Update:

> It seems they can't even get their chips above 1.2 GHz.

Now it's official: only up to 1.2 GHz for Mamba.

https://myapmold.apm.com/MyAMCC/jsp/public/productDetail/product_detail.jsp?productID=APM86190
https://myapmold.apm.com/MyAMCC/jsp/public/productDetail/product_detail.jsp?productID=APM86290
https://myapmold.apm.com/MyAMCC/retrieveDocument/SNP/APM86190_PB_20120216.pdf
https://myapmold.apm.com/MyAMCC/retrieveDocument/SNP/APM86290_PB_20120216.pdf

Well Andreas,
Looks like we've got two threads in one here.
I lost interest in the blue vs. red thing a long time ago.
And David is a friend so I'm inclined to side with him and state my desire for increased cross platform cooperation.

Now as to APM, its a pity they can't get anything to scale up.
The successors to Gemini aren't bad, but they're slow.
Even if the T5s only operate at 1.8-2.0 GHz (when introduced), they'll have a solid 50% lead over Applied Micro's best.
It makes picking a favorite manufacturer a no brainer.

> Even if the T5s only operate at 1.8-2.0 GHz (when introduced),
> they'll have a solid 50% lead over Applied Micro's best.

True, and even a much bigger lead if we regard not only the clock frequency but also the per-clock performance of the e6500 core vs. the PPC465 core. This should amount to an overall performance lead per core of at least 150%.

> It makes picking a favorite manufacturer a no brainer.

I wonder what ACube will do for the successor to the 1.15 GHz Sam460ex, if there'll be one that is. Let's see if they'll stay with Applied Micro and opt for the 1.2 GHz Mamba and maybe overclock it like they did with the PPC440EP and PPC460EX chips on the previous Sam boards so they can get it to some 1.4 GHz, or turn to another CPU vendor.

> It's incredibly complex designing any core but it'll be a lot simpler
> than trying to design an x86. It is a frighteningly expensive business:
> [...] Intel and AMD and the likes of Nvidia each spend around $500 million
> to design a new core. - All that before they go into production.

Applied Micro share that estimation:

"to develop a core from scratch would be a $400 million to $500 million type of investment, minimally"
http://www.sec.gov/Archives/edgar/data/711065/000119312512164604/d334600dex992.htm

Update:

> while the e5500 core is claimed to be able to reach up to 2.5 GHz, no such part
> has been announced yet.

They're getting close. QorIQ P5040 and P5021 announced running at up to 2.4 GHz:

http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1692130
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=P5040
http://www.freescale.com/files/graphic/block_diagram/P5040_BD_IMG.jpg
http://www.freescale.com/files/32bit/doc/fact_sheet/P50405021FS.pdf
http://www.freescale.com/files/32bit/doc/reports_presentations/P5040OVRVWPRES.pdf
http://www.freescale.com/files/training/doc/dwf/DWF13_APF_NET_T0560.pdf

SerDes lanes count has been upped to 20 (up from 18 on P5020/P5010) with 2 of them reserved for SATA, but only 3 PCIe controllers (down from 4 on the P5020/P5010) with one of them fixed at x8 speed.
Sampling starting in June 2012, volume production in Q1 2013.

Edit: Price estimation:

"The company has not announced pricing; we estimate $200 to $300 for chips clocked at 1.6GHz to 2.4GHz."
http://www.linleygroup.com/newsletters/newsletter_detail.php?num=4840


Edit: added another PDF link

[ Edited by Andreas_Wolf 26.11.2013 - 00:34 ]

Good Sam, my current Sam to work programming.

The discussion of e5500 chips is interesting, but I am more interested in seeing what is done with the e6500 designs that are reported to scale from single core, low power, low cost SOC products, up to multi-core, multi-threaded workstation class products.

But most people here agree that it is not cost effective to port MorphOS to any new PPC board designs and the only new boards and chips that porting should be considered for should be x86, x64, or ARM, because those three will be mass produced designs intended for computer use, not an embedded market design, or an expensive custom low quantity design.

I have to agree that it is hard to compete with the currently available G4 & G5 Apple models, if MorphOS is going to be ported to any more PPC designs. The 2.7GHz G5 PowerMac is really hard to beat if you are looking at performance per dollar/pound/euro, as the prices continue to come down.

If the MorphOS Development Team will agree to announce the port MorphOS3.1, or 3.2 to the G5 PowerMac's (including the 2.7GHz model and promise to complete the port), I will agree to raise the money to purchase 10 dual 2.3GHz, or faster G5 PowerMac's for the MorphOS Development Team to use, or hand out to which ever members they choose to give them to.

