@bernd_afa
Quote:

and i still not see wy many need a X1000 and a xorro slot for 1500Eur and what IO Card that cant do as a PCIe Card or USB 3 Card

1st: you don't know the price so stop stating your speculations as facts.

2nd: the xorro and Xmos chip probably add a total of ¤20 to the X1000 price, so I don't understand how your price implications have anything to do with the recently released faq (which is about xmos).

Xmos is a costless playful thing that can add a lot of fun to the most powerful Amiga machine ever released, and comes built in instead of "add-on" so that is safe for developers to use it. Think of the X1000 as french fries and the xmos chip as some free ketchup, to go along with it.

_________________
SamFlex Complete 800Mhz System + AmigaOS 4.1 Update 4
Amiga 2000 DKB 2MB ChipRam GVP G-Force040 Picasso 2 OS3.9 BB2
AmigaCD 32

@bernd_afa Quote:

i think they (should) use a switch, so Xena can connect to CPU or Xorro with full wire

Quote:

but when they have hardcode it, then you can only transfer 25% (39 megabyte)of the 156 megabyte(with xmos and 500 MHZ) with the CPU to the Xmos Chip.

So you think it would be better that someone could accidentally program the switch wrong, and then send wrong voltages onto the CPU's local bus? Ah yes, fry the whole X1000 due to a small programming mistake - that sounds really great!

(sarcasm off)

Having the CPU bus hard-wired means the software tools can ensure that the wires connected to the CPU bus are always set correctly. Then you can't damage the X1000 by software alone.

Last edited by ChrisH on 14-Mar-2010 at 11:22 AM.
Last edited by ChrisH on 14-Mar-2010 at 11:21 AM.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@tiffers
I'm sure that bernd_afa already knew those FAQ answers. Not to mention people (inc me) have already given similar answers as the obvious explanation some time ago.

Last edited by ChrisH on 14-Mar-2010 at 11:23 AM.
Last edited by ChrisH on 14-Mar-2010 at 11:21 AM.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@KimmoK

One good example should be studio quality audio processing.

From 32bit one can get nought 24bit channels. If enviroment is enought quiet (X1000 motherboard), I mean no signal noise and add DSP functions for that with good software, we have killer platform.

Digital oscilloscope ??

Or anything with DSP needs.

My random thinking again :)

@tiffers

Quote:

Quote: From the FAQ: Quote: By including Xena on the motherboard we ensure that every X1000 owner has one, and that makes it a baseline feature that is safe for 3rd parties to support.

But Bernd is right --- there is still no mention of what benefit there is to having it as a "baseline feature".

As has been pointed out, it's not suitable as "extra muscle", i.e. for purely computational tasks. Not only is the raw CPU power negligible compared to the main CPU, but there is also the issue of communication. And what the FAQ cleared up is that the *maximum* number of I/O lines on Xorro is 46 (98 pins on a PCIe-8x connector, 47 taken for power, 5 for JTAG), which is apparently three times as many as are connected to "the CPU local bus" (whatever *that* is). So something like 16 I/Os connected in a usable way to the CPU --- which is way too few for any reasonably fast data transfer, given what those I/O pins can and cannot do. Add the scarcity of internal memory, the maximum per-thread throughput of 125MIPS, and there simply is no point, ever, to pass computational stuff (and that includes "emulation" of any kind) from the main CPU to the XCore chip.

Which leaves what the chip is actually designed to be --- an intelligent I/O controller, which can react with low latency to external events.
But on the X1000, the XCore chip has no externally accessible interfaces. So without an additional plug-in card for Xorro, it can only react to events created by the CPU --- and if you leave event creation to the high-latency CPU, your low-latency I/O processor is kinda pointless.

So in order to do something useful, one needs something to plug into the Xorro slot. Which means a double-sided, shaped PCB, which then provides some connectivity between the proprietary-as-all-hell Xorro slot, and whatever one wants to connect on the other side.
Let's repeat that --- despite the XCore chip being "baseline", the only way to get it to do something useful involves buying an additional, decidedly non-trivial, custom-made PCB!

Yes, as long as the chip on the motherboard is actually what the hardware developer needs, the PCB will be slightly simpler, because it won't need to include an XCore chip. On the other hand, if the hardware developer's needs are not met by that chip, too bad, that's the one you get.... And the slot is internal, anyway, and not hot-pluggable, so it's not like your average user is going to have half a dozen "Xorro cards" lying around, swapping them in and out all the time, and enjoying the savings from using the one chip between all of them.

