Good question.
There are of cause several options to solve this.
As we know the EFIKA will comes with new GFX chips on board. Genesi has all drivers for so these chips so they could be supported by MOS too.
And creating a CyberGFX VESA driver to even provide basic functionality on all GFX-cards should be possible too, right?

We will see.

The same for me. OSW without MorphOS is out of my interest. But there is other problem except of drivers. MorphOS as we know it today, is a single-CPU, 32-bit operating system. It means it will use a single core out of 4 ones and also will force CPU to run in 32-bit mode. If Pegasos III is clocked at 1.25 GHz, it means it can be slower running MOS, than Pegasos II with G4 (well probably it will be somewhat faster thanks to better memory controller). So, if we are talking about MorphOS, it is possible that we have no (or little) gain for $700. I doubt that ABox environment can be made using SMP. QBox is another story, but it seems it has been forgotten forever.

http://samsung.com/Products/OpticalDisk ... _H553A.asp
http://www.plextor.be/products/px-755SA ... e=PX-755SA

Easily. There is no reason to move away from our choice of Samsung for burners, but there are at least 3 other burners coming to market in the next months (even more later, since Intel don't put parallel ATA controllers in their chipsets anymore, this is driving the market now).

We would hope to see XGI drivers for MorphOS at some point. We have the documentation for the chip on the EFIKA, plus a Linux driver with 3D.

For the OSW it will have PCI Express so your choice of cards comes from that.

There is a possibility still of a legacy PCI slot in the OSW (see the block diagram) so at the very least it may be possible to ship a PCI 9250

I'm sure the cores will run with 2.5 GHz each.
Designing a G5 based server with only 1.2 Ghz
would be really odd in the year 2006 IMHO.

Cheers

Not really. Multicore, low-frequency chipsets are really common in 2006.

You think too much about MorphOS, we are thinking of current design methods and Linux and workstation usage :)

Yes, if MorphOS would run on it.

I am not interested in this pool.

I will buy one Pegasos III/OSW when I will got some funds...

...But I will not buy anything if I will not see SMP MorphOS announced, presented, demoed for real, and finally running on it for true.

I have no need for 64 bit computing on CPUs at its first time, It is good for me even if MorphOS will run on G5 in 32 bit emulation in the very beginning...

...BUT the new MorphOS it should be SMP from the scratch and run flawlessly on any core of the dual G5.

This is all I ask for!

Of course i would upgrade,
but only if mos supports it!

I am guessing that most of the already supported (and supportive) Linux distros will work out of the box. Even better, with the chip used in the OSW you could get some 64 bit joy out of this. Some distros will do a 64 bit kernelland, 32 bit userland, and then Gentoo can do it all 64 bit kl, and 64 bit ul.

With the possible HPC applications of the OSW, getting the most out of those procs and dealing with large applications/datasets, 64 bit use is going to be a must.

Now, here is where my ignorance of MorphOS sets in, I'm not trying to trash it but I don't know what it's capabilities are here when it comes to the 64 bit 970. (the first and last time I played with MorphOS, there was not even a TCP/IP stack included). I am just trying to show a voice of support for Linux use on the OSW. +1 on the poll.

Cheers!

I'm a yet another linux+multimedia developer, having the possibility to work on the G5 exploiting 64bit and altivec to improve some new codecs like h264, snow, dirac could be quite nice.

Another interesting feature is the possibility to test mutithreaded applications, like a certain rtsp server named fenice,
on real cpus and not just stressing the kernel scheduler, well stressing it in a different way and the cpus =)

First: 32bit isn't emulation like ia64 or a crippled legacy of the past like on amd64 (on 32bit mode you won't have access to a number of gpr iirc), you just miss the 64bit but nothing else.

Second: Having an improved I/O would already give you some speed boost, even if not as the one the linux people will experience (think about having a full aiglx/XGL setup with lots of eye candy and still enough cpus to do something more serious)

Looking at the OSW Diagram i see users ending up with one HD and a CDRom.
There should be at least the option for a second HD...



That is true for IBM POWER 4 and POWER 5 but the PPC 970 was designed to clock higher than the "Big Iron" Server Processors with 144MB of on Board Cache. For me this means the OSW with Core Frequency at 1,25 GHz will totally suck in single thread performance (slower (maybe equal) compared to the some year's old G4 at 1.25 GHz). Parallelisation does not always work and always only to some degree.

Don't get me wrong i really like the idea of the OSW but some parts i would like to have changed.

So please, consider at least 4 x Serial Ata channels and the option to buy higher clocked OSW's (Exchangable CPU modules?). Didn't vote for now because i have to make up my mind first...

>>>If Pegasos III is clocked at 1.25 GHz, it means it can be
>>>slower running MOS, than Pegasos II with G4.

>>Designing a G5 based server with only 1.2 Ghz
>>would be really odd in the year 2006 IMHO.

>Not really. Multicore, low-frequency chipsets are really common in 2006.

Still, G5 is no Niagara! :-)
Intel is going 3 Ghz with Woodcrest and this will be your competition.

Matt, what are we looking at here for the maximum wattage on the OSW?