MorphOS on Power Mac G5

[Iggy]

Aren't the G3 and G4 PowerMacs to slow to run MorphOS or just UAE?

or most G4's and then you would have to be careful what you got because of motherboard changes and upgrade ability (with the exception of some pricey processor upgrades)

And then the other question will MorphOS eventually make the move to intel as Apple has done?

Its unlikely that G3s are going to be supported. G4s are certainly fast enough (the fastest common Pegasos uses a G4 at 1Ghz). Right now, they've only shown Powermac support for the last G4s, the MDD/FW800 models. Since they top out at 1.42Ghz they should be a upgrade to most Pegasos owners.

Most of the users of the Mac Mini seem quite satisfied with their machines (which top out at 1.5 Ghz).

All these G4s are going to be faster than the boards manufactured by Acube (who's Applied Micro PPC based boards use less powerful CPUs).

The G5 is going to be a brute. G5s processors are slightly more powerful then the PA6T and are available at speeds up to 50% faster then that of the X1000 (at a much lower price).

[Iggy]

Untrue. The PA6T outperforms the G5 quite comfortably at the same clock, at least for floating point and is on par for integer performance. Try your google fu and you'll soon see.

http://www.amiga.org/forums/showpost.php?p=566253&postcount=31

The advantages of the G5 mac are cost and maximum clock rate. Both of which still equate to a better price/performance ratio than the PA6T, of course.

That doesn't actually agree with figures I've received from other sources. In fact, while the PA6T clearly beats Applied Micro's new Titan cores and other Applied Micro products the drystone figures I've been given (which is an alternate if fairly artificial measurement) reflect different performance figures than your floating point comparison.

These comments were post by Andreas Wolf on MorphZone.

Just for the purpose of overview I compiled a list of the DMIPS/MHz per core figures for various PPCs in ascending order (AltiVec capability is generally ignored, obviously):

e300: 1.9
PPC440: 2.0
PPC460: 2.0
Titan: 2.0 (presumably PPC450 based)
PA6T: 2.2
PPC750: 2.3
e600: 2.3
PPC470: 2.3 or 2.5 (varying with information source)
e500mc: 2.5
PPC970: 2.9
e5500: 3.0

What seems really weird is the rather low value for the PA6T (which was provided by P.A.Semi themselves btw), especially compared to the PPC970, which I thought it should have been on par with. Or maybe the PPC970 figure is too high (i.e. with AltiVec)?
[unquote]

While PA6T figure may be low, its important to remember the figures came from PA Semi.

Frankly, I don't think the PA6T demonstrate a "quite comfortable" advantage and if these figures are even close to correct then even a G4 at comparable speeds may offer competition for the PA6T.

Once you balance both these measures, you come to realize that at similar speed the performance of the PA6T and the G5 are going to be close. And the PA6T is a far more energy efficient processor. Of course, on the other hand, the G5 (PPC970) operates up to 50% faster.

The real shocker is that while it doesn't support nearly as good floating point performance, Freescale's new 64bit e5500 communications oriented core may outperform all of these (at similar speeds).

[Iggy]

The G5 was in no means a disaster. Apple's primary dissatisfaction was in IBM's inability to get this chip to scale up in speed high enough. The decision to move to X86 had been considered tor years. Blaming it on the G5 isn't completely fair.

>Quote:
But it's better to pay 1/4 of the price for one kind of "obsolete" hardware, than paying £1500-£2000 for another "obsolete" hardware that performs **worse** than the cheaper one...>

Damned right! Frankly, I consider this a hobbyist pursuit, so the less I pay the better. And as a user of AMD processors, performance per watt was not really a concern I was worried about (I have no problem with 125watt processors - my water heater uses 5500 watts). Overall performance, yes. Electricity costs concerns making me want to buy a computer that costs $2000 more and doesn't perform better, no.

      >Quote:
                    Finally, what will you do on your obsolete PPC platform of choice that requires either a G5 or PA6T that you can't already do more than comfortably on a G4?>

What are you trying to do Karlos, make our point? I have an eMac that was purchased dirt cheap. Sometime this summer I'm likely to see my Powermac (also purchased dirt cheap) supported under MorphOS.
And, if 7448 processor upgrades are supported, I'd love to run some real benchmarks against the X1000.

Right now, except for the AmigaOne and the Pegasos, most AOS machines are based on Applied Micro based SAM boards. Even DVD decoding is a challenge to these boards (although they can do it, which is probably better than the Natami will do).