How much more sensible would it have been to provide a 38 pin 2.54mm DIL header, with the "CPU local bus" signals, JTAG, and a variety of power lines, allowing hardware designers easier connections, and the ability to use whatever XCore chip they feel like? What was gained by making the chip "baseline" that would be lost by just having that header?


Oh, and as an electrical engineer with some real world experience, I am amazed by the idea of speccing Xorro with 5V across all the power connections --- on an actual PCIe-8x slot, some of them are 3.3V, and some are 12V. So plug a Xorro card into a real PCIe slot, bzzzzt, 12V where the card expects 5 (or worse, if the designer connected all the Xorro 5V pins, shorting 3.3V and 12V in the motherboard's PCIe slot). Plug a PCIe card into the Xorro slot, bzzzzzzt, 5V where the card expects 3.3V.
Furthermore, whoever decided 9 VCCs and 38 GNDs was a good combination, presumably because it's roughly what PCIe-8x has, hasn't quite grasped the reason for all the ground pins in the variable section --- they are shielding, around the 2.5Gbit/s differential signal pairs. Xorro does not *have* any signals even remotely like that, so presumably the ground connections are mainly used as a return path for the VCC ones... at which point, having 4 times as many is a tad silly, and wasteful.

Last edited by umisef on 14-Mar-2010 at 11:32 AM.

@bernd_afa

Quote:

and i still not see wy many need a X1000 and a xorro slot for 1500Eur and what IO Card that cant do as a PCIe Card or USB 3 Card

You dont see *ANY* difference between USB3 and Xena real time signal processing capabilities ??

With that chip, if x1000 mobo iks enought good made, there is many different areas, where it can be used in pro-level.

Good IO cards to PCIe are pricey. Consumer motherboards in PC world are crap.

But, if you think, that never anyone anymore will use Amiga to any pro work, it is pointless.

But with that chip it is possible :)

@umisef

Quote:

umisef wrote: Oh, and as an electrical engineer with some real world experience, I am amazed by the idea of speccing Xorro with 5V across all the power connections --- on an actual PCIe-8x slot, some of them are 3.3V, and some are 12V. So plug a Xorro card into a real PCIe slot, bzzzzt, 12V where the card expects 5 (or worse, if the designer connected all the Xorro 5V pins, shorting 3.3V and 12V in the motherboard's PCIe slot). Plug a PCIe card into the Xorro slot, bzzzzzzt, 5V where the card expects 3.3V.

While I have no experience as an electrical engineer I have never seen a motherboard with a PCI-e x1 inline with a PCI-e x8. Nor have I seen a standard board with a PCI-e x8 slot so far back.

I'm going to assume (which we both know makes an ASS out of U and ME, but "You pays your money, you takes your chances!") that the inline x1 slot will be industry compliant, especially with respect to voltages etc. I'm guessing that sticking a Xoro card in anything but a x1 slot will be impossible, and in a x1 slot will merely be useless on anything other than the X1000 and future possible generations of hardware from A-Eon, but certianly not disasterous for either the Xorro card nor the 'alien' board hosting the PCI-e x1 slot.

I also assume it will be the way to get acess to the extra IO paths (?) on the Xoro slot, however un-helpful this will be. This is why I asked in my 'Official X1000 Questions' thread if there will be a 'simple bridge card' sold with all systems, so that the extra IO on the Xorro slot will be accessible via the PCI-e x1 slot immediately with every system sold. This card could be removed when another Xoro card was installed, which would either provide the same bridging, or itself make use of those lanes, removing the necessity for the bridge.

I understand your concern with randomly changing something that is industry standard and making it totally customised. I also understand anyone with any intention to develop a Xoro card would have enough nouse to get it right. A-Eon should provide proper documentation to developers so that the design of the slot is very clear. The physical aspect of the Xorro card should make it impossible to insert into a standard x4 x8 or x16 slot, so the dumb end user shouldn't be affected.

It's easy to wave around your professional qualifications and experience, but a little logic and common sense is also a good idea

@TrevorD
I'm interested to know about the write once memory. Will this be writeable to 'everyone'? You really don't want someone writing to your XCore write once memory, without being totally aware of what's going on.

