The Xena module

Is the purpose of the Xena established yet? If it has I have not been up to pace and this might be a bit late.

According to the X1000 webpage the contents of the Xena was up to "us". I gather that what that means is that they will sneak around the Amiga-forums to see what we wish, and for now, I haven't seen any particularly wishes.

The reason for this post, is that I want you to throw ideas around in what you want your Xena to do.

Remember in the old days, it was these extra modules that made Amiga better. Usually by offering greater graphics and multitasking.

My suggestion is simple but since I am inept in programming is just an idea that might not be possible.

The original processors of the Amiga was graphics, but that is not a problem anymore, with good graphics cards you can do much the same. Multitasking with good sound quality isn't a problem either.

But I remember the saying that "The Amiga can emulate everything!" was a particular statement to listen to.

Can Xena be programmed easily and maybe on the fly to emulate all sorts of stuff? Like ANY software from anyware to be run like it would on a native hardware system? Because there might be a future in that! It would also take care of our legacy-problems. Imagine running everything on your Amiga fast by just teasing the Xena-module properly with some small piece of software. Imagine running exotic software in a window on your screen in the workbench. "Xena- the holistic emulator"...

Or maybe you could program it entirely to be more consistent on running rich graphics that day, and program it to be more consistent on running many programs smoothly the next day? Like an "up to you environment" ? As versatile as the bootblock back in those days.

I have no idea what is possible with that Xena hardware, as you probably have gathered in my novice sense of hardware understanding. But I am just trowing out ideas because it makes things make sense. You do the same now.

I hope it's not to late and there is a possibility for this.

@Geoshaper

Don't think of it as something that would accelerate stuff the cpu or the gfx subsystem would normally do. It not suited for such tasks at all.

Think of it as a IO hardware emulator. Being able to 'emulate' stuff like:

- An A1200 clock port
- A midi port
- A network card
- Digital joystick ports
- Oscilloscope I/O
- Robotics controller
- Audio chip
- USB ports
- LED display driver
- CIA emulator
- .. and so on.

Most people will probably have little use for it. You'll still need some kind of adapter card to have that clock port (or whatever) on, which in such low low production counts will cost just as much as a dedicated pci card anyway.

If you want to look at the 'strong' points of the x1k platform you should not focus on the xcore stuff, since it won't matter much to most people anyway. It's a nice addition for the geeky hardware people but little more than that.

@Geoshaper

Read Here:
http://amigaworld.net/modules/newbb/v ... topic_id=30985&forum=42&2

I found their ( A-Eon ) speaking a bit confusing ...
It was firstly said that Amiga X1000 will be a bit like Amiga was in the past in what it had of particular ... The particularity of Amiga's of the past was that the architecture was precursor of actual video game console (and with more power than other computers available )... Now, they said Xena was that particularity ..
if Xena if only suited for external devices I/O ... it is not really developed for Amiga OS 4 but more for industry (robotic, domotic, etc ... )

I'm a but disappointed when they said somewhere in another post "Emulating AGA is not a good use of Xena" ... So that mean that using Xena for specifix codings things is not good ( ie : Physic engine, IA calculation, CPU/GPU emulation ) ... But then, what is the interest if XENA for actual Amiga OS 4 users and developers ? We are all hobbyist without the needed financial ressources to create external card/components .. so ... Xena will then be useless for us ... and for our Amiga OS 4

I think they should really give us more information on what the Xena 8 cores can/cannot do !

Without this, buying a X1000 will be haZardous for all actual Amiga OS 4 users ...

@freddix

As I understand Xena, and conclusion i draw from some presentations of its using on YouTube.

I have seen it used as oscilloscope I have seen it used as audio processor, and some one also pointed out the limited space on chip for storing code.

The conclusion for this is that its best used as io interface for anything really, a clock port should be easy to do, ciaa and ciab should be easy to implement, but its to small for 680x0 emulation, its possible that it can be used for byte conversion, GBAR TO ARGB, 16b PC to 16b AMIGA, simple bit conversions, p2c or c2p might be some it can do, stuff that require little code, but can give big bust.

@freddix

I don't see how that can be the case. The (as yet) unused FPGA on the Sam boards hasn't stopped any purchases AFAICS, so I doubt that anyone will not buy an X1000 just because there is a lack of Xena programming.

Try to think of the XMOS chip as a programmable buster chip. In the classic desktops, the Zorro3 slots were controlled by it. These allowed all sorts of diffent cards to be plugged in and utilised. With the XMOS chip on-board, software can be written that makes the XMOS chip utilise connections on the Xorro slot, and any outside devices that may be used. As Orgin described earlier, this opens up a wide world of I/O possibilities with only a simple Xorro card with the necessary connectors on it. The XMOS software can then use the hardware as an I/O device.

