Post a Comment
Sweet, I wasn't expecting that! Honest! ;-)
No, seriously, this is great news for anyone that has been following Amiga for the past couple of years... we now have to 2 very capable pieces of hardware. Sam440 for the low-end and now X1000 for the high-end.
Well done Hyperion! :-)
Alex.
Nope, Xenon and G5 are very different. The G5 is an out-of-order quite wide superscalar processor with a deep pipeline and no hyperthreading.
The XBOX 360 Xenon is an in-order not so wide superscalar processor with a shorter pipeline and hyperthreading.
Instead, the Xenon is rather similar to the Cell PowerPC core.
Edited 2010-01-05 18:32 UTC
Potential I see, lots of potential. The only computer that allowed itself to be bleeding edge in 1994 although you bought it 8 year earlier was the Amiga 2000 thanks to its incredible expandability potential (accelerators boards that fitted a special expansion slot).
It is indeed from those (lost)times that I don't see such potential in a modern computer.
Sure it has just already minimal Linux to run developing tools, benchmark and computing stress tests.
See for example this teasing image of X1000 running Processor stress tests brought to public by A-EON firm:
http://a-eon.com/gfx/hidden.png
In this age, when you bought large number of processors in quantities, manufacturers will give you the software developing tools bundled with it and usually these tools are already written in Linux environment (even if minimal kernel distro).
Any Portings could be done in not so much time...
Edited 2010-01-05 14:45 UTC
I was just talking about this with a co-worker recently, and also read about a speech by a previous MS employee. I was wondering if CPU's would soon have a "SPU", streaming processing unit, like a mini CUDA or Larabee: unit built in, much like the FPU of old. It seems to be gaining momentum. The Cell Broadband / 8i did it, and could sustain massive throughput. China is doing it with the Godson / Longsoon 3. The original Godson will be quad core, with 4 CPU's. The later generation is for supercomputing, and will be 8 cores. Four cores will be a "next generation" of the Godson 3, and the other four will be for streaming data.
Clearstream tried to do this with a daughter card with 8 or 12 (two models) parallel MIPS processors. It's main limitation was limited support (they were very Sun Solaris friendly, less so otherwise) and a bajilion and one dollars per card. There are Cell BBE and 8i cards, but the price for those is close to that bajilion US$ price. (Actually, closer to $10,000. One could buy a nice Fermi workstation for that.) Intel said they would probably do this with Larabee "real soon!", but no solid plans yet.
If finances weren't so tight here, I'd get one. Personally, I'd be interested to see what can be done on this with BOINC. Surely a BOINC / SETI@Home client will appear within minutes.
If Intel's Larabee and AMD's Fusion projects are any indication, then its heading in that direction. First step will be to get the CPU and GPU on the same die, then the next inevitable step will be to make the GPU more general purpose (GPGPU). Or more likely, we'll have both steps happening simultaneously.
Intel's goal with Larabee (which is more ambitious - effectively merging the CPU & GPU concepts) may be a tad early (thus why its debut as a consumer product has been delayed), but with those projects, and the momentum/desire behind CUDA and OpenCL, the writing is now on the wall: soon enough, GPUs won't be just for graphics anymore.
For a number of years now graphics cards have been used by advanced music production software as a means of rendering real time audio.
I forget the technical name for the technique, but essentially even older graphics cards were good for *some* music filters (due to the video RAM and the way how GPUs were designed to process some graphical filters efficiently.
IIRC the technique involved creating an image map for the sound and running it through the graphics buffers in a similar way as you would with 3D modeling.
This technique allowed for the more humble system to truly parallel process audio in a time before dual core/processor PCs were common place on the desktop / smaller studio.
Apologises that it's as most in depth link as you were looking for (it was a quick and lazy Google at work), but the following link from 2004 shows the concept and some of the guys behind the work:
http://www.tomshardware.com/news/audio-supercomputer-hidden-graphic...
Very true.
In fact, now you mention it, you've reminded me how the practice of converting photos to audio is/was fairly common among some electronica acts (one of the more known instances is how Aphex Twin's face appears if you play the intro of "Come To Daddy" in a spectrograph).