So, no I don't need any better computers than the G4's. And I'm not fixated on new hardware, but some AmigaOS users are. And that leaves them with a choice between under powered and over powered.

I'm looking forward to Nemo, as I've been in contact with the designer in the past. I've been impressed with Varisys since I started using XMOS components. But the X1000 is not based on the most powerful PPC, that's still the 970/G5 (unless you want to count the cell and its related products or future Freescale releases).

And this isn't a pissing match, its just healthy competition. Someone out there is going to be able to afford to buy those high priced X1000s with AOS4 (I hope). And I'm still going to be happy using something more economical.

[Iggy]

That was a well thought out and succinct reply. I find myself, once again, in agreement with you..
I'd rather see a focus on better video cards and software development. Otherwise I'm happy with where we are now.
If we focus solely on the high end of our markets' hardware, we snub developments in systems people can afford to buy.

Right now, I'm not nearly as excited about the announcement of eventual G5 support as I am about the latest statements that All AGP G4 Powermacs are likely to be supported by MorphOS (as soon as this summer).

This type of hardware is everywhere and, to me, it seems better suited to MorphOS than 64bit multi-processor speed machines.

And AOS vs. MOS, its an apple vs apples argument. Who cares?

[Iggy]

Yes Red, its going to be fast and they are cheap, but this summer we may see the next revision of MorphOS supporting all the AGP G4 Powermacs which are cheaper still and more than fast enough.
You've been using an eMac similar to mine. The Mac Mini and the eMac are already pretty fast aren't they?
I'd just like to get this massive one piece unit off my desk and get back to using my LCD display (and recover some of my desk space).

Currently, the development teams next move is to support the AGP G4 Powermacs (apparently pretty soon).
After that, we'll probably see Powerbook support some time in early 2011.

Who knows when we'll see G5 support? But you're already moving a few eMacs and I've already built a G4 Powermac.

Rather than spending a lot of time discussing potential hardware, I wouldn't mind focusing on what's here or close to here.

[Iggy]

What about iBooks?  I have this iBook that I've replaced with a MacBook and it is just waiting for Morphos to give it a second life!

You're amongst a large group of people who'd like to see support for those. But, right now it looks like only Powerbooks.

The MorphOS developer you need to bug goes by the name Bigfoot. Apparently none of the developers feels like writing the necessary routines for ADB keyboard/touchpad support.

[Iggy]

It could open the door to port more powerful programs and games that may struggle on a G4. But, as you already said "Why not just use a $50 PC and be done with it! "

Primarily because I already have a PC (with a power hungry AMD processor that's also useful as a space heater). It plays back HD video content fine.

Now I'd like similar capability on my PPC hoobyist computer.

What is actually more of a problem with HD content on Amiga related PPC systems is the lack of HD video decoding functions built into the older video cards we support. While my X86 processor could still decode these videos, CPU usage would be much higher without the video card I'm using.

While MorphOS plays DVD level content fine on G4s, I have no doubt it would struggle with HD. However, should anyone ever get ATI or Nvidia to fully document their later GPU releases, designing drivers to enable this playback on lower processors (thasn the G5) should be possible.

You don't think I want to waste all my CPU power on HD playback do you?

[Iggy]

Sure, but the list of apps you just cited are all surely available for your PC already and, unless it is equally retro, it should be able to run them better than the G5?

I don't know. My slowest X86 PC is a dual core with a standard clock of 2.6Mhz running at 3.0.
The fastest G5 runs at 2.7 (and while there are dual and quad machines, MorphOS can only use one core).
For tasks that aren't heavily threaded and when not running a lot of concurrent apps, the G5 could potentially match the performance of the X86s.
It would definitely outperform most Intel Atom systems.

Flash...I was wondering if I was the only one who missed that (instead of just attacking it). Well, I can't get that (or a least a good variation of that) under PPC Linux either.

[Iggy]

The latest iBooks G4 had kdb/TP by usb and not by ADB. For these models the adjustemnet of the drivers may be an effort worthwile to consider.

With all the noise about them, I wouldn't be surprised to see them supported. Thanks for the point about those Zylesea. It doesn't sound like there's a lot separating the final ibooks from the Powerbooks. What video GPU did they use?

[Iggy]

I downloaded the linux 64-bit version whilst it was available, but it seems they withdrew it again. I'm still using it though, it's fast enough

Most people need flash for youtube but to be fair the HTML5 stuff seems to have that base covered nicely. Especially in OWB/MOS.

