> The Freescale Roadmap shown in that first document is not encouraging
> for future PPC development.

Scanning NXP FTF 2016 documents for Power Architecture relevant news reveals:

1. mention of e6500-based B5T10 SoC:
https://community.nxp.com/servlet/JiveServlet/downloadBody/331519-102-1-267267/DES-N1841%20Porting%20Software%20from%20Power%20Architecture%20to%20ARM.pdf (page 7/8)

Also mentioned there (B5T10/B5TXX):
https://il.linkedin.com/in/nir-baruch-30b4421a

Unfortunately, it has been cancelled according to this:
https://il.linkedin.com/in/shai-tzur-8a3772111

2. Power-based QorIQ roadmap vs. ARM-based QorIQ roadmap:
https://community.nxp.com/servlet/JiveServlet/downloadBody/331653-102-1-267403/NET-N1863%20What%27s%20Next%20for%20QorIQ%20-%20Portfolio%20and%20Roadmap.pdf (page 11/12, page 13/14)
https://community.nxp.com/servlet/JiveServlet/downloadBody/331747-102-1-267497/FTF-NET-N1863%20What%27s%20Next%20for%20QorIQ%20-%20Portfolio%20and%20Roadmap.pdf (page 11/12, page 13/14)
https://community.nxp.com/servlet/JiveServlet/downloadBody/333451-102-1-270712/EUF-NET-T2471.pdf (page 8/9, page 9/10)

ARM-based roadmap contains next-gen SoC while Power-based roadmap doesn't. From this can be concluded that from NXP there will not only be no successor to the e6500/e6501 core (which was already evident before) but also no new SoC based on e6500/e6501 core.

As recently as 2 months ago, NXP's QorIQ roadmap still contained future LT1/LS2/LS3 Power Architecture SoCs for 1st half of 2018:

https://community.nxp.com/servlet/JiveServlet/downloadBody/330251-102-2-263914/EUF-DES-T1741%20Scalable%20Multicore%20QorIQ%20Layerscape%20Processors%20Based%20on%2064-bit%20Software%20Environment%20for%20Enterprise,%20Home%20and%20Industrial%20Applications.pdf (page 8/9)
https://community.nxp.com/servlet/JiveServlet/downloadBody/330317-102-1-263920/EUF-NET-T1744%20Powerful%20QorIQ%20Multicore%20Processors%20for%20the%20New%20Virtualized%20Network.pdf (page 8)

Strangely, there're Freescale/NXP documents claiming the existence of a T2040:

http://www.nxp.com/files/FTF_2015/FTF-2015-QorIQ-LS-Slides.pdf (June 2015, page 7/8)
https://community.nxp.com/servlet/JiveServlet/downloadBody/330315-102-1-263918/EUF-NET-T1742%20An%20Overview%20on%20QorIQ%20Trust%20Features%20for%20Securing%20Embedded%20Systems.pdf (March 2016, page 11/12)
https://community.nxp.com/servlet/JiveServlet/downloadBody/330317-102-1-263920/EUF-NET-T1744%20Powerful%20QorIQ%20Multicore%20Processors%20for%20the%20New%20Virtualized%20Network.pdf (March 2016, page 8)


Edit: added more links

[ Edited by Andreas_Wolf 30.11.2016 - 23:08 ]

Quote:

The PPC future looks very bright.

Quote:

Actually, that was a pretty good run at the end when you consider how many generations of Power have been created since Freescale was involved in creating cpus for desktop use.
I, for one, am not crowing at the idea of returning (once again) to Intel.
It isn't as painful as the last time, but there is still the lingering worry that if Intel's marketshare becomes too high that their creativity will stagnate.

I hope we don't see another fiasco like the P4 in the future.

@ Jim

Quote:

Come on, when Apple pulled the plug was when everything interesting stopped regarding PPC for our context. Essentially, it has merely been a long line of SoC's for printers and routers since then, more or less high end, but the same nevertheless, while true desktop developed in a different path. Since Apple moved away, PPC has been a very bad architecture for a desktop OS to have its future tied to...



Quote:

Why should anyone care about CPU, as long as it's powerful enough, available enough and cheap enough?

Quote:

There are illuminating examples in this very thread how the ARM architecture is catching up and starting to chase Intel with a blow torch in its ass in various market segments, with a trend that extrapolated suggests that this will become even more true in the future.

Quote:

OMG, did you just put more emphasis on the P4 than a forgotten parenthesis in CPU history?

Sorry Grandma,
We have a multitude of different opinions here.
First, the best PPCs we support only use Power PC elements from up to Power4.
IBM is promoting Power8 as an open standard, and is readying Power9.
Both of these simply lay waste to any ARM processor.
So dead? Hardly.
I would love to see a Power8 MorphOS platform.

And as to Intel's occassional misstep, the Pentium 4 was hardly the first one.
Anyone remember the Pentium Pro?

Ever since AMD introduced RISC cores to the X86 I have been very fond of them, and they did invent the X64 instruction set.
Further, their built in gpus continue to surpass Intel's in performance.
It will be interesting to see how their next core design fairs.

All that being said, you are right about one thing, if its fast enough who cares what ISA is featured?
Although, for my part, the P50XX is fast enough.

[ Edited by Jim 31.05.2016 - 00:38 ]

@Jim

I'll take a Power4,5,6,7,8 or 9 over an ARM IF they manage to produce a low/mid power chip.