[ Edited by amigadave 07.05.2012 - 14:56 ]

> The discussion of e5500 chips is interesting, but I am more
> interested in seeing what is done with the e6500 designs

Same here, but I found it worth reporting that Freescale seems to be able to eventually (almost) accomplish its design goal regarding the e5500 core's clock frequency as announced 2 years ago. That renders it even more probable that the 2.5 GHz design goal for the e6500 core in the QorIQ T5 will be met as announced a year ago.

> that are reported to scale from single core, low power, low cost
> SOC products, up to multi-core, multi-threaded workstation class
> products.

Actually, within the QorIQ AMP series, the family with the highest per-thread performance, which is the T5, is announced to not be the one with the highest amount of cores (that will be the T4) and to only have single-threaded e6500 cores (while the e6500 in T1 through T4 will be dual-threaded).

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7001&start=665
https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7183&start=250

Well, it will be interesting to see if Trevor Dickinson can learn from his first experience in creating a new computer and make huge improvements in keeping his costs down by avoiding delays and choosing more economical components. Also, will he be able to create a low cost, low power design that will be appealing to hundreds, or thousands of remaining, or returning former Amiga users, to generate enough sales to make a profit from this second effort. I was frankly surprised when he announced at last year's AmiWest Show that he was working on a new design, again with Varisys. I thought that he would back away from designing computers for the tiny Amiga market after the losses he sustained with the X1000 project. Since he has decided to go forward with a second design, I can only conclude that he has seen new information that has given him more hope that his second attempt will be more financially successful than the X1000 has been, as I do not think that Trevor has any hopes that he will ever recoup his investment in the X1000, or if he is even going to try, since he has already stated that he is not making any profit for himself from the X1000 (at least not with the "First Contact" systems). I admire his dedication to the Amiga community and his optimism about the Amiga market. When he does release another product, I hope it will make him a profit and not be a donation to the users who buy it, some of whom can sometimes be ungrateful and very critical without a complete understanding of all that went into the creation of a new computer design.

No matter what his critics say, or write about him and the X1000, I will always support him and his efforts, and wish him all the best in what ever he comes up with in the future. I for one will be listening/reading intently, when his next announcement(s) is/are made.

Good Luck Trevor!

[ Edited by amigadave 08.05.2012 - 11:44 ]

> he announced at last year's AmiWest Show that he was working on a new design,
> again with Varisys. [...] he has decided to go forward with a second design

Actually, he said that A-Eon contracted Varisys to develop no less than 3 new products, one of which will be an XCore processor card.

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7001&start=682

"Varisys to develop no less than 3 new products, one of which will be an XCore processor card."

1) XCore processor card (for PCI)
2) Xorro card that gives easy access to xmos I/O + prototyping area.
3) nemo successor with some Freescale P* or T* chip
4) ?
(I calculated 4 HW things being done from the news feed, but I do not have any details)

There's also XMOS SDK being done for AOS4, but that's not "HW". And netbook/notebook does not seem like A-Eon thing.

I think A-Eon will consentrate on high end systems only.

[ Edited by KimmoK 09.05.2012 - 17:24 ]

KimmoK,
Quote:

Although it is not available for sale to the public yet, I believe that #2 above has already been completed, but might need more work before it is ready to sell. I think they showed it somewhere already.

I don't agree with your last comment about A-Eon only concentrating on high end systems only. If the e6500 core design chips are available as announced in the near future, with a single core SOC, low cost chip design, I could see A-Eon using it to produce an equivalent to an A500 kind of product for the Amiga market. That is what is needed for OS4.x to spread out to more of the majority of remaining and returning former Amiga users. Of course, if Hyperion comes out with their announced netbook at a price of only $300 to $500 and it can run OS4.x fairly well, that will be a huge hit and big seller for people interested in trying out OS4.x for the first time, plus all the existing OS4.x users who want something that is portable and can run off battery power.

In the long run, this could actually be beneficial to MorphOS3.0 on the G4 PowerBook as well, because once hundreds (or hopefully thousands) of current and former Amiga users get a taste of running a Next Gen Amiga system on PPC, many of them will want something more, better, faster, and might try MorphOS3.0 on the G4 PowerBook as a step up from their low powered Hyperion netbooks.