Valid point. I have no interest in playing Flash based games. I wish I'd caught the 64bit version you mentioned, I still regularly dabble with Linux (probably because even with my dissatisfaction with it, I appreciate its UNIX underpinnings).

[Iggy]

It's open, I can probably send you the .so

That is a good point and it would be useful.
Its always good to see that even when we've got open debates raging, we're still (mostly) a civil lot.
I'm not even vaguely aware of how to send a PM on Amiga.org, but I can get you may e-mail address somehow.

Thanks

[Iggy]

I'll agree with all the pro G5 points I've seen except for Dave's one on price. Older G4 AGP Powermacs are routinely given or thrown away (rather like some Commodore hardware).
There are later G4 Powermacs (933Mhz and above) starting at $.99 on Ebay. I have a Quicksilver Powermac that built from components that cost me well under $100.
Unless your talking bout 1.6 or 1.8 Mhz G5s, the prices are almost always significantly higher.
For truly fast G5 of 2.5 to 2.7 Mhz, while they are at good prices, generally they command about $500 or more.
While I agree with you that there's nothing wrong with more power, I can't agree with you that G4s and G5 have similar pricing.
We're just lucky that Apples switch to Intel processors has so devalued the G5s (since their still remarkably powerful computers).

[Iggy]

As far as the OSX graphics overhead. At an Apple Con an Apple employee told me OSX interface is totally PDF vector based... So there you go. Still amazing the fucntionality of it on the Imac G3400 with tiger.

I thought Apple graphics were rendered via OpenGL. Are you sure that they weren't referring to graphics functionality in documents. Even before OSX, Apple seemed to rely on PDFs for printing (or at least supported postscript).
Somehow, I just can't see what PDF format would have to do with screen display and  vector functions (while they may be the wave of both the past and future) would not convert efficiently in OpenGL calls.

[Iggy]

You tell 'em Red!!!

AS for OSX pdf thing... I know, it didn't make sense but the way he explained it to me was that I should look at the icons... they never pixelate when you use the genie effect becuase they're rendered in vectors. When I aske the goy some other questions he cas totally clueless and told me apple flew him to the convetion to fill in for someone else.

Long live the obsolete hardware!
Commodore diehards prove that obsolete doesn't mean it can't still be useful.

What, an Apple expert that may not know what he's talking about? No....

[Iggy]

G5s have definitely gotten a bad rap as far as their thermal properties go. Apple didn't help by using liquid cooling on the last units. It makes it look like that was necessary, it isn't, it's quieter. G5s at 2.7 Ghz have no more rigorous cooling requirements than an Intel or AMD processor (in fact, compared to some of those, their cooling requirements are lower).

And I understand Karlos' argument really well. I fact, since I recognize that I'm not going to be doing anything with a G4 Powermac that I can't do with a 1.25Ghz eMac, the first Powermac I've put together is a 1 Ghz Quicksilver. I'm pretty sure I can overclock this to 1.2 Ghz (which is close to my current 1.25) and if third party CPU upgrades are supported I can go as high as a 2.0Ghz 7448 (the fastest MDD upgrade is a 7447 which would be less powerful).

I don't expect to be able to decode HD video without high CPU usage or a video card based decoder. And, as before, there's going to be some applications that the system is either not powerful enough for or that software doesn't exist for under MorphOS. I also not pinning my hopes on hardware that's not going to be available in the immediate future and I'm glad to see that AmigaDave thinks the same way.

Will G5 support be cool? Will we be able to keep pace with AmigaOS4 hardware? Of course, its going to seriously kick ass, and we'll be paying less for our hardware.

The thing that Karlos is aware of and that I must admit I'm painfully aware of (as someone who worked for a company selling 68K based computers in the late 80's and early 90's) is that even if we have reasonably powerful hardware we're facing an impossible task.

Its simple, with the number of Windows and OSX computers sold each year the market for software for those machines is very large. As the time and energy necessary to create good software is also quite great, developers gravitate toward those markets with the greatest potential to make a profit.

As such, many great ideas and even better alternatives (to the current dominant platforms) have failed in the past. We need to keep our feet on the ground and remain realistic. The chances of a mainstream revival are slim. Our market does seem to be enjoying a period of renewal and growth. But, our systems and our software still serve a hobbyist market. Whether they will ever serve as primary systems of production remains questionable.