I also have wondered at the possible use of the Xoro slot, as you can really only stick a single card in the slot. I really wouldn't want to have to swap cards a lot, as these slots do have insertion count ratings. If developer A comes up with a brilliant idea using XCore and makes a card, and developer B has a totally un-related idea using XCore and makes a card, how do I use both cards? WIthout co-operation between developers and maybe A-Eon there are potentially many possible conflicts (assuming many developments using the XCore).

@Kronos

"just that sofar noone has come with a single plausible problem that could be solved with the chip onboard a full-size ATX motherboard"

Perhaps there is no problem to solve.

xena seems more like enabler to me.
You do not have to get xcore board separately.
One step less for Amigan to start using/experimenting xcore.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@thread

One thought is that the Xorro slot may not actually be aimed at the home / desktop market. Not everyone who buys a computer sticks it on a desk and runs a word processor on it. Statements like "this is hardly a deal-breaker as far as the AmigaOne X1000‭ ‬is concerned" (although related to the cost in the original context - it can mean: 'it's not costing you heaps more to be included, so if you don't use it, it's not really a reason not to buy the system') or "We‭’‬re making the most powerful AmigaOne yet regardless of Xena"; "you could say that Xena is a gift to the Amiga‭’‬s geek heritage" and "it‭’‬s non-conventional" and "We expect people to use this to make interfaces to all kinds of legacy devices,‭ ‬as control hardware and so on" make it clear this isn't 'just another expansion card', nor that it will be used for everyday computational purposes. It's ewasy customisable IO. You can offload the IO to the XCore and let the CPU go about running the rest of the system. Just like the classic chipsets did. Admittedly we're playing with multi GHz processors now, not single digit MHz ones, but it's also a 'geek port' which is there for 'possibilities. And come on... tell me how selling possibilities doesn't fit in with modern IT equipment?

Industrial applications (which is where a lot of good money is to be made I'm guessing) may well see the Xorro slot and pounce on the idea, Producing a single card so the system could be used in a dedicated role might be exactly what gives the X1000 the market share it needs to gain popularity for both it and AmigaOS.

We all need to stop thinking inside our home desktop paradigm (How can I do the same thing as everyone else, but faster so I can brag how mine is better/bigger), and actually think of things and ways to use this new system and it's possibilities.

A-Eon may be selling the X1000 as a complete system, and that's a great way to get a standard development system into the hands of people who can take it places. They have the ability to produce and sell the board as standalone, or to even develop other purpose-built enclosures for the system. The board could be re-designed with essentiall the same hardware but with less slots for dedicated purposes, and maybe this is some of the stuff they've done, but haven't and probably won't release as general knowledge.

The mindset of people who zero in one one thing and just pick it to pieces without even considering anything outside their narrow pre-conceived view point never fails to amuse me.

You know what? It's pointless even arguing why it's there. It's there. You can't change it. EVER. What will you achieve by arguing against a business decision that wasn't yours, and has already been made? Are you trying to somehow teach the 'fools' a lesson by driving down interest and possible sales? I don't see _anyone_ on here as feeling that way. Perhaps you just want to show how much better you are than the 'fools' who made the dumb decision? Not so much better as to actually put your money on the line and atually produce something though.. right?

If the X1000 had no Xena, it wouldn't cost significantly less (according to TrevorD)... so if you buy one and don't use the Xena/Xorro combo... so what? What is the real point of all the nay-saying? Could it be the "it's-not-a-real-amiga-if-it-doesn't-have-a-custom-chipset-with-100%-accurate-timing-with-the-classic-harware" crowd trying to prove Xena just isn't a real custom chipset and they're not going to be fooled by some marketing mumbo jumbo? Seriously, if you want an old Amiga with an OLD chipset, go to ebay. The Xena is at least an attempt to give modern, commodity hardware a custom-like chipset that not only can do cool things no one has dreamed of (look at the hacks people did with the copper), but it can change the way it does those cool things? You see someone's implementation Z and you can make it better, you have that power! How many AGA bugs did developers wish they could fix? The possibilities of Xena are tremendous, especially with a soruce-sharing community. Can anyone suggest a way to make the Nemo board a better PPC board without blowing out the cost to produce the board (which I guess is the aim of the Xena/Xorro inclusion)?

