That's odd Andreas as I can see multiple applications for it. A significant part of the encoding and decoding of A/V and audio data involves compression and decompression.
Until we have the full specs on how this components works, it would be foolish to dismiss it as not having broader applications.

> I can see multiple applications for it.

Me too, but none of those involve A/V decoding or encoding.

> A significant part of the encoding and decoding of A/V and
> audio data involves compression and decompression.

True, but which specific A/V codecs make use of any of the algorithms supported by the DCE?

> Until we have the full specs on how this components works, it
> would be foolish to dismiss it as not having broader applications.

You mean the list of DCE-supported algorithms presented in the documents that have been published so far may not be complete and the DCE may support more algorithms, including such that are useful for A/V encoding or decoding?

Yes, Andreas,
I saying that until we have a full understanding of all that this unit can do, we can not count out the possibility that it could aid in conventional encoding and decoding applications.
What interests me more is the possibility that new methods, better suited to this component, might be created.






[

>> You mean the list of DCE-supported algorithms presented in the documents
>> that have been published so far may not be complete and the DCE may support
>> more algorithms, including such that are useful for A/V encoding or decoding?

> Yes

So you think that Freescale conceals what the DCE really can do at this point in time when you can already purchase the hardware?

> until we have a full understanding of all that this unit can do, we
> can not count out the possibility that it could aid in conventional
> encoding and decoding applications.

I don't understand. What you mean "conventional encoding and decoding applications"? Encoding/decoding means using a specific (set of) algorithm(s) to transform a bitstream into another bitstream under certain rules. In case of the DCE, algorithms mentioned specifically by Freescale have been implemented in hardware to accelerate the transformation. What's "conventional" about specific algorithms?

> What interests me more is the possibility that new methods,
> better suited to this component, might be created.

Huh? You mean that somebody should invent A/V codecs that use algorithms that the DCE implements, establish those new codecs as a standard and make the world use those codecs, so that subsequently the DCE could be used for decoding/encoding A/V data? Sounds like a plan ...not ;-) Thing is that the algorithms implemented by the DCE may not be suited for encoding of A/V data. Refer to:

http://en.wikipedia.org/wiki/DEFLATE
http://en.wikipedia.org/wiki/Gzip
http://en.wikipedia.org/wiki/Zlib
http://en.wikipedia.org/wiki/Base64

As you can see, the first 3 are essentially just applications of LZ77 and Huffman. All of them, i.e. LZ77, Huffman and Base64, are *lossless* encoding algorithms. Thus, I can only repeat that I can't see where A/V encoding or decoding could fit here.

Addendum:

> The T4240RDB has already been announced as "Low Cost 1U development platform",
> (to be) developed by NEXCOM.

NEXCOM has introduced its first T4240-based hardware but still no sign of the T4240RDB.

>> Since at least one month ago, the PowerQUICC and QorIQ Processor Selector Guide
>> (last page) lists the P5040RDB (reference design board) at 1995 USD.

> That is a much better price. Thanks for keeping track Andreas. I hadn't thought
> to revisit that.

The above-linked document still has the same price of 1995 USD for the P5040RDB, but now that the product page and fact sheet are available, the website shows a price of 2995 USD, which is even further afar from "low cost" than the other price. Strange.

http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=P5040RDB&tab=Buy_Parametric_Tab

Update:

> NEXCOM has introduced its first T4240-based hardware but still no sign of the T4240RDB.

"T4240RDB: 1U, available Q3-13, Nexcom"
http://www.youtube.com/watch?v=U3cSmYV1mJ8

In the same video, the narrator says about the T4240RDB in comparison to the T4240QDS:

"We are working with partner Nexcom in development of a 1U reference design that brings out most of the same functionality in a much smaller form factor at less than half the cost, which will be available in Q3-13."

This means the "low cost" T4240RDB will be below 2000 USD.

>> You mean the list of DCE-supported algorithms presented in the documents
>> that have been published so far may not be complete and the DCE may support
>> more algorithms, including such that are useful for A/V encoding or decoding?

> Yes

http://www.freescale.com/files/32bit/doc/prod_brief/T4240PB.pdf (pages 28/29) has more info on the DCE from which you can see that this is not the case.

Definitely not useful in A/V, but still interesting as a way of increasing network bandwidth.
Thanks for the documentation.
I started researching the compression/decompression formats supported.
Fascinating.

Update:

> This means the "low cost" T4240RDB will be below 2000 USD.

T4240RDB has been made available:

http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=T4240RDB
http://www.freescale.com/files/graphic/block_diagram/T4240RDB_BD.jpg
http://www.freescale.com/files/32bit/doc/fact_sheet/T4240RDBFS.pdf
http://www.freescale.com/files/32bit/doc/user_guide/T4240RDBQS.pdf

And indeed it is only 1445 USD.

Interesting info from the fact sheet (page 2):
"This board is also available from Nexcom in production volumes."


Edit: added quote

[ Edited by Andreas_Wolf 01.12.2013 - 15:10 ]

That price is at least way cheaper than some certain custom ppc board. Two PCIe ports is little, but at least enough to add a gfx card and an audio card. The board is also availavble in volume.

Still expensive, but maybe worth a closer look.

