@feanor

I sent an email to you. I would love to see a new board based on the 8610 processor, something like the red tail with more pci-e and pci slots. What would the chances be of considering an VIA Envy 24 for audio? The sound quality is excellent and its the same chip used in M-Audio Revolution cards like the one in my PowerPC Mac and should be supported in Linux. Also have you considerded posting in Macintosh forums as many of them have experience with developing for PowerPC? Sure you can buy Macs with faster PowerPC processors fairly cheap now but mine is already 8 years old and I have no intenetion on buying an Intel Mac. Also have you considered talking with Fixstars, maybe they can help.

@Sloar

Quote:

Also have you considered talking with Fixstars, maybe they can help.

I think he covered that.
@feanor has threads in all major forums, but the one that talks about Fixstar is here

Also a good link page here to help you see who he's talking to:

http://lists.ozlabs.org/pipermail/linuxppc-dev/2009-September/076132.html

#6

Last edited by number6 on 05-Oct-2009 at 12:08 AM.

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@minator

Dear Nicholas,
Do you even read other posts or you just assume its content? Exactly where do you see me state a specific ratio between the CPUs? I explicitly say a _minimum_, which is calculated from a set of benchmarks I've done on my own machines- then scaled from 2.4ghz to 1.6ghz- the atom is more than 3 times slower on all tests, _after_ frequency scaling.

---

And theoretically btw...

Intel Core 2 can complete up to 4 instructions per cycle (per core).
Intel Atom can complete up to 2 instructions per cycle (per core).

Optimally the Atom would be 50% slower per cycle, but due to lack of instruction reordering it's not even particularly likely to complete 2 instructions per cycle successfully..

Last edited by Frek on 05-Oct-2009 at 12:30 AM.

@Frek

This isn' a thread about Core 2 vs Core i7, have a look here however:

http://www.tomshardware.com/reviews/Intel-Core-i7-Nehalem,2057-17.html

I don't know what benchmarks you ran, but I guess I could write a custom benchmark of my own that made the i7 look ridiculous against a G4. The fact IS, that in almost every benchmark the i7 completely trounces the Core 2 architecture.

Anyway, I didn't say that the Cortex A9 would surpass Core 2 Quads or some Xeon CPU. I said they are *comparable* in performance (even if it was 50% slower, it would be better solution at 1/10th the power consumption).

Btw the XEON 3220 is NOT the same chip as a Core 2 Quad 6600, just the same core inside the chip, which is a totally different thing, please google around, they even support different SSE extentions, iirc. Oh the 3220 is also a tiny bit faster esp for server use.

Anyway, as I said, the point is not a starting a thread of Intel vs Intel chips, but how good would a A9 stand up against a Core 2 DUO and Atom (not every model in the line, and not a Core i7 -which is in a different league).

Regarding Atom, I completely agree, the Atom is overrated, even the 8610 would totally destroy it.

Please get your facts straight before you do any such claims the next time. This is not meant to be offensive, nor a flamestarter, just an advice.

Last edited by feanor on 05-Oct-2009 at 08:15 AM.

@Mechanic

Oh, oh that's a nice dream question -or rather a question with a dream answer :D

What I would be nuts about is this:

1.5Ghz MPC8640D (I don't care if the e600 is EOLed)
with SATA2, USB3, and all the goodies (audio, etc, lots of PCI/PCI-e slots)
a PCI-e 8x for a gfx card AND:

A SpursEngine SE1000 connected to a pci-e bus (1x-4x) that would act as a graphics/math/whatever coprocessor and would make such a system kick some serious ass. :)

And a netbook with a MPC8610 +SpursEngine. With the 2D builtin display and the spursengine chip, the system would provide really good 3D performance (using Gallium3D, which already has a 3D driver for the Cell, I guess porting it to the SE1000 is not that hard, it's the same thing underneath anyway).

These would be my dream systems, in fact, these systems were in exactly this form in some Genesi documents, they wanted to build these, but several factors made them switch to ARM.

Quote:

feanor wrote: @Frek This isn' a thread about Core 2 vs Core i7, have a look here however: http://www.tomshardware.com/reviews/Intel-Core-i7-Nehalem,2057-17.html I don't know what benchmarks you ran, but I guess I could write a custom benchmark of my own that made the i7 look ridiculous against a G4. The fact IS, that in almost every benchmark the i7 completely trounces the Core 2 architecture.