I'm all ears.

Last edited by tiffers on 14-Mar-2010 at 12:59 PM.
Last edited by tiffers on 14-Mar-2010 at 12:51 PM.
Last edited by tiffers on 14-Mar-2010 at 12:43 PM.

@bernd_afa

"...But 1500 Eur and maybe more is really very very much. ...
and i still not see wy many need a X1000 ..."

To run AOS4.x and it's applications in 4...8 times the speed of SAM440ep?
(For that, x1000 costs "only" twice the amount of SAM's price.)


I would really love to see some new lowcost PPC board as the lowend AOS4 machine.
Perhaps with PA6T(or some cheaper SoC)+1GbRAM+natamiGFX+AMDSB600+1PCI x16 (or x8)+1PCIx8+xena+case+harddrive+AOS4, for 500¤?

Last edited by KimmoK on 14-Mar-2010 at 12:54 PM.
Last edited by KimmoK on 14-Mar-2010 at 12:53 PM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@tiffers

Quote:

It's ewasy customisable IO. You can offload the IO to the XCore and let the CPU go about running the rest of the system. Just like the classic chipsets did.

There is no IO to customize. It is like Amiga custom chipset without video and audio output. You can add video or audio output but it costs extra money.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Yes, we all know that. Having the Xena onboard doesn't give you extra IO ports on the back/front of the machine. Unless the CF slot is connected? Yes, you would need a Xorro card to get those IO lanes actually doing something useful.

As I said, it's not being touted as 'everyone will/can use this' but as 'everyone has one'. From a hardware developer's point of view it would be good to know everyone has one if I found a groovey way to use it, and to produce something much more quickly and cheaper than if I had to do it a more traditional way.

Are we looking at it from the right perspective? Everyone's talking "How can it make my computer better" and Trevor seems to be saying they included it to "make developing for user's computers better".

You see, the point of all the arguing is that people aren't looking at it the right way.

@tiffers

The original Amiga came with all custom chips, so everyone had it... The thing is it was included because it had uses. This doesn't seem to be the case with that XMOS chip. I guess some people are wondering if it isn't there just because it makes it look like more "Amiga"... But: what's the use if it has not benefit of having it ?

_________________
http://www.warpdesign.fr/

I have a question: The xmos chip has 64kb only, but it is possible that could access the main ram of the system for bigger things?.

_________________
Amiga 1200 desktop. Apollo 030/50 Mhz 8mb ram + ClassicWB + Wb 3.1
Amiga 500 + ACA500plus 8mb + 30gb CF
Raspberry Pi 3b+ and Amibian 1.4
Mac Mini G4 1GB Ram with the butterfly!!

My opinion follows. (Note that this is an opinion and not a statement of fact)

IMHO I find it difficult to believe that someone would intentionally include an XMOS chip when designing an AmigaOS4 compatible motherboard. IMHO it would explain a lot if this motherboard was initially designed for some specific market, (e.g. manufacturing robotics), and that A-Eon adopted it. IMHO there is nothing wrong with that if you get what you want, and having that something extra is always better than nothing.

I notice that the chip spec has increased to 500mghz. I wonder if this has been coupled with an improved chip design. A next gen XMOS chip with more data lines, more memory and an FPU would be nice. I see no news on this development on the XMOS site. (or I missed it)

_________________
Leo Nigro, CTO Commodore USA, LLC
Opinions expressed are my own and not those of C= USA.
Commodore/AMIGA "Beautiful, High-Performance, Home Computers for Creativity and Entertainment."

@ChrisH

>I'm sure that bernd_afa already knew those FAQ answers. Not to mention people >(inc me) have already given similar answers as the obvious explanation some time ago.

no, there is a big diffrence.

The answer in the FAQ sound now good to me.

But when you look here on this offical Page you can read

http://hyperion-entertainment.biz/index.php?option=com_content&view=article&id=136:hyperion-entertainment-cvba-and-a-eon-technology-cvba-announce-strategic-partnership&catid=36:amigaos-4x&Itemid=18

""""
and will also appeal to those looking to exploit the incredibly versatile onboard functionality offered by the unique “Software Defined Silicon” technology from XMOS (www.xmos.com) which, in true Amiga tradition, provides the AmigaOne X1000 with a custom chipset of potentially unparalleled flexibility and scalability.
""""

and compare the XMOS Chip with Custom Chip is really not ok.