[ Edited by amigadave 09.05.2012 - 10:36 ]

> I calculated 4 HW things being done from the news feed

I didn't "calculate" anything but just listened to what Trevor said himself at AmiWest 2011:

"I'm looking at several new hardware developments with Varisys and we just signed the contracts for three new developments. Two of them are quite exciting, one will just [...unintelligible...] a PCI board with XMOS stuff on it [...]. We're looking at two other areas for new hardware which will hopefully help to bring again a new community together in the future."
https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7183&start=383 (video links therein)

> There's also XMOS SDK being done for AOS4

I wonder if that will ever come, and I still remember Trevor's statements on that from July 2010 in Essen/Germany:

http://www.amiga-news.de/de/news/comments/251521.html

(Working links to video:
http://www.boingsworld.de/video/AmigaEvent2010Essen.ogg
http://www.boingsworld.de/video/AmigaEvent2010Essen.mp4 )

Some statements issued by "the son of the guy that runs XMOS" some weeks before:
http://moobunny.dreamhosters.com/cgi/mbmessage.pl/amiga/173959.shtml
http://moobunny.dreamhosters.com/cgi/mbmessage.pl/amiga/174092.shtml
http://moobunny.dreamhosters.com/cgi/mbmessage.pl/amiga/174154.shtml
http://moobunny.dreamhosters.com/cgi/mbmessage.pl/amiga/174180.shtml

> netbook/notebook does not seem like A-Eon thing.

True, it's been announced as a Hyperion thing. But I suspect that sales are supposed to be handled by AmigaKit.

> I think they showed it somewhere already.

Yes, 2 months ago there: http://www.youtube.com/watch?v=S5LLqe62_J8

Update:

> I wonder what ACube will do for the successor to the 1.15 GHz Sam460ex, if there'll
> be one that is. Let's see if they'll stay with Applied Micro and opt for the 1.2 GHz Mamba
> and maybe overclock it like they did with the PPC440EP and PPC460EX chips on the
> previous Sam boards so they can get it to some 1.4 GHz, or turn to another CPU vendor.

No overclocking of 1.2 GHz Mamba to 1.4 GHz needed as APM86691 and APM86692 ("Black Mamba") got announced running at up to 1.4 GHz:

http://investor.apm.com/phoenix.zhtml?c=78121&p=irol-newsArticle&ID=1702760
http://www.apm.com/products/embedded/multicore460/apm86692/

So for now the PacketPro family of chips has those members:

APM86190 (Green Mamba, single-core)
APM86290 (Green Mamba, dual-core)
APM8639x (Diamondback, single-core and dual-core)
APM86491 (Catalina, single-core)
APM8669x (Black Mamba, single-core and dual-core)
APM86791 (Keelback, single-core)

I assume that there's an APM8659x still to come ;-)

Correction:

> the e6500 in T1 through T4 will be dual-threaded

First T1 chip got announced:

http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1705881
http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1743554
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=T1042
http://www.freescale.com/files/32bit/doc/fact_sheet/AMPT1042PFS.pdf
http://www.freescale.com/files/32bit/doc/fact_sheet/T1FAMILYFS.pdf
http://www.freescale.com/files/training/doc/dwf/DWF13_QorIQ_Portfolio_SanJose.pdf (SerDes lane and PCIe controller configs at pages 36/37 and 39)
http://2014ftf.ccidnet.com/pdf/0163.pdf (SerDes lane and PCIe controller configs at pages 21-23)
http://www.freescale.com/files/graphic/block_diagram/T1040_T1042_BD_IMG.jpg
http://www.youtube.com/watch?v=V6V3cKlZW0o
http://www.youtube.com/watch?v=hyU3VUT2Vpc

...and quite to my surprise (and opposed to what is being said by Freescale on other occassions*), it doesn't have the new e6500 cores but the older e5500 cores, thus neither threading nor AltiVec.
I think it's notable that those are e5500 cores shrunk to 28 nm (like e6500), as opposed to the 45 nm e5500 cores of the P5 series.
Furthermore, the T1042 has a DDR4 memory controller (T2080 and T4240/T4160 have only DDR3).

* http://www.freescale.com/files/training_pdf/FTF/2012/americas/WBNR_FTF12_IND_F0122.pdf (page 9)
* http://www.freescale.com/files/training_pdf/WBNR_FTF11_NET_F1176.pdf (page 12)
* http://www.freescale.com/files/32bit/doc/brochure/PWRARBYNDBITSSKD.pdf (page 3)
* http://www.freescale.com/files/32bit/doc/brochure/PWRARBYNDBITS.pdf (page 57)
* http://www.freescale.com/files/microcontrollers/doc/brochure/BRPWRARCHCOR.pdf (page 3)

T1022 seems to be next (maybe it has a DIU like the P1022):
http://www.freescale.com/files/training_pdf/FTF/2012/americas/WBNR_FTF12_NET_F0009.pdf (page 59)
http://2012ftf.ccidnet.com/pdf/0009.pdf (page 52)
http://www.freescale.com.cn/cstory/ftf/2012/pdf/0009.pdf (page 52)

Edit1: Added Youtube video link.
Edit2: Added another PDF link.
Edit3: Added T1040.
Edit4: Added another PDF link.
Edit5: Added another PDF link.

