> there is something like the e5500 a 64 bit core derived from e500 for QorIQ in
> the pipeline, but no e700 (which would AFAIK a 64 bit derivate from the e600) .

Seems some Wikipedia editor's opinion differs from yours:

"Freescale have used the e700 and NG-64 monikers to refer to this core since 2004. [...] PowerPC e700 - The former codename for what probably was to be this product."
http://en.wikipedia.org/wiki/PowerPC_e5500

"The PowerPC e700 or NG-64 (Next Generation 64-bit) [...] was eventually revealed as the e5500 core."
http://en.wikipedia.org/wiki/PowerPC_e700

"The e700 is not based on e600."
http://en.wikipedia.org/wiki/Talk:PowerPC_e700

I was extremely surprised to actually find a Freescale PDF file (dating September 2005) supporting his view:

http://web.archive.org/web/20060623094707/http://www.freescale.com/files/abstract/overview/FTF_BN112.pdf (page 35)

Can it really be that e700 was supposed to be based on e500 (and not on e600) from the very beginning? For instance, I couldn't find one single document listing AltiVec as a planned e700 feature, which is further supporting the Wikipedia editor's view.

Edit: some more documents indicating that e700 was supposed to be a continuation of e500:

From June 2005, few weeks after Apple's switch announcement:
http://www.freescale.com/files/community_files/MCUCOMM/1033_e300_e500_e600_comp.pdf (page 7 indicates e700 to become the basis of future PowerQUICC processors)

From September 2007:
http://www.power.org/devcon/07/Session_Downloads/PADC07_Maguire_Sept25_MPC5121e_Freescale_final.pdf (page 8 lists e700 together with e500 in "Book III-E" column)

[ Edited by Andreas_Wolf on 2011/3/4 13:13 ]

Quote:

I just looked up my older Fresscale documents (from 2005). I found the e700 only in the context of network processors. And while it isn't stated that the e700 would include Altivec the evolution diagram suggests that.
There are two major branches on the diagram, one for the e500 core with several 85xx chips on it. That branch just continues with an arrow to the future.
The second major branch starts with 74xx and branches itself in 2005 into two subarrows, one continuing teh 74xx series and one integrated host processors starting withe the 8641D and directly continues with the 87xx (64 Bit, 2.5 GHz).
*I* would understand that diagram in a way that the e700 would be the continuation of the e600. But it is nowhere explicitly stated in my documents.

Just another addition: When Apple announced their switch, freescale put the e700 quite immediately on hold and during a talk a Freescale guy used Apple's switch as the reason for putting it on hold. The e700 was aimed at Apple, Therefore I akways thought it was just the successor of the e600 with Altivec. Because I think Apple would not have been pleased without Altivec. Plus, somewhere in the back of my mind I think I heard of Altivec being part of the e700 core. But my old, rusty brain may be wrong though...
Maybe after Apple left and Freescale saw teh network communication as their new major target they decided to make the e700 a Altivec-less core which then could better get derived from the e500. Since teh e700 was mostly paper work at that time, I think it is pretty possible, that Freescale just changed the basics of the e700.

> I just looked up my older Fresscale documents (from 2005). I found the e700 only in
> the context of network processors. And while it isn't stated that the e700 would include
> Altivec the evolution diagram suggests that. [...] integrated host processors starting withe
> the 8641D and directly continues with the 87xx (64 Bit, 2.5 GHz).

Is that 2005 document from before or after Apple's switch announcement (June)? (Apdf -> "Ansicht" -> "Informationen" -> "Letzte Aenderung")

Quote:

I think there a serious issues with APDF feeding a *paper* folder Maybe i should send out a bug report...
While the event took place briefly after Apple announced the switch I think the talk (i.e. the ppt slides) was prepared before.

[ Edited by Zylesea on 2010/6/24 23:19 ]

> during a talk a Freescale guy used Apple's switch as the reason for putting it on hold.

Yes, I've known that story from you.

> I akways thought it was just the successor of the e600 with Altivec.

Me too. That's why I'm so astonished by what the Wikipedia articles say in that regard.

> somewhere in the back of my mind I think I heard of Altivec being part of the e700 core.

I thought so too. But maybe you and me were just implying that by assuming it would be derived from e600? Otherwise I'd find it odd that I can't chase down an (official) online document explicitly stating AltiVec as an e700 feature.

