> http://moobunny.dreamhosters.com/cgi/mbmessage.pl/amiga/160792.shtml

Yes, I know the e700 route isn't pursued by Freescale anymore for a long time. My question wasn't really meant seriously :-) Also power.org's Power Architecture Silicon Roadmap doesn't mention it anymore.

From your moobunny posting:

> the e600 as current top edge Freescale ppc core just handles its
> tasks well

Then why are Freescale going to put the e600 EOL (according to BBRV)?

> to me the most [...] promising way freescale will continue with ppc
> is the QorIQ network processor route - IIRC there the e500 core s used.

Right, it's e500. And that's as well the reason the QorIQ is not the least promising for me as a desktop Power Architecture (i.e. MorphOS) user. The e500's (and e200's) FPU implementation is incompatible with the one of the e300 and e600 (and the PPC4xx, the PPC7xx and the PPC970 for that matter).

Interesting comment on LimePC and CherryPal by Mr. Campbell dating January 4th:

http://greentechgirl.com/green-computing/cherrypal-limepc-architecture/

Jack Campbell has a vivid imagination and lots of energy, but Matt and Andre contributed more to what the LimePC was supposed to be than Jack. Here is a an image that should help substantiate the interaction. MorphOS was supposed to have been featured on the device.



The MyMate became what was later known as the LimePC. One other thing...

Jack also has a big problem with the whole truth, but that never seems to get in his way.

R&B

he didnt even mention genesi at all...

anyway, is he talkin about morphos in this?

"We had a radically unique new user interface/desktop OS version sculpted for this project, one that would have minimized the UI load on the little MPC5121e CPU, and that had a super high level of optimization work done to make the graphics/framebuffer/display subsystems use as few clock ticks as possible"

> is he talkin about morphos in this?

I'd rather say he's talking SymphonyOS/Mezzo things there. Or is he? BBRV?

It is hard to say what Jack is talking about now. Originally, it was all about MorphOS. We also considered a dual boot option with GNU/Linux+Qtopia. After the cache-coherency issues surfaced at the end of October and the complexities of completing a port of any EFIKA-working software became more apparent, the switch was made to SymphonyOS. Of course, the other reason was that THTF did not want to pay for any of the software development. Read this...



In the end, the Investment Banking Firm was able to get commitments for the $20 million, but THTF did not deliver from their side *and* tried to steal the whole package to themselves. There is more, but that ought to convince most folks the opportunity was real.

Looking ahead, the i.MX515 is going to end up much better though not PowerPC based. The same Investment Banking Firm remains interested and we will see what happens over the next few months. There have been considerable changes in the financial markets since September when that letter was written. In the meanwhile, you can be sure Gerald is up to his eyeballs in technical documentation for the new chip already, while the rest of the team has alpha level access to all the general details. We post more about all this on PowerDeveloper when we can.

R&B

|||

[ Edited by Velcro_SP 27.04.2011 - 06:59 ]

@ Velcro_SP

As efficient as the Atom N270 is (at 1.3-1.6Ghz the specs say 2W-4W) it is always coupled with the Intel 945GC/945GM northbridge, which has specs topping 25W in use.

You cannot use the Atom without that bridge, and Intel are not moving Atom to the "integrated northbridge" model they are pushing for their new high end quad-core chips (possibly because it would increase the die size way past their target, and die size directly impacts cost).

Then of course on any design, you have to add RAM power costs; this can be anything from 3W to 10W depending on how much you have.


Compare an equivalent PowerPC - the MPC8610 or MPC8640D at the lower clock rates - and you are looking at chips which do all the Atom and 945 combination do, in a single chip package. This makes layout easier, which makes PCBs easier to design and smaller, and of course even the dual-core PowerPC chip uses less power with zero power management than the Atom and 945.

Of course the MPC5121e does better, but it's nowhere near as good as CherryPal say it is. Not by a long shot.


To contrast the i.MX515 uses a lot less than the MPC5121e plus it has highly aggressive power management as all ARM chips do.

[ Edited by Neko on 2009/1/25 19:02 ]

Some additions to my own posting:

> it's e500. And that's as well the reason the QorIQ is not the least
> promising for me as a desktop Power Architecture (i.e. MorphOS) user.
> The e500's (and e200's) FPU implementation is incompatible with the
> one of the e300 and e600

The above is only true for the QorIQ P1 and P2 platforms, which use the e500v2 core, but not for the eight-core P4 platform, which uses the new e500mc. Up to now I wrongly assumed the e500mc to be just the multi-core (with "multi" greater than 2) enabled variant of the e500v2. As a look into this nice document revealed, that's not the case.
Obviously and to my surprise, the e500mc got the pre-e500mc's embedded FPU implementation replaced by an e300/e600 compatible FPU implementation. Furthermore, the pre-e500mc's SPE SIMD (not to be confused with Cell's SPEs) is gone in the e500mc. But unfortunately, it didn't get replaced by AltiVec SIMD.
In summary, as a MorphOS user I have mixed feelings about the QorIQ. The variants matching an OS that is not SMP capable are of no use due to the incompatible FPU implementation, and the variant matching the way FPU support is implemented in the OS is an overkill eight-core beast. And AltiVec is missing generally. A single-core e500mc based CPU (as suggested as an implementation possibility in the linked document) would be nice, though.


Edit:
Link to video because PDF document seems to be taken down.

[ Edited by Andreas_Wolf on 2009/9/24 5:41 ]

Quote:
Unfortunately the FPU is halfclocked only. The QorIO is rather for I/O ops and not that much for mathematical stuff like media processing.
Quote:
I don't see the QorIO as a primary MorphOS target, but from Freescale's view I see there the biggest potential for the Power architecture. One thing which might be cool for MorphOS though is the guest OS handling provided by the chip. Maybe this could be useful somehow. Though I don't know a (at least not ad hoc) scenario where MorphOS would be beneficial to have as guest OS (except making MorphOS lovers happy).