[ Edited by Andreas_Wolf 29.10.2014 - 14:04 ]

Update:

> Revised QorIQ T roadmap as of August 2011 (page 4):
> http://www.freescale.com.cn/cstory/ftf/2011/pdf/0772.pdf
> http://2011ftf.ccidnet.com/pdf/0772.pdf
> [...]
> T2: T2080 [...]

Announced:

http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1705881
http://media.freescale.com/phoenix.zhtml?c=196520&p=irol-newsArticle&ID=1743554
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=T2080
http://www.freescale.com/files/32bit/doc/fact_sheet/AMPT2080PFS.pdf
http://www.freescale.com/files/32bit/doc/fact_sheet/T2080FS.pdf
http://www.freescale.com/files/graphic/block_diagram/T2080_BD_IMG.jpg
http://www.freescale.com/files/graphic/block_diagram/T2081_BD_IMG.jpg

T2080: 2x PCIe 3.0, 2x PCIe 2.0, 16 SerDes lanes (8x 10 GHz, 8x 8 GHz), 2x SATA2
T2081: 1x PCIe 3.0, 3x PCIe 2.0, 8 SerDes lanes (8 GHz), no SATA

"T2080 is primarily intended to succeed [...] P2041 and P3041. T2080 offers double the performance of these devices while maintaining about the same power and pricing."
http://www.youtube.com/watch?v=_qOP9muZAQ8

"T2080 is primarily intended to succeed [...] P2041 and P3041. T2080 [...] offers 2x or better in core capability, cache size, SerDes bandwidth and Ethernet connectivity of these devices while maintaining about the same power and pricing."
http://www.youtube.com/watch?v=JRhAq1XHcBo

Pricing of P2041/P3041 at Avnet:
P2041: 202...337 USD
P3041: 230...364 USD


Edit: added T2081

[ Edited by Andreas_Wolf 10.10.2012 - 22:13 ]

Addendum:

>> some (see Wikipedia) even think the e5500 to actually be what was
>> announced as e700 in 2004.

> Answer from Preet Virk, strategic marketing director for Freescale's
> Networking Processor Division:
> "The e5500 is [...] unrelated to a previously mentioned e700 family"
> http://blogs.freescale.com/2010/06/23/what%e2%80%99s-up-with-64-bit-embedded-computing/#comment-292

Wikipedia has been changed recently to reflect this. The statements

"Freescale have used the e700 and NG-64 monikers to refer to this core since 2004."
and
"The former codename for what probably was to be this product."
and
"It was eventually revealed as the e5500 core."

...have been changed to

"Freescale have used the e700 and NG-64 monikers to refer to a very similarly speced core since 2004, but they are not the same product"
and
"Freescale eventually released the e5500 core but it was not the same core even though they are very similarly speced."

http://en.wikipedia.org/wiki/PowerPC_e5500
http://en.wikipedia.org/wiki/PowerPC_e700

Update:

>> I have not read up on which PPC platforms AROS has been ported to

> So far, AROS for PPC runs natively only on Efika 5200B and Sam440.

Seems it's being worked on for Sam460ex as well:

http://gitorious.org/aros/aros/commits/sam460
http://aros-exec.org/modules/newbb/viewtopic.php?post_id=72012#forumpost72012
http://aros-exec.org/modules/newbb/viewtopic.php?post_id=72034#forumpost72034


Edit: Added links to aros-exec.org postings.

[ Edited by Andreas_Wolf 01.08.2012 - 18:17 ]

What's the fastest T1?
What core is to used in T2-T4?

> What's the fastest T1?

The only one announced so far:

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7001&start=736

> What core is to used in T2-T4?

"e6500 core running up to 2.5 GHz (T2-T5 families)"
http://www.freescale.com/webapp/sps/site/overview.jsp?code=QORIQ_AMP_SERIES

Update:

> There's also another error in there, as it reads "P3080" where it should read "P3060"

P3060 cancelled:

https://lists.ozlabs.org/pipermail/linuxppc-dev/2012-July/099047.html
http://lists.denx.de/pipermail/u-boot/2012-July/127806.html

Update:

>>> Just for the purpose of overview I compiled a list of the DMIPS/MHz per core
>>> figures for various PPCs in ascending order (AltiVec capability is generally
>>> ignored, obviously):
>>> [...]
>>> e5500: 3.0

>> The table on page #2 of the QorIQ AMP fact sheet PDF indicates that
>> the e6500 core didn't improve in that regard (comparing e6500's per-thread
>> performance to e5500's per-core performance).