There's no denying the Core i7 (especially recent ones such as the i7 920 is fast).
Read my post again.

Quote:

Anyway, I didn't say that the Cortex A9 would surpass Core 2 Quads or some Xeon CPU. I said they are *comparable* in performance (even if it was 50% slower, it would be better solution at 1/10th the power consumption).

They don't compare at all, and that's my point. The Atom is not useful for computional tasks, The Core 2 is, The Core i7 is. the Atom is not. Even a Pentium 4 easially outruns an Atom.

Quote:

Btw the XEON 3220 is NOT the same chip as a Core 2 Quad 6600, just the same core inside the chip, which is a totally different thing, please google around, they even support different SSE extentions, iirc. Oh the 3220 is also a tiny bit faster esp for server use.

Seriously if you try to be authoritive on the subject, why not research a bit about it?
There's absolutely no secret that the only difference between Xeon and Core 2 (or whatever architecture it's based on) is temperature specification; Ie the Xeon chips can be pushed harder without hitting the roof, The technical architecture is identical.
Some Xeon chips also use a different socket that allows for SMP, however in the case of the X3220 it uses the same LGA755 package as the Q6600 so there's technically no difference at all- both are based on the kentsfield platform.
NO there is no SSE difference at all, the penryn was the first hw platform to introduce SSE extensions- and the X3220 is based on kentsfield as I said.

http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=0&SearchKey=Q6600

http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=0&SearchKey=X3220

You can look it up here if you want to, but I assume with your logic Intel is wrong.

Btw the same applies to

X3230 = Q6670
X3300 = Q9400
etc.
Just look at the architecture and you can quickly determinate they're the same chipsets;
If in doubt just check the processor finder at intel...

Quote:

Anyway, as I said, the point is not a starting a thread of Intel vs Intel chips, but how good would a A9 stand up against a Core 2 DUO and Atom (not every model in the line, and not a Core i7 -which is in a different league).

As already stated, it will not stand up against the Core 2 Duo.
It may stand up against the dual core Atom however.

Quote:

Please get your facts straight before you do any such claims the next time. This is not meant to be offensive, nor a flamestarter, just an advice.

I'm pretty confident my facts are straight, how about yours?

Quote:

And a netbook with a MPC8610 +SpursEngine. With the 2D builtin display and the spursengine chip, the system would provide really good 3D performance (using Gallium3D, which already has a 3D driver for the Cell, I guess porting it to the SE1000 is not that hard, it's the same thing underneath anyway).

Why on earth would you want SpursEngine on a laptop when there's significantly more suitable GPGPU hardware that could easially offer a significantly better performance per watt.

Even a slow chipset such the GMA950 beats the Gallium 3D Cell driver for that matter,
If you don't believe me why not download Mesa and try it out on a PS3 ?
Performance is somewhat disappointing.

@Frek

that one is easy, the spursengine is much more than a 3D GPU. It might lose on 3D by any 3D chipset, but it will decode/encode HD H264 video without a sweat, try that on a GPU.

@feanor

What is the current statut of your project? Like others I really wish you success, and hope to get some good news soon!

Petrol

@Frek

Quote:

Frek wrote: There's no denying the Core i7 (especially recent ones such as the i7 920 is fast). Read my post again.

I did:

Quote:

And regarding i7 the Core 2 architecture is competitive in more tasks than it's not- ie. the i7 rarely outruns the Core 2 (so far- this will ofcourse change with future revisions).

You're contradicting.

Quote:

They don't compare at all, and that's my point. The Atom is not useful for computional tasks, The Core 2 is, The Core i7 is. the Atom is not. Even a Pentium 4 easially outruns an Atom.

I don't know about the last one, I never saw benchmarks between those at the same frequency. And anyway, the point is not about performance, but about power/performance ratios.

Quote:

Seriously if you try to be authoritive on the subject, why not research a bit about it? There's absolutely no secret that the only difference between Xeon and Core 2 (or whatever architecture it's based on) is temperature specification; Ie the Xeon chips can be pushed harder without hitting the roof, The technical architecture is identical. Some Xeon chips also use a different socket that allows for SMP, however in the case of the X3220 it uses the same LGA755 package as the Q6600 so there's technically no difference at all- both are based on the kentsfield platform.