> Maybe after Apple left and Freescale saw teh network communication as their new
> major target they decided to make the e700 a Altivec-less core which then could
> better get derived from the e500. Since teh e700 was mostly paper work at that time,
> I think it is pretty possible, that Freescale just changed the basics of the e700.

Interesting thought :-)

> I think there a serious issues with APDF feeding a *paper* folder

Oh, old-school paper document. Nice but less chance to determine the exact age. Pity.

> While the event took place briefly after Apple announced the switch I think the
> talk (i.e. the ppt slides) was prepared before.

Makes sense. That would support your thesis of the e700 having changed its basis from e600 to e500 mid 2005 (i.e. due to Apple's switch to Intel). Reminds me of the "Motorola G5" (MPC7500) project in 2000/2001 where they eventually released only the "Embedded G5" (MPC8540/8560) in 2002/2003. Really disappointing.

http://en.wikipedia.org/wiki/Motorola_G5_project
http://web.archive.org/web/20000816012045/http://www.mot.com/SPS/PowerPC/overview/newroadmap.pdf (September 1999)
http://web.archive.org/web/20030803100646/e-www.motorola.com/files/archives/doc/roadmap/PPCRMAP.pdf (November 2001)
http://www.geek.com/articles/chips/powerpc-roadmap-g5-g6-g7-2001123/

Quote:

Don't worry Ulrich, many in the audience are old and rusty too! I guess we love this debate anyway. But you guys are talking about either abandoned, crippled, or forthcoming products. Don't you think it's a pity not talking about something real?

> you guys are talking about either abandoned, crippled, or forthcoming products.
> Don't you think it's a pity not talking about something real?

The announcement of the e5500 core *is* real ;-) And I find it interesting to discuss if that core is what was once announced as the e700 core because there seem to exist completely opposing answers to that question with good arguments on both sides.

Quote:

Yes, I'm not saying it isn't. What I mean is that, precisely, it's just an announcement. Lucky people using regular computers have the joy of talking about things that they can actually touch.

Quote:

Of course, me too!

> Lucky people using regular computers have the joy of talking about things that
> they can actually touch.

People discuss announced things all the time. Maybe you had heard or read at some places of a device called "iPad" way before it was actually available to buy? ;-)

|||

[ Edited by Velcro_SP 20.04.2011 - 08:02 ]

> it is more proven than these other things being talked about. [... ] it seems
> much more established to me than this other low-wattage, small stuff.

With "other things" you include the PowerQUICC II Pro? If yes, let me tell you that it's massively used in the mid-range embedded network and storage space. In those domains it's a quasi-standard CPU. It can't really be proven more. I think there's magnitudes more devices using PowerQUICC II Pro than devices using MPC5121e (and probably even MobileGT in general). The difference is that really being embedded, in most cases the user simply won't notice that the device is driven by a PowerQUICC II Pro at all.

"PowerQUICC II Pro is used as networking processors for routers, switches, printers, network-attached storage, wireless access points and DSLAMs."
http://en.wikipedia.org/wiki/PowerQUICC#PowerQUICC_II_Pro
(note that MPC837x and thus the 800 MHz capable variants are missing from the article)

> I'd like to know how far performance can be pushed on slow, low-wattage
> processor by [...] dedicated graphics chip, like PowerVR

With the MPC5121e there's no need for a dedicated PowerVR chip because the SoC has a PowerVR core on-chip already. The discussed driver problem for the PowerVR core wouldn't magically be eased if that core was additionally put to a dedicated chip (resulting in two unsupported cores).

Quote:

But it would perform different. I was told that, being the PowerVR core an internal device of the CPU, its usage does block the CPU, not very different to what happens in the original Amiga design, when you ask too much to its custiom chips. The graphics core does steal bandwith to the CPU.

If the graphics core is setup outside the CPU, connected to a PCI bus, these wait states would not cause blocking.

If you want me to explain it better technically, sorry, I can't. But I believe the source that explained it to me years ago.

Then, how come that there are many SoCs with internal graophics cores, and living so happy? I don't know...

|||

[ Edited by Velcro_SP 20.04.2011 - 08:01 ]

> The people on this forum know

Speak for yourself.

> I said and meant absolutely nothing like what you purport to respond to.

I responded to the following words of yours:

(1) "it is more proven than these other things being talked about."
(2) "I'd like to know how far performance can be pushed on slow, low-wattage processor by way of more RAM and dedicated graphics chip, like PowerVR"
(3) "it seems much more established to me than this other low-wattage, small stuff."

