wouldn't be better suited to say running Linux or BSD on the other cores of a Quad core G5.

But hey, even that is the future.

Frank, Mark, thanks for posting.
I like being reassured, even though, so far, you and your associates seem to have made all the right moves.

> albeit not the fastest

Yes, and also not the 2nd or 3rd fastest.

> there are a bit more powerful Cortex-A15's as well

Yes, devices being on the market today include Cortex-A15 up to 1.9 GHz. Linearly scaled, this would perform about on par with a 2.25 GHz G5, so still slightly below the 3rd fastest G5.

> Konstantinos Margaritis of libfreevec fame claimed he used one as his main system
> at some time.

It may be helpful to know that he also happened to be of Genesi fame when he made that statement ;-)

> In 2014, the A53/A57 will leave these territories *far* behind.

A57 yes, but A53 announced as having one third of A57 performance probably not.

Quote:

Sure, but it doesn't make it less true. I have personal experience from the Efika MX, it's not a speed daemon but it's very far from being "dead slow", and it's perfectly possible to use it as ones main system for most daily usage. And the referenced Cortex-A9 machine is so much faster, hence my skepticism about *that* being "dead slow" (although I haven't got any personal experience from this).

Quote:

The main point with the A53 and the scenario where it will probably mostly be used, will be to function in a big.LITTLE configurations together with one or several A57's, as an idle running engine when the required CPU power falls below a certain threshold. I think the A53 is designed to be a side kick, not the star of the show. Even if it will play the first violin in some chips.

>>> Konstantinos Margaritis of libfreevec fame claimed he used one as his main system
>>> at some time.

>> It may be helpful to know that he also happened to be of Genesi fame when he made
>> that statement ;-)

> Sure, but it doesn't make it less true.

Sure, it doesn't make it less true that he made that claim ;-) I don't remember this claim being made by him but I have no reason to believe that your recollection is faulty here. What I mean is that his Genesi involvement and in particular his involvement in the development of the device he talked about may have led him to make a somewhat biased statement. In any way, it's easier for me to believe such statement when it comes from a real customer instead of an employee.

> it's perfectly possible to use it as ones main system for most daily usage.

I know I couldn't and wouldn't want to use a system based on a single 800 MHz Cortex-A8 as my main system today but I have no reason to doubt your statement is true regarding yourself. To each his own.

>>> In 2014, the A53/A57 will leave these territories *far* behind.

>> A57 yes, but A53 announced as having one third of A57 performance probably not.

> The main point with the A53 and the scenario where it will probably mostly be used,
> will be to function in a big.LITTLE configurations together with one or several A57's,
> as an idle running engine when the required CPU power falls below a certain
> threshold. I think the A53 is designed to be a side kick, not the star of the show.
> Even if it will play the first violin in some chips.

Thanks for (long-winded) confirmation that the Cortex-A53 won't have a stake in leaving G4 and G5 performance territories behind.

>Thanks for (long-winded) confirmation that the Cortex-A53 won't have a stake in leaving G4 and G5 performance territories behind.

;) !

No the A53 will not be a powerhouse, and yet I still expect it to be used by itself in some products. One third of an A57s processing power should still be pretty impressive.

> I still expect it to be used by itself in some products.

Yes, at least one product that uses Cortex-A53 without Cortex-A57 has already been announced:

http://newsroom.altera.com/press-releases/nr-altera-arm-a53.htm

And more:

http://en.wikipedia.org/wiki/Snapdragon_%28system_on_chip%29#Snapdragon_410
http://en.wikipedia.org/wiki/Snapdragon_%28system_on_chip%29#Snapdragon_610
http://en.wikipedia.org/wiki/Snapdragon_%28system_on_chip%29#Snapdragon_615
http://www.gsmarena.com/mediatek_working_on_cortexa53_quad_and_octacore_chips-news-7846.php
http://www.marvell.com/company/news/pressDetail.do?releaseID=5078

[ Edited by Andreas_Wolf 10.03.2014 - 23:52 ]

This is snaphots of the Alchimie X:

https://www.facebook.com/media/set/?set=a.464610800326398.1073741826.100003324455952&type=1&l=9477be6bc9

More available on http://www.meta-morphos.fr/index.php?op=edito

thanks Papiosur for getting back to the original topic...