> The QorIQ AMP T4240/T4160 fact sheet claims "6.0 DMIPS/MHz per core"
> and at the same time for dual threads "1.7 times the performance of a single thread".
> To me this sounds strange as I think it should be [...] about 5 DMIPS per MHz and core
> at best in case of dual-threaded execution.

According to page 11 of

http://www.freescale.com/files/training_pdf/FTF/2012/americas/WBNR_FTF12_NET_F0027.pdf
http://www.freescale.co.jp/event/ftfj/2012/technicalsession/pdf/FTF12_NET_F0027_Don_rev2-180.pdf
http://www.freescale.com/files/training/doc/dwf/DWF13_AMF_NET_T0027_Ottawa.pdf
http://www.power.org/wp-content/uploads/2013/11/QorIQ-T4xxx-Slides.pdf

it's indeed 6.0 (or 5.4, or 7?) DMIPS/MHz per e6500 core, but 3.3 instead of 3.0 (or 2.7?) DMIPS/MHz per thread, which amounts to a factor of 1.8 (or 1.6, or 2.1?) instead of 1.7.
Taking these new information into account, this is the revised list of the DMIPS/MHz per core figures for various PPCs:

e300: 1.9...2.0 (depending on version)
PPC440: 2.0
PPC460: 2.0
PA6T: 2.2
PPC750: 2.3
e600: 2.3
e500mc: 2.5 (or 2.7?)
PPC470: 2.7
PPC970: 2.9
e5500: 3.0 (or 3.3 (32-bit) and 3.4 (64-bit)?)
e6500 (1 thread): 3.3 (or 3.4 (32-bit) and 3.5 (64-bit)?)
e6500 (2 threads): 6.0 (or 5.4, or 7?)


Edit: added another PDF link
Edit2: added alternative value for e6500 (2 threads)
Edit3: added alternative values for e500mc, e5500 and e6500 (1 thread) from https://www.nxp.com/files-static/training/doc/ftf/2014/FTF-NET-F0139.pdf (page 3/4)
Edit4: another source for alternative value for e6500 (1 thread)
Edit5: added alternative value for e6500 (2 threads)
Edit6: https://www.nxp.com/docs/en/supporting-information/DN-Digital%20Networking%20solution-2017.pdf (page 17) confirms "3.4 (64-bit)" value for e5500

[ Edited by Andreas_Wolf 17.05.2023 - 14:34 ]

>> bPlan are very close to finishing Aura development on the EfikaMX, thus providing
>> an high level abstraction to the devices/drivers of the device in an OS-agnostic way.
>> Of course, an appropriate Linux kernel is also developed at the same time.

> Thanks for the clarification Konstantinos. So, what Genesi/bPlan/Pegatron are doing
> is making a new computer and custom operating system. Well, almost... But in many
> regards, it is. Once done, it's all a matter of convincing the world to write kernels for Aura.

According to Genesi's statement from June 2011, they are more successful than IBM in this regard at least:

"Genesi also possess the proprietary technology, "Aura." Competitors like IBM have worked diligently and spent substantial amounts of time and capital to replicate Aura. However, until this date, they have not been able to succeed in that endeavor."
http://www.genesi-tech.com/cache/baylor/China.pdf (page 20)

Does anybody have any information about IBM's alleged unsuccessful endeavour to "replicate Aura"?

Update:

>> when I discussed PPC based systems with the owner of Varisys
>> (the designers of the X1000 motherboard) I got this response:
>> "In terms of PPC we are thinking QorIQ here for higher end designs."

> Apparently, they've started to follow suit:
> "XMC/PMC [...] VM400 (XMC) QorIQ Pxxxx"
> "COM Express [...] VPX1 QorIQ P1022"
> "PC104 [...] MC104PRQ QorIQ P2022"
> http://www.varisys.co.uk/products.html

The VM400 has now been revealed as:

"XMC/PMC VM400 (XMC) QorlQ P3041/5020"
https://www.box.com/index.php?rm=box_download_shared_file&file_id=f_3209347735&shared_name=si8iwtiq2ag3oyykkqvb (page 9)

PDF file seems to be 4 months old but was apparently publicly released only now.