Something suitable for a laptop (even if that never sees the light) or a small footprint desktop (miniITX,MacMini that kind of style). Sure a proper desktop would also be nice but not another Power4 server-CPU shoehorned into a 2ton tower (aka PowerMac_G5).

> I'll take a Power4 [...] IF they manage to produce a low/mid power chip. [...]
> not another Power4 server-CPU shoehorned into a 2ton tower (aka PowerMac_G5).

Well, there's always the iMac G5 for a start :-)

Which is anything BUT low power !

Quote:

Yeah, a high power chip stuck in an all-in-one design with gpus that have a fairly high failure rate.

Low power Power8...
I wonder if that is possible considering how massive the design is?

A quick note - The ARM Cortex A73 looks interesting.
By next year these should feature a new 16 core gpu that looks like a nice step forward as well.

[ Edited by Jim 31.05.2016 - 14:48 ]

> The ARM Cortex A73 looks interesting.

Indeed. Thanks for the note. I missed the announcement.

> a new 16 core gpu

It's up to 32 cores in the Mali-G71.

>Its up to 32 cores...

I missed that, I only saw an announcement that it would be up to 16 cores by the end of the year. 32? Great.

Hmm, I seem to remember say I was in favor of a move to ARM instead of X64 (at least its still RISC).
But then again I'd rather stick eith PPV for awhile as well.

X64, hey...
Since they have decided.

But man, look at the negligible power draw of this latest 64bit ARM.

>> http://www.cavium.com/newsevents-Cavium-Announces-Availability-of-ThunderX.html

> Wow! :-o

http://www.cavium.com/newsevents-Cavium-Announces-ThunderX2.html
http://www.cavium.com/ThunderX2_ARM_Processors.html

The only concern I have with an ARM switch if it were to happen is the lack of full featured hardware. You run into the same problem OS4.1 is having with custom high priced boards.

@Acill

I don't think the Team would make such a mistake. The point of the switch is to have access to cheap, easy to find, of the shelf hardware.

Quote:

I would hope so. Looking more into it the only ARM boards I see are the small mini systems and some servers. The servers are not low cost for sure.

Quote:

There are all sorts of lower cost ARM systems.
You just have to look around to find them.
The newer chips are usually only found on development boards or devices meant for hobbyist.

The problem as far as I understand it that it's usually one vendor that makes a certain ARM based board (usually RaPi clones), and once they are dropped from the market the Team would have to start over again.

Raspberry Pi clones?
No, even if the team targeted ARM that would be a poor choice.
The original, while it has a decent GPU, has a weak cpu.
The latest Pi does partially remedy that, but its still not an impressive device.

As we have repeatedly covered available ARM boards, I'm not going to rehash that topic.
But even my two year old A9 quad core Android box would be a better choice.

Raspberry Pi clones?
No, even if the team targeted ARM that would be a poor choice.
The original, while it has a decent GPU, has a weak cpu.
The latest Pi does partially remedy that, but its still not an impressive device.

As we have repeatedly covered available ARM boards, I'm not going to rehash that topic.
But even my two year old A9 quad core Android box would be a better choice.

Update:

> New list of DMIPS per MHz and core figures

ARM Cortex-A7: 1.9
ARM Cortex-A8: 2.0
ARM Cortex-A35: 1.8...2.1 (depending on source)
Qualcomm Scorpion: 2.1
ARM Cortex-A53: 2.3
Marvell Sheeva PJ4: 2.4
ARM Cortex-A9: 2.5
Marvell Sheeva PJ4B: 2.6
Qualcomm Krait 200: 3.1...3.3 (depending on source)
Qualcomm Krait: 3.3
Qualcomm Krait 300: 3.4
ARM Cortex-A12: 3.0...3.5 (depending on source)
Broadcom Brahma15: 3.5 (unsure whether or not this core is just ARM Cortex-A15)
Apple Swift: 3.5
Qualcomm Krait 450: 3.5
Qualcomm Krait 400: 3.4...3.6 (depending on source)
ARM Cortex-A17: 2.8...4.0 (depending on source)
ARM Cortex-A15: 3.5...4.0 (depending on implementation)
Applied Micro Potenza: 4.2
ARM Cortex-A57: 4.1...4.8 (depending on implementation)
Samsung M1/Mongoose: 5.1
Qualcomm Kryo: 6.3
ARM Cortex-A72: 4.8...7.4 (depending on source)
ARM Cortex-A73: 4.9...8.5 (depending on source)

Is this per core or for all cores ? Ex a7 monocore vs exynos whith for cores ?

>>> New list of DMIPS per MHz and core figures

>> [...]

> Is this per core or for all cores ?

As I wrote: "per MHz and core"

There's been quite a bit of news recently, surprised it's not been mentioned.

Intel to make ARM-based smartphone chips.

Fujitsu to build ARM v8 Supercomputer.

This will require a new technology, funnily enough this appeared:

The Scalable Vector Extension (SVE) for the ARMv8-A architecture.

Oh, and there's this.

Quote:

A lot of interesting news there, like Intel going ARM instead of Atom!

I said early on in this thread that ARM will scale upwards much easier than x86 will scale downwards. ARM is entering new market segments upwards all the time, wile continuing to rule the low power segments (and soon dawns IOT for real)...

> Intel going ARM instead of Atom!

That's a misleading statement in my opinion. Atom has been Intel's property, whereas for ARM chips they will only act as make-to-order manufacturer (just like TSMC and GlobalFoundries).