Quote:

I recently bought a new PC to replace my old. A quad core fourth generation Core i7 4770 (Haswell) CPU with 8MB cache, and a motherboard based on the Z87 chipset with plenty of PCI-e and very nice specs and features, costed me equivalent of $430 USD (excl tax). Not very cheap for a motherboard/CPU combo today, I know, but it's a major step up from my old first generation Core i7 860 "el-cheapo" CPU, and I really need that. The full system (motherboard, CPU, case, PSU, 8GB RAM, Noctua CPU cooler, Radeon R9 270 GFX card, and 256GB Samsung SSD 840 PRO) landed at equivalent of $1360 USD (excl tax).

When the Pegasos 2 G4 was introduced a decade ago, I think I recall its introduction price was €499 EUR (~$680 USD in todays rate) excl tax, but I can be wrong about that? I do recall that I thought it was kind of expensive though for a motherboard/CPU combo, especially by the value of the money back then, but it was a small volume product for a narrow market, so high prices are to be expected, right? After a while the price came down a bit though (as Genesi products usually did after a while), in at least 2 price drops IIRC. And when the Pegasos 2 was finally discontinued, the remaining stock was sold out at $399 USD.

Evaluation motherboards has always been extremely high priced from a consumers perspective. But these kind of motherboards aren't meant for end-users anyway, so it doesn't matter. But if we think that $1445 is cheap, then maybe the community has become "speed blind" regarding prices because of the insane prices of "certain custom ppc boards"? And is a "24-virtual-core" CPU really that relevant in a MorphOS context anyway?

takemehomegrandma,
Quote:

Yes you can !!

499€ was including VAT (299€ for the G3).

Which was indeed more than one had to spend on a similar speced (low end) x86 mobo, but also less than anything G4 Apple offered at that time (but then those were all full systems clocked at > 1GHz).

> if we think that $1445 is cheap

What was said in this thread is that it's cheap compared to

- "some certain custom ppc board" (referring to Nemo, I reckon),
- the "below 2000 USD" mark (which could as well have meant 1999 USD),
- other Freescale Power Architecture RDBs, especially considering the specs.

Anything else would be a product of your imagination.

> is a "24-virtual-core" CPU really that relevant in a MorphOS context anyway?

Being Power Architecture, at least more relevant than your "quad core fourth generation Core i7 4770 (Haswell) CPU" :-) And yes, this thread was certainly started in the wrong subforum as MorphOS never ran on the MPC8610, so should belong into General Discussion instead. But as this thread began without a MorphOS context, I think it's legitimate to continue it this way. You may request this thread be locked, though :-P

Since I no longer expect content about the 8610 (which is obsolete anyway) to be posted here, I would prefer that this remain (a I am following it).
Andreas' new about the T4240RDB is cetainly welcome as other PPC based reference boards have cost over twice as much.

It is certainly possible to modify the subject of a discussion thread to better reflect its growing content, whenever this may be necessary.

takemehomegrandma wrote:


Quote:

They also don't generally work very well and have terrible firmware. Just ask AmigaOne owners...

Quote:

Cheap? You can get *8* Arndale Octas for that price.

> They also don't generally work very well [...]. Just ask AmigaOne owners...

You may want to limit that statement to Eyetech AmigaOnes as I don't think it applies to the A-Eon AmigaOne X1000 or the ACube AmigaOne 500.

Quote:

I am all for an ISA switch, no need to tell me about the x64 beneffits... But this board is comparatively way better positioned than most other _ppc_ stuff and especially Nemo and its annouinced successor.
And regarding being it a dev board - indeed it is one, but it is also availbale in volume and IIUC the basis for some servers. Just follow the links in the freescale document.

Plus, if th edev board is "that cheap" it may be an indicator that (hypothetical) derived products will have a lower price tag, too.

I am not stopping to evangelize for the ISA switch though. But currently MorphOS is PPC only and hence news of the ppc world are of some importance in the context of MorphOS.

Well, I am in no hurry to get to the ISA change, as I like the OS in its current form and I am concerned about what direction a 64bit version would take.
Further, i do not know about the rest of you, but I (and I am sure Andreas) have been following the developments of products based on the e5500 and e6500 cores since they were announced.
in fact, these two have been about the only bright spot in this ISA.
They were announced, they can to market, and they are available.
This impresses me as others building on this ISA have not fared as well.

What good would a 12 core hyper threaded 64 bit PPC be to us?
Not that much. Its not useful with our current OS.
But the curious thing is, even without an ISA change, it would definitely be suited to where we want to go.

And, thanks to the fact that this is the only reasonably affordable reference design I have seen Freescale offer (at least since they offered the HPCN-MPC8641D at half price), I may buy one to use under Linux.

> it is [...] IIUC the basis for some servers. Just follow the links in the freescale document.

In case you refer to the Nexcom NSA 5640, it seems to have a very similar board as the T4240RDB, but not exactly the same. For instance, the NSA 5640 seems to have a Mini PCIe x1 slot where the T4240RDB has a normal PCIe x4 slot.

>> T4240QDS has been made available

> I'd love to see how well the dedicated hardware could handle [...] video codecs.

Just stumbled upon this from two years ago:

"We also did some x264 performance benchmarks on a current Freescale/NXP T4240 Power CPU (24 e6500 cores available) with Altivec support enabled. Up to a parallelization of 8 to 10 threads you get some performance benefits with each additional thread on this platform. Beyond that parallelization grade you see quite significant performance saturation effects - presumably due to Altivec resource and CoreNet bandwidth limitations."
https://mailman.videolan.org/pipermail/x264-devel/2016-January/011583.html

That should read 12 cores available or 24 concurrent threads (although Freescale/NXP literature is equally misleading referring to "24 virtual cores"), but that post is interesting.
Thanks.
I'm trying to find a vendor with a NSB 5640 right now.