And when they change their minds or accept that, they should not do make hope on the Hyperion Page that this Chip can do great as the Amiga Custom Chips from the A1000 in compare to PC.

Many do not read the FAQ, only the Hyperion Page.

Last edited by bernd_afa on 14-Mar-2010 at 02:46 PM.
Last edited by bernd_afa on 14-Mar-2010 at 02:46 PM.

@BigBentheAussie

There has been a lot of updates in xcore specs in the wiki pages. I think 500Mhz appeared there around january/february.

btw. It has been said several times that the board was designed ((mainly)) for AOS4 niche.
Addition of xcore + xorro added very little extra cost.

"It was suggested to us by our‭ ‬hardware partners who have a lot of experience working with XMOS products,‭ ‬and we were convinced.‭ ‬As for why Xena,‭ ‬the answer is simple.‭ ‬It‭’‬s a cheap addition to the board,‭ ‬so this is hardly a deal-breaker as far as the AmigaOne X1000‭ ‬is concerned.‭"

Last edited by KimmoK on 14-Mar-2010 at 02:46 PM.
Last edited by KimmoK on 14-Mar-2010 at 02:45 PM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@mlehto

>Or anything with DSP needs.

>My random thinking again :)

when you look at DSP Code, there happen many data access.and when the transfer rate of the data is slow or have few ram work not good.

its lots more easy to code such DSP stuff when you have floating point.
also when want do software synthesizer, delay or a reverb you need lots delay buffers.

But the Chip have only few RAM.

you can look at the hd-rec sourceforge page that the high quality reverb use lots more than 64 kb ram, also the Delays.an thats only 1 delay

and when there is not enough room in the xmos chip to store the data or samples, then all run slow and is lots more work to code.

the xmos chip is maybe able to do a parametric EQ

this calc is so simple that a 1 GHZ CPU can calc over 200 with native code and programming is more easy with float.so the CPU can also do whats need.

also mp3 decoding is possible but this cost on a 1 GHZ CPU only 3-4% CPU load.
And there is a second Core better.

also more important speedup you get when have altivec.when process left and right channel at once speed up 2*.

>To run AOS4.x and it's applications in 4...8 times the speed of SAM440ep?
>(For that, x1000 costs "only" twice the amount of SAM's price.)

ok, here you are right, when look at a SAM System (800 Eur) then the X1000 is better performance/money also if it get only A Titan CPU and no Altivec.

but when user use OS4 as a secondary system because of missing apps, then there is better to make a board at the price of SAM with a more powerfull CPU and PCIe i think.

Last edited by bernd_afa on 14-Mar-2010 at 02:53 PM.

@KimmoK

Quote:

KimmoK wrote: @BigBentheAussie btw. It has been said several times that the board was designed ((mainly)) for AOS4 niche. Addition of xcore + xorro added very little extra cost. "It was suggested to us by our‭ ‬hardware partners who have a lot of experience working with XMOS products,‭ ‬and we were convinced.‭ ‬As for why Xena,‭ ‬the answer is simple.‭ ‬It‭’‬s a cheap addition to the board,‭ ‬so this is hardly a deal-breaker as far as the AmigaOne X1000‭ ‬is concerned.‭"

Yes, I know very well what was said. My opinion still stands however. As with everything related to the X1000, I doubt we are being told the full story. I don't say this to troll as I really don't mind either way. Better to have an XMOS chip than not I guess. However, please note that it was "suggested" by the hardware partner and NOT a "deal-breaker". IMHO the hardware partner wants to go after a specific non-Amiga market with the board that may actually benefit from XMOS tech. Again, nothing wrong with this approach, and it is similar to what ACube wanted to with SAM. Leveraging another market for the mobo is a good idea, and I would actually be relieved if it was the case.

_________________
Leo Nigro, CTO Commodore USA, LLC
Opinions expressed are my own and not those of C= USA.
Commodore/AMIGA "Beautiful, High-Performance, Home Computers for Creativity and Entertainment."

Quote:

btw. It has been said several times that the board was designed ((mainly)) for AOS4 niche. Addition of xcore + xorro added very little extra cost.

Little cost and hardly any use as of now... What's the point ?

_________________
http://www.warpdesign.fr/