XMOS is not designed to be a super-CPU, but it IS designed to be a super I/O chip. Look at how many classic expansions there were for I/O devices (IDE, sound, serial/parallel etc), and all of these can be handled via the XMOS chip and a suitable backplane, and all without buying expensive hardware.

Simon

@Geoshaper

Go read up on xmos's web site and some forums at
http://www.xmoslinkers.org/projlist/
and
http://www.xcore.com/forum/

Some discussions about this from a while ago at AW.net with some answers for you

http://amigaworld.net/modules/newbb/v ... _id=30943&forum=33#544974

http://amigaworld.net/modules/newbb/v ... _id=30893&forum=33#542053

http://amigaworld.net/modules/newbb/v ... c_id=30398&forum=2#539872

@freddix

Quote:

If you look on the aminet, you'll see a whole series of hardware hacking projects. The Xorro port + Xena provides a great way for hardware hackers to create their own add on hardware, without having to create a PCI/PCIe device or a USB one. As such, it is for Amiga OS 4 users, but will obviously be of more interest to those who like making stuff with electronics.

Hans

@Hans
I understand this, that Xena is good for hardware support writing ... but hardware support by hobbyist is a bit hasardous for support/development ...
We can see that for example video card drivers always need updates ... and are never "definitive" version ... and, seeing that they are developed by far more than 1 or 2 developers .... hardware support by hobbyist can takes ages to see light ...

I understand it can be positive for Amiga "on the paper", "on the theory" ... but for real app ... it will takes ages before we can see valuable drivers using it ...

Personally, I hope the Xena can help maybe in these ways (that can be usefull for the AmiDARK Engine) :
- Asynchronous loading of various objects ( 3D models, sounds, images ) by accessing HDD and memory
- Asynchronous support for playing audio
- helping in the process if sub tasks (like IA, Physics, I thkn physics is too heavy for 1 xena chips)

Maybe we can merge 2 or 3 Xena cores for the same tasks splitted in 2/3 parts ?

EDIT : In some links provided, it is mentioned that Xena can be used for emulation ... but A-Eon said it is not a good use of Xena ... explaining more my disappointment.

Regards,
AmiDARK.

@freddix Quote:
I'm not convinced these are sensible uses of Xena, when the X1000 already has (by Amiga standards) a super-powerful dual-core CPU. (Although playing audio would certainly be possible for Xena.)

In fact, Xena is not intended to replace the CPU, it is suppose to augment it. More specifically, it is not so good at brute force calculations (for which we already have a powerful CPU), but can instead take the load off the CPU by handling I/O, and do so better than the CPU could hope to. Xena is also extremely good at timing (down to a few 10s of nanoseconds in the best case, IIRC), which again makes it good at both generating output signals AND at processing incoming input signals.

When A-Eon say it is up to "us" to decide what Xena should do, they mean it is down to developers to think-up & make/sell ingenious uses of the Xena chip, just like they used to extend the old Classic Amigas with ingenious hardware (the Toaster being the ultimate example).

@freddix

Quote:

HDD to memory loading, I'm not sure Xena has useful access to either of those things to be superior at that tast compared to a CPU or to SATA chip with DMA. Perhaps if you use it as an HDD controller chip and have it use DMA feature, but I think that is already present elsewhere.

Considering what the Xcore chip can connect to on the motherboard other than the Xorro slot, I think you'd lose performance by making it responsible for certain things like this.

Audio might be reasonable, make it be an audio chip with DMA... But there's probably already an audio chip in there, I think it's been said to be HD audio even.

@all :
Quote:
Do you really think it exist enough Amiga developer that can elaborate cards for Amiga ?
I don't think so.

I've never said "more powerful than CPU" ... but for example, you play a game an go through 2 areas ...
The CPU Can process the global render preparation and Xena could load/release objects from memory ...

Imagine in the game engine instead of direct load of object, it can call a Xena function I will develop that will load the object (like a background task) and when it is finished, will add it to the 3D render list ...
It's the kind of purposes I think ... No need for much power ... but simply working at a backgrouns task ... with signals between main/background tasks to not crash ;)

@freddix

The way I understand the XMOS chip is that it has a serial data hook-up and many ports for I/O hook-up. The Xena is just an interface to external devices. It doesn't have enough bandwidth to connect to main memory quickly.

I think one of the first things we may see is a MIDI port.