I apologize, I had read in some forum -inaccurately it seems- that the Xeon 3220 offered SSE4 on top of Q6600.

Quote:

NO there is no SSE difference at all, the penryn was the first hw platform to introduce SSE extensions- and the X3220 is based on kentsfield as I said.

I stand corrected on this one.

Quote:

As already stated, it will not stand up against the Core 2 Duo. It may stand up against the dual core Atom however.

As I already said, it does not matter if the A9 has 50% of the performance of a similar-frequency C2D, if it is 1/10th of the power consumption, it's already a winner. If I were to build a server farm and I had to choose between those two, I know which one I'd choose (if A9 delivers what it promises at least).

Quote:

I'm pretty confident my facts are straight, how about yours?

They were not and I apologize. I still insist though that the A9 will beat the C2D on a power/performance ration at a close performance too.

Quote:

feanor wrote: @Frek that one is easy, the spursengine is much more than a 3D GPU. It might lose on 3D by any 3D chipset, but it will decode/encode HD H264 video without a sweat, try that on a GPU.

Most graphics chipsets manufactured in recent time has dedicated hw support for MPEG decoding (encoding is not always accelerated, that's true. but it can however be performed with general GPGPU like OpenCL (CUDA etc))

@feanor

Quote:

If I were to build a server farm and I had to choose between those two, I know which one I'd choose (if A9 delivers what it promises at least).

Like a ARM based blade system?

_________________
Dammy

@Frek

For that matter, OpenCL could be easily ported to the Cell architecture as well.

@damocles

ARm would be a change, but again as already stated in other threads, how many current amiga programs would actually run on ARM?

New hardware though is always good, the more the better, the more choice we can than have.

I am stil awaiting a netbook or laptop amiga.

_________________
Courtesy of SAM440Flex & Amiga OS4.1 only
Flex is 800mhz
A1000 with Classic 520 Amiga OS3.2.1
AmiKit 12
MorphOS PowerBook G4 (which can play youtube vids)

https://blitterwolf.blogspot.com

@petrol

still waiting for mails to arrive. I'm also having some investor meetings the following weeks, to see if I can convince them with these numbers. I'll also send some mails to Hyperion and MorhpOS team today or tomorrow.

@Gebrochen

Quote:

I am stil awaiting a netbook or laptop amiga.

Assuming one existed how much would you be willing to pay for it? This is a semi-serious question. :)

@damocles

sounds like a good business plan.

@feanor

Personally (if and only if it runs AmigaOS 4 at the time of the purchase) I'll be able to spend upto 1000 Euros, I would not need not top class graphic card (but decent performances still) nor would I need a super big screen either, of course enough USB ports, an Ethernet one, a basic DVD reader, and some entry level Audio capabilities would satisfy me.

_________________
AOS 4.1 : I dream it, Hyperion did it !
Now dreaming AOS 4.2...
Thank you to all devs involved for this great job !

@feanor

Quote:

feanor wrote: @Frek For that matter, OpenCL could be easily ported to the Cell architecture as well.

Sure it can be adapted for the Cell BE,
But it doesn't map good on the cell; Infact thats a bit what gallium is about,
generating SPU code for shaders.
Honestly while Cell is a bit more flexible today,
GPUs are way stronger, and with recent changes in the pipeline
it's even possible to extract the power.
Personally I think the Cell was introduced several years too late.

Just compare the Xbox 360 to the PS3 check somewhat recent game engines,
like the cryengine 3; it's not significantly behind the PS3,
still the Xenon is significantly weaker on paper.

@feanor

Quote:

1.5Ghz MPC8640D (I don't care if the e600 is EOLed) with SATA2, USB3, and all the goodies (audio, etc, lots of PCI/PCI-e slots) a PCI-e 8x for a gfx card AND: A SpursEngine SE1000 connected to a pci-e bus (1x-4x) that would act as a graphics/math/whatever coprocessor and would make such a system kick some serious ass. :) And a netbook with a MPC8610 +SpursEngine. With the 2D builtin display and the spursengine chip, the system would provide really good 3D performance (using Gallium3D, which already has a 3D driver for the Cell, I guess porting it to the SE1000 is not that hard, it's the same thing underneath anyway).

kick some serious ass .

Oh Yeaaaahh !

Although I have no need/want for a laptop (at this time) I am willing to pay for power.