So you now say you didn't say or mean anything of that?

> I don't want to respond to these misrepresentations or anything else from you but
> neither should my silence be read as acquiescence.

If you think I misrepresented what you said I can only appreciate your attempt at seeking dialogue.

> You're also using ellipsis incorrectly.

I don't think so.

> It's supposed to be used benignly for clarity

I use it to clarify reference, i.e. what I refer to in particular if the statement I respond to covers several aspects and I don't refer to all these aspects. To illustrate: Regarding your sentence

"I'd like to know how far performance can be pushed on slow, low-wattage processor by way of more RAM and dedicated graphics chip, like PowerVR"

...I only referred to your mention of "dedicated graphics chip, like PowerVR", not to "more RAM". That's why I used ellipsis for "more RAM". And that's a genuine use of ellipsis.

> not in such a way that the person you are quoting has to come
> back and say "that's not what I said.

Then stop failing to clarify what exactly you *meant to say* by the words I quoted from you. I'm not able to *mind*-read, after all.

Did "other things being talked about" and "this other low-wattage, small stuff" include PowerQUICC II Pro, which was talked about? Yes or no? (That's been my simple question from the start.)
What did you mean by "dedicated graphics chip, like PowerVR", if not a dedicated PowerVR (or similar) graphics chip?

> It's not supposed to resolve ambiguities in the quote in a way
> to benefit your response

I didn't use it to resolve any ambiguities but only for clarification as to what I'm referring to in particular.

> Please just don't do it all with my words.

I'll continue to use ellipsis in the same proper way I've been doing it.

Here we go again...

http://drlacxos.files.wordpress.com/2008/04/popcorn_soda.jpg

Quote:

Luke^H^H^H^HAndreas, that's the next step in learning the ways of the Force. You will find that in time, you will be able to "see" and feel other people's minds, like open books. The final step will be mind-writes, but beware of the Dark Side, it will tempt you to force your will upon weak minds much too often...

|||

[ Edited by Velcro_SP 20.04.2011 - 07:43 ]

Quote:

please do continue. pile on what? Unless, that's a phrase I'm not aware of -I'm not a native English speaker anyway. Still, I'm sorry you are not humoured. It was meant as a joke, and I fail to see how my post could even remotely annoy you.

|||

[ Edited by Velcro_SP 20.04.2011 - 07:10 ]

Andreas,
Given that the e5500 has improved fpu performance, what is the impact of the missing Altivec instructions (in particular in regard to MorphOS).
Also, as a 64 bit processor, could MorphOS be ported to the e5500 and how would performance compare to a G5?

[ Edited by Jim on 2010/7/2 2:40 ]

> Given that the e5500 has improved fpu performance

As far as I understand, the e5500 has improved FPU performance towards the e500mc, where the FPU is crippled by running only half-clocked. Thus, the e5500's full-clocked FPU would be no better than what's considered a usual FPU elsewhere (including other PPCs with real FPU).

> what is the impact of the missing Altivec instructions
> (in particular in regard to MorphOS).

That question is best answered by a MorphOS Team member. But to get an idea about usage of AltiVec by MorphOS components:
http://www.google.com/search?q=site:morphos-team.net+altivec
Of course, there is also 3rd party software leveraging AltiVec if available, like MPlayer or dnetc client. But I think there's hardly any MorphOS software that really demands an AltiVec unit to be present. It's just that in case of missing AltiVec generic code is used und therefore the program is running slower.

> as a 64 bit processor, could MorphOS be ported to
> the e5500

I think it could. And it could even use the 32 bit mode of the e5500. So no need to port MorphOS to 64 bit.

> and how would performance compare to a G5?

Good question. According to Freescale the e5500 delivers 3.0 DMIPS/MHz, for the PPC970 I found 2.9 DMIPS/MHz (presumably ignoring AltiVec).

Thanks for the quick concise response. I still can't get data out of Applied Micro on the APM 83290 SoC and 1.5 Ghz does not look that impressive. Freescale has always been pretty open with me (even providing me the MPC8641D samples for my aborted PPC motherboard).
I'm going to check out the e5500. At 2.5 GHz, even without Altivec, it might be a match for the PA6T. And its available from a company that doesn't make it difficult to obtain technical information or components.

Quote:

WHAAAT? Tell us more! Those samples alone are seriuos money themselves...