UltimatePPC rumor

[KimmoK]

Now that sounds interesting, and being x86 it would be cheaper to make. I'm in :-)

I doubt you find usable x86 SoC that is cheaper than PPC.

I bet ultimateppc people have changed P10xx chip to a T10xx. (eur40...50 & 1.4Ghz).

[KimmoK]

Actually, an X86 SoC would be a lot cheaper...

Do you have more info/link at hand?

BayTrail is not as powerfull as desktop x86, but there might be some USD30 component that I have missed. (it has been USD100+ on BOM that I saw)

CherryTrail might be closer to 50% of i7 performance per/Mhz, their initial price was:
Atom x5-8300 PCIe 2.0 x1 (max 1080p OpenGL3.1, max 2GB RAM, USD150...249?)
Atom x5-8500 PCIe 2.0 x2 (max 4k OpenGL3.1, max 8GB RAM, USD250...349?)
Atom x7-8700 PCIe 2.0 x2 (max 4k OpenGL3.1, max 8GB RAM, USD350+?)

So, to me it seems PPC is alot cheaper. (even though might perform only less than what cherrytrail core does)
(PPC prices start from USD30 or so, if cheapest 1Ghz part (without GFX output) is used)

PPC SoC example for ultra low end 1Ghz:

APM82181-SKE1000T USD18,5
http://avnetexpress.avnet.com/store/em/EMController/Processors/AMCC/APM82181-SKE1000T/_/R-5000740400286/A-5000740400286/An-0?action=part&catalogId=500201&langId=-1&storeId=500201&listIndex=-1&page=1&rank=1

https://c1170156.ssl.cf3.rackcdn.com/UK_AMC_APM82181-SKE1000T_DS.pdf


UPDATE:
"The price listed on Z3740 and Z3770 pages on Intel website is respectively $32, and $37."
Still more than cheapest 1Ghz PPC.
But I have no info what is cheapest available intel SoC... and for what real price.
I know that for example some BayTrails (if not all of them) are available only for existing customers. (and not for long)

[KimmoK]

@Iggy
>So...here is a four core AMD Soc that runs at 2.4 GHz AND has a built in GPU. http://www.semiconductorstore.com/cart/pc/viewPrd.asp?idproduct=50134 $97, is that expensive (especially considering the added value of the gpu)?

Thanks for link.
It is expensive if someone is "just" building an accelerator.
But if also GPU planned to be used, then it's ok.
(it just might be super hard to implement driver for it, for Amiga environment)

(PPC SoC with simple 2D output cost about USD50. AMD does 3D.
AMD without GPU is USD38 while PPC is about USD18, so PPC is cheaper but with fraction of AMD performance))

>You should be comparing SoCs, not whole X86 systems with PPC SoCs.

I was.

>At only a little over $100, the T1042 could give us a four core system at up to 1.4GHz
Freescale lists:
-USD38 (+ tax and stuff) for T1024 and their material indicate that they might release also single core T1014.
-USD60 (+ tax and stuff) for T1042

As those chips are pin compatible, one should analyze what is target of the accelerator build.
68k CPU emulator would use only one core, AOS4.1 classic uses only one core etc...

But sure, if there is some use for more cores quad core version should be selected because the price difference is so small vs the total price of such accelerator.


Addition, more about the AMD SoC:
https://www.amd.com/Documents/AMDGSeriesSOCProductBrief.pdf

This seem like superb value for money, if you do not need the GPU (2Ghz 38USD):
http://www.semiconductorstore.com/cart/pc/viewPrd.asp?idproduct=50128

[KimmoK]

@"almost every PPC hardware project ... has ended in failure"

Let's just say "many Amiga related PPC HW projects ended badly", because there are many succesfull projects as well.

And at the same time one should remember that guys like Edison failed some 11 000 times with his lamp experiments before he succeeded.

(+The faster you fail, the faster it is possible to find the working solution.)

@"ppc hardware doesnt make any sense when there is nothing to run on it"

With original design they could only have run some linux/embedded/RT OS.
+ If they were considering it to function as 68k emulated CPU, then it would have run 68k SW.

If they have changed the SoC, then it might be possible to run a lot of SW with it.

Info refresh:
Official from January: "We have restarted our efforts to get a working demo ready before the Amiga30 event in the Netherlands this summer!"
About the CPU from y2012: "We knew that the FPU in this processor is incompatible with the ones found on older PowerPC boards. Our original plan to fix these incompatibilities in software turns out to have more impact that we had foreseen. For this reason, we will reconsider the choice of PowerPC CPU, while continuing the development work of other important stuff on the current prototype."