Two more weeks
It is still in final design stages. We may post up some benchmarks of a test configuration as and when we have them (for instance to show performance of database apps, we use this to pick the final processor speed etc.)

One thing people are worried about with the OSW is the CPU speed as above; don't worry. We will be determining it based on price/performance. For instance there is very little likelihood of the OSW being a 2.7GHz power box like the PowerMac was. If that was the case it would cost you $3000 or more just like the PowerMac did.

But we can accurately determine what processor speed you would be happy with. And if you can get 60% of the performance of that, for at the end 30% of the price, I guess you could always buy 3 if you really wanted the power.

That's why we don't mention numbers yet; it simply hasn't been comprehensively benchmarked. The same kind of benchmarking, we did for the 7448, which is why you haven't seen one of those yet.

_________________
Matt Sealey

was it the poor benchmark performance (according to what was expected) for the 7448 that prevented Genesi from making 7448 based CPU modules for Peg2 ? Was it that bad ? A lot of Pegasos users (including myself) would still appreciate a Pegasos 2 CPU module update for linux needs.

Is the CPU soldered onto the motherboard? Or can the customer choose the CPU that comes with the motherboard?

--jc

_________________
--
Jimen Ching (jching@flex.com)

It doesn't have poor performance. The problem is that for the double-sized cache compared to the 7448, the cache is a little slower (by around 10 to 15%) than the one on the 7447A. That means you get twice the space to put data in, but can't access it as fast.

However it does run at clock speeds 300-400MHz faster than the fastest 7447A. But that comes at a cost to your wallet.

So it's a trade-off. The 7447A is cheaper, and runs better as a desktop chip. Like I said in the previous post, if you can get 60% of the performance for 30% of the price, you can always buy 2 or 3 to make up your performance goals.

As for why we didn't make a 7447A yet, we just didn't run out of 7447 chips, and it seems fairly nutsy to buy a batch of 7447A's just to throw a batch of perfectly good 7447's away.

If someone wants to order 7448 chips on CPU cards, we can do this, for example if a customer or developer or another board vendor wishes to do performance testing with a certain application to see if it is the right chip for them.

@jching: The 970MP will be soldered down for the time being.

_________________
Matt Sealey

"But we can accurately determine what processor speed you would be happy with. And if you can get 60% of the performance of that, for at the end 30% of the price, I guess you could always buy 3 if you really wanted the power. "

No problem, processor speed can be lower than 2.7Ghz.

If arround 1.8 to 2.2Ghz the machine is not draining to much electrical power. A 4xG5 à 1.8Ghz would be a real overpowered dream machine !

I own a 2x1.25Ghz G4, this machine runs very very nice ( speed is really impresive compared to PCs ) and cool.

Personnaly, I use my machines to surf, mail, video chat, sometimes video and photo editing. I take much care to electrical power consommation than to absolutily peak max horses power. Four G5 is well far over my needs now and in a near futur. So, less Ghz ( and less heat ) would really be fine.

And... woudl it be possible to have two version of the OSW ?
One with less Ghz and another with more Ghz for Ghz fans ?

Will the Peg2 will be upgraded soon ( dual G4 .. ? ) in ordrer to create a complete panel of Genesi offer ?

So, one week to go ?


Are we really going to have news about the super-G5 computer, before the end of the upcoming week ?

Any chance down the track it would be socketed. and also which type of socket would be used?

If my memory serves me right Neko once mentioned that maybe there will be a socketed version of the OSW....

If we bother him enough he surely will give in

Two more weeks followed by a smiley does most likely mean: We do as fast as we can, but a date cannot be given yet.

That's good arguments, but there's one factor you didn't mention: Heat.
Doesn't the 7448 run cooler?
What if "performance" was not measured in SPECint, but in dB and Watt? Would a higher "performance" system (read passively cooled) be possible?

I don't know about anyone else here, but I would GLADLY pay Genesi more money for a faster processor. I'm not saying that a 1.5 processor isn't plenty fast for 99% of what I would be doing on this system, but from an end-user enthusiast perspective why skimp on it? I can understand of course from Genesi's point-of-view that you want to save as much money as possible while producing a good product to sell... which is why there should some type of processor upgrade path for this system. I'm sure I'm not the only one around here who is interested in having the fastest Power-based computer on the block, and I'd like for that system to be the OSW.

_________________
Joshua Purcell

I second that

Our point of view?

What Sun, IBM and others are realizing is that open source software yields better outcomes than proprietary software for both producers and customers. The biggest and best example for now is of course GNU/Linux. Nokia and Finland are taking this collaboration farther. We are too. Solaris is next.

Computers are commodities? We don't think so.

The first phase of acceptance for open source began as being a way of laying off a lot of new development expense on developers who were not inside the commercial organization. The largest hardware vendors cannot afford to hire enough software engineers to do all the development of software that they would like to do. The way for them to address that problem is to insure the availability of an open source platform. The way that Genesi has encouraged other people to do this development is to provide them with machines.

That is what the OSW is, a development platform, and what the ODW has been. An integrated Power host and target with a full toolchain and a myriad of environment options. The EFIKA is actually something more, but it is that too. Like the Pegasos, the EFIKA can be deployed in many different configurations. A good example of PegPower:

http://bbrv.blogspot.com/2006/05/road-warrior_17.html

What has been the ultimate objective? Getting the bigger players to understand the approach. It was a challenge with "microelectronics types" and CPU salesmen that might not understand that system enablement meant more silicon sales, so we started fishing for bigger fish (or said in another way being caught ourselves in a bigger net). In any case, today we have Power.org and the intent in building an ecosystem around Power, while not completely understood by 90% of the people proclaiming its virtue, it is still being proclaimed. Enter the PowerProjectCenter.

Channeling the enthusiasm and involvement into useful collaboration is the key. With the OSW we are where we were with the Pegasos/ODW a couple of years ago. But, having a solid entry in the 32-bit to 64-bit transition is not a bad thing. After all, the only time that a dominant platform gets toppled is when the hardware platform changes out from under it, and the biggest driver of hardware platform changing out from under it is best with the changes. We saw one change in the 8- to 16-bit transition. The Atari of today is not the Atari that made 8-bit computers.

The 32-bit to 64-bit transition going on now, which we think is going to be the best window for a long time to achieve Power market share. There will be a dominant 64-bit operating system and that could be Linux, or it could be something else. The opportunity comes in that we are changing the "rules." The field of play is over here now. That is the transition we need to leverage (with some help from IBM). We can do that with the OSW, building off the work we have done with the ODW.

The coolest thing though is the EFIKA. The Pegasos was sold to hobbyists, developers, the embedded market and academia. The OSW will be more of the same, but as it is out on the market edge. We think the benefit to Genesi will increase by a factor of three to ten times, depending on the excitement we can create through Power.org. Having said that, the EFIKA is MUCH, MUCH more.

We will explain that another time...

R&B

_________________
http://bbrv.blogspot.com