Chris

@freddix Quote:
Anyone with a bit of electronics experience will hopefully be a potential "Amiga developer": I have suggested several times that someone (hopefully A-Eon) create a "general purpose" Xorro card that would make it easy to experiment hooking Xena up to anything they want. BTW, ever heard of the Minimig? Natami?

Failing that, we at least have Individual Computers...

Of course, no-one really knows, but there's no need to be negative about these things! Anywaym, A-Eon put the Xena/XCore chip on the mobo in the hope that people would find uses for it. If they don't, well, it didn't add much to the price of the X1000, so no need to worry.

Quote:
That really wouldn't make much sense, even if Xena was as powerful as the main CPU! Do you know why the Phase5 PPC cards were *so slow* for most things? It's because they had to flush the CPU caches when switching between 68k & PPC. The same issues would happen here - the PPC's cache would need to be flushed before/after Xena did her job, most likely rendering the benefits of using Xena moot on such a small task.

The only way the P5 PPC cards were able to do anything useful was when you gave the PPC a big job which didn't require communicating with the 68k. Same would apply to Xena, assuming she was powerful enough to offload the CPU's work... But since Xena isn't as powerful as the main CPU, it doesn't make sense to give Xena brute-force tasks.

@All :
All your speeches confirm me that Xena if for only a few people here ... I have some knowledge in electronic and microprocessors but I don't have money to spend to create a new interesting card for Amiga ...
Xena is then useless for many people as it's only useable for external cards ...
So, it's interest to make Amiga OS 4.x interesting comparing to other OS is really low ...
Why ?
Simply because now, PC computer are dual/quad/multi cores that are times faster than what X1000 will be and even if these PC use main CPU to handle PCI/PCI-e cards (what Xena can do technically), they will be faster than X1000 ... What is the interest then ?

Recreate something that will be slower than what is done on other computers ...
It's not for me the *genius* idea I expected to see ... to make Amiga different and more attractive than PCs ...

@freddix

Quote:

I don't think that you quite understand the hobbyist electronics thing. Users making custom hardware for themselves don't have to worry about updating their drivers, because they're the ones designing and building the driver. There won't be a new version of the hardware until they decide that they need something different.

What this does is make hardware hacking more accessible to people. Connecting the Xorro slot to sensors, switches, and outputs isn't particularly difficult. In many cases, so long as the signalling voltages match, things can be hooked up directly. This is far far easier than having to create a USB driver or a PCIe card and driver, and puts it within reach of people with just basic electronics skills.

So where's the value for those who aren't developers? That will probably come when people have made stuff that they wish to share with others. Plus, there's nothing to stop users from learning how to connect stuff up and program it, even if it's only at a basic level.

Hans

Some years ago I designed an FSK decoder (in hardware) to receive weather facsimile transmissions from HF and display them on the Amiga. These days, of course, you can get the same info from the Bureau of Met over the 'net.

However, it was a design exercise and fun. I built the tone decoders into a box along with a dedicated 68HC11 that generated a bit stream at 9600 baud for transmission to the Amiga through the serial port. I still use it at times, although the RF transmissions are so much more difficult to receive these days with the pervading RF broadband noise caused by so much "recent technology".

Application #1: I could throw away the 68HC11 in the decoder box and simply run a few signals directly to an XMOS chip. I would then be able to decode the data and pass it directly to the CPU for display.

Application #2: I wonder if the XMOS chip has sufficient grunt to be able to perform real-time FFTs? I could then do away with the tone decoders as well, leaving only an audio cable from the receiver output to the Xorro board (which would have just an ADC).

When you think of it, sort of like what you can get on any PC with a DSP cound card.

This XMOS feature is going to give us the technology to catch up with the hardware of PC cards. Someone with the ability to make small, cheap Xorro plug-in cards could make and and sell audio, video or data capture or interfacing equipment, and sell it with suitable support software for the XMOS chip.

Before someone says that you can get all that for a few dollars on a PC card, you can't get the necessary software for it and never will. With the XMOS chip and the Xorro slot, we'll be back in the field ourselves.

@Geoshaper

According to my understanding. To take use of xena is 100% up to third party developers. Do not expect A-eon to deliver more than the needed developer tools (hopefully).

You all speak about theory "many people can develop card to make PC card compatible, to handle audio input, etc ..." ... but like some said .. "only if developer want to share" ... and developer that'll do that kind of just are few ... and releases will be more rare ...

It can have a big interest in industry I think ... and it's there that it can take its full potential. But for Amiga OS 4.x users ... low interest ...

So finally, Xena is just a "toy" in an Amiga ...