Quote:
Oh really ? I'm surprised you couldn't do that :p

Btw showing here screenshots of your incoming box, and scans of confidential docs: how serious is it for your company ?

|||

[ Edited by Velcro_SP 19.04.2011 - 16:38 ]

|||

[ Edited by Velcro_SP 19.04.2011 - 16:22 ]

> there is some proprietary issue with the Linux display drivers so
> they plan to ship fit-PC2 with a previous Ubuntu version that has
> drivers

...which *is* the proprietary one, incompatible with latest Ubuntu.

Quote:

Intel Atom netbook platform with the 945GSE Express chipset has a specified maximum TDP of 11.8 Watts.

ASUS Eee PC 901 uses 945GSE (Calistoga) Express chipset.
ASUS Eee PC 4G Surf (701) uses Intel 910GML (Alviso-GM) chipset and it has 10.5 Watt max TDP.


The problem here is between the keyboard and the chair.

[ Edited by rvalencia on 2009/4/5 7:49 ]

[ Edited by rvalencia on 2009/4/5 8:03 ]

|||

[ Edited by Velcro_SP 19.04.2011 - 16:16 ]

Quote:

Well, here is someone trying to sell it to the Amigaworld.net crowd:

http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=28546&forum=2&start=0&viewmode=flat&order=0

|||

[ Edited by Velcro_SP 19.04.2011 - 16:34 ]

Quote:

An underpowered 5121 at ~400Mhz is the best ppc response to a ~1.0Ghz ARM CPU with a NEON SIMD unit? Riiight... I'm -or rather I was- a powerpc advocate myself, but seriously, spec-wise, the Limebook looks like a joke at this moment. Personally, I wouldn't bet my money on it. Face it, powerpc has no future anymore, at least none that would interest desktop users.

Quote:

The problem is that the LimePC is broken by design, for all intents and purposes a stolen Genesi product proposal made it come into fruition. Except for overexcitement on the part of one executive, MorphOS was not in their interests at all, and it's no surprise that the department at Freescale that was dealing with THTF no longer exists.

If they did a deal with Hyperion for OS4 it'd be out of desperation on their part (THTF have nothing to sell right now). I would feel very sorry for Hyperion if they did it... it'd be another Mai Logic style situation all over again. We all know who came out better for that fiasco.. Genesi, who quit while we were ahead and found a better opportunity.

Quote:

Please don't latch onto a processor line for no good reason.

Cortex-A8 and Cortex-A9 cores are really something special. It may not run MorphOS but if you're looking to make a Netbook or thin client, PowerPC is probably the worst choice in the current market.. it would need significant cooperation from the CPU vendor and other parties to make it work, and that cooperation - because of the complete lack of excitement about the architecture these days - is not forthcoming.

[ Edited by Neko on 2009/4/8 19:12 ]

> Except for overexcitement on the part of one executive, MorphOS was
> not in their interests at all

BBRV said on several occasions that the LimePC devices were originally to run mainly MorphOS. How does that fit your statement?

|||


[ Edited by Velcro_SP 19.04.2011 - 16:33 ]

@Velcro_SP
>You signed up for MZ to post that?

I support ASUS.

>At least the guy in the test report hooked up a power meter,
The guy tested a desktop chipset.

>you just run your mouth.
You didn't look at the fine details. RMclock tool also shows the power consumption.

Must I post ASUS W3J with it's mobile 945 chipset wattage?

"Mobile Intel? 945GSE Express Chipset"

http://ark.intel.com/chipset.aspx?familyID=35553

Note the "Max TDP 6 Watts"

Ordering items
1. 82945GSE Graphics and Memory Controller Hub (GMCH) FCBGA998 A3 6 Watts
2. 82801GBM I/O Controller Hub (ICH7M) PBGA652 B1 3.3 Watts
Total: 9.3Watts

Take note of "Valid Processor Combinations"
"Intel? Atom? Processor N270 (512K Cache, 1.60 GHz, 533 MHz FSB)"

Now for ASUS. http://en.wikipedia.org/wiki/ASUS_Eee_PC
Note "Intel 910GML series" and "Intel 945GSE" for ASUS EeePC.

Try again Velcro_SP.



[ Edited by Hammer on 2009/4/9 3:39 ]

Quote:

please, spare me the lecture about resource efficiency. I wrote libfreevec and am expanding it to a full-blown libc. In fact, my work is *exactly* on software optimization, getting old enterprise software and making it orders of magnitude faster and more efficient. I've optimized several algorithms for AltiVec/SSE and working on ARM NEON now. I think I know pretty well what resource efficiency is.

The fact remains that 5121e being power efficient does not make it usable (esp. by the vast majority of users who are used to Vista or "sluggish" Linux. Btw, Linux can actually be very fast and efficient, you just skip KDE/Gnome.

But still, you're missing the point. My argument was not about efficiency, it was about using PowerPC and the 5121 in particular. However efficient LimeBook is, it just won't compete with the Atom and even the ARM-based netbooks. Face it, PowerPC is out of the picture.

[ Edited by feanor on 2009/4/9 11:43 ]

Intressting topic I must say

Quote:

I agree calling Linux sluggish is not really true, just look at distros such as Damn Small Linux or vectorlinux if you want speed.

Quote:
Some questions that I find intressting.
I not very good at CPUs but is there any other ARM CPU than Freescales i.MX515 that is in the range for a netbook?

What makes ARM intressting instead of going with Intel Atom?

Is anyone using ARM in a computer today (as the main CPU?)?

[ Edited by robjoh on 2009/4/9 9:32 ]