Excitement about NatAmi

[Iggy]

Thomas, Gunnar and the rest of the Natami team probably could have released something awile ago if they'd stopped moving the target.
I still think using a real 68060 would make more sense than incorporating the CPU into the FPGA. And now they're talking vector units and other additions.
Good grief! What is the point?

[Iggy]

The LX board was made not to be sold but to prove that the MX board could be built.  In that time it was decided having ethernet and USB on the production board would be a big positive.  Why is this so hard to comprehend?


I am not on the team, perhaps you should direct your questions to a team member on the project's forum instead of speculating none shall exist - if you are indeed truley curious.  The plan is to release a fully functional PC.  To think less is to spread FUD.

So let's be clear, just because the board is developed and produced doesn't mean it will be for sale until a working Amiga OS(and/or AROS 68k) installation is running on it.  It also helps to have hardware in hand when writing drivers, no?

No one has suggested Thomas won't make good on the promises, but to claim that the targets haven't changed isn't fair.
I would have prefered the 68060 board that was announced be considered for the final production.The 68070/68050 development time has undoutedly slowed the introduction.

I also don't remember a statement being made about the board only being offered as a whole system.

[Iggy]

As I see it, that makes it less "multi-core" and more "multi-processor".

Could you stop posting your opinion and consider providing facts?
Xenon is very much an enhanced tri-core Cell PPE derivative.

And since both the Cell and the Xenon are in order processors, their actual performance per cycle is a little lower than many other PPC cores.

Multi-core or multiprocessor? Well, as I see it, if its on the same die its a single processor. Otherwise, where do you draw the line? Math co-processors used to be off die, and even then they were only considered co-processors. That may all the SPE should be considered.

[Iggy]

Very well said.

No problem Nicholas. I know you've been following the general discussions covering PPC systems. Over the last couple of years, I've frequently seen the Cell and the Xenon held as as superior architectures.
Its not that they're bad designs, but if you compare their performance to other PPCs they have a higher top frequency and a lower IPC.
If we were to see MorphOS or AmigaOS ported to these platforms, their performance wouldn't be much better than hardware we already have.

I'll wait for PA6T or G5 performance. Lower clock speed, but better performance.

[Iggy]

No-one is going forward with a design like the Cell for any platform, it's an evolutionary dead end. Better to spend more on the CPU than be stuck with something so awkward to develop for. Besides tasks/jobs are a common way of feeding multiple threads which is simply what the task/job system for the SPEs hid from you by using the SPURS library.

Instead of SPEs we'll see more fully formed AMP systems where the SPEs can actually access main memory and handle more than 32k in or out at once.



This is the relevant part, extending it with SIMD, pipelining the N68050/70 better, making it OoO etc are probably all  a better idea right now than making the system SMP capable.

There's good argument for having another CPU to offload tasks to like the GP2X's second ARM cpu be used for or the PlayStation2's IOP. That's more like treating the second cpu as a co-processor.

OK nothing 'Snarky' to this comment. Its interesting to see other groups/camps within the Amiga community considering asymmetrical multiprocessor applications. We've discussed this idea on MorphZone a few times as a method for utilizing additional cores without breaking compatibility with legacy software.
Unlike SMP though, it seems more suited to OS operations than as a method of running additional threads.

You guys have impressed me with your grasp of the potential applications for this.
Should this approach become common in Amigiod platforms, how do you all think it will complicate cross platform development?

[Iggy]

Is this story still going...


The PPE was an *existing* PowerPC core that was taken and rejigged for the Cell using new circuit design techniques.  There's nothing spectacular about it other than it runs at 3.2GHz and only uses something around 25W on 90nm.  Something nobody else got even close to.

IBM took the PPC part and used it in the 360's Waternoose processor (yes, that's the actual codename).  

It's not actually that important though.  All the action is in the Vector units.  In the 360 they added on a heavily customised version of AltiVec.  In the Cell they used SPEs.

You can see the similarity here:




BTW the design techniques are not that spectacular either, domino logic has been around a long time and it's been used in ATI GPUs, and is in use in the iPad / iPhone CPU.  It was going to be used in the AMCC Titan but it got canned.  It'll likely appear in more Apple chips as they bought one of the companies who specialised in it.

Better, but still not 100%. Yes the design created for Microsoft has several improvements related to floating point instructions.But both processors, while fast, still suffer from an in order execution pipeline that standard PPCs (which can execute instructions out of order) do not have.
The designs have similar cores, but these are not standard PPC cores.
Virtually all current PPCs have a superior IPC which allows them to perform well even at lower frequencies than the Cell or Xenon.

While I like the Xenon, were I to design a new PPC based PC motherboard I'd look at Freescale's QorIQ rather than a Cell derivative.

[Iggy]

Whoa! That would be nice to hack. A big improvement the Wii. Might be a good base for a Linux hosted version of AROS.