IIRC the process is very straightforward too as all you need is a grey-scale image and you can (for want a better term) "reverse spectrograph" it (in that the image light waves are converted to sound waves and the 2D width element of the image would be translated as time in the output audio so the wider the picture the longer the audio clip).
Needless to say, unless you're careful, the output is usually just noise. But many artists used this trick to place hidden images in their work
The specs are only slightly better than a netbook, but I wonder if the pluggable processors could provide some interesting opportunities? I had one of the first A1000s back in the day, but I think I'll give this a pass for now. I've got so much time and money invested in OS X and iPhone, that this is too little too late for me.....
"Capable of eight concurrent real-time threads with shared memory space, at up to 400 MIPS"
400 MIPS? A Pentium Pro from 1996 does 541 MIPS. http://en.wikipedia.org/wiki/Million_instructions_per_second#Timeli...
"Reference boards have been made with up to 256 cores, offering a theoretical 102400 MIPS."
A single Intel Core i7 965EE CPU does 76383 MIPS.
AHAHAHAH... Are you joking?
Define please a netbook hosting dualcore PPC powered at almost 1.5, or 1.75 GHz as AmigaOne X1000 (rumors from one of the developers).
A Dual Core PPC architecture with enough clock speed around 1.5 GHZ or 1.75 It is sure capable to outperform any Intel DualCore upto 2.0 GHz and to compete with Intel DualCore clocked at 2.2 2.4 GHz per core.
The PPC offers few architectural advantages over a modern Intel x86-64 chip an certainly any company producing this new PPC won't have the technology to implement their chip so that I could outperform intel on any level.
A dual core 1.5Ghz PPC is going to perform no better than the cheapest budget CPU from intel.
Running Logic Pro on my 1.5Ghz G4 and 2Ghz Athlon64 showed that the G4 was half as powerful as the Athlon64.
And mainly it will outperform direct opponents if it will be capable to run AmigaOS 4.x.
Infacts AmigaOS is truly sooo lower resource consuming that leaves almost the processor free of number-crunching.
And also this PowerPC fortunatley has not to deal with MacOS-X that contains fat-binary code (executables mixing PPC code with IntelX86 code) that in MacOS was really an iron ball tied to the legs of pooor innocent PowerPC processor.
If Apple had had leaved out IntelX86 code from MacOS-X fat-binaries and had made two different versions of its Mac-OS X, one optimized for intel, and the other for PowerPCs, then sure PPCs had doubled their performance, and had had being still usable nowadays with a PPC native Snow Leopard version.
I wonder why thuis anomlaous fat-binary solution.
The OSX fat binary solution requires more HD space, but only one chunk is loaded into memory... The chunk with the correct instruction set... The PPC version of OSX only loads the code relavant to it, it has no effect on speed of execution...
It seems to me that Apple made all in its possibilities to let Macintosh users to believe that PowerPCs were slow in performance by keeping deliberately them slow with fat-binary solution.
The PPC couldn't get enough money behind it to keep development equal to Intel and AMD, the CPU architecture offers few advantages over the x86-64 and has not had the same level of implementation development work done it... No PPC solution got offer the same price/performance ratio as x86-64... Apple were wise to get off that platform!
Edited 2010-01-05 16:36 UTC
As long as ancient G4 have Front Side Bus stuck just at 266MHz it does not surprise me.
G4 had suffered for any data transfer waiting for raw data to be input and elaborated data to be output.
BTW Was not Athlon64 FSB running at 800MHz?
Modern PPCs have more performant FSB running at decent speed so they do not suffer anymore of bottlenecks in data transferring with their peripherals and RAM.
Edited 2010-01-05 17:30 UTC
The Athlon64 has an integrated memory controller and as such lacks a conventional Front Side Bus. It's hypertransport interconnect does run at 800mhz, but this was mostly irrelevant for single socket performance.
While the PowerPC ISA is more elegant than the AMD64 ISA, comparing the G4 micro-architecture with the Athlon64, the G4 has a nice vector unit but loses in things such as maximum instructions in flight and issue width. So it is no big surprise the Athlon64 is faster. It should be expected to remain faster even if the G4 had a fast FSB.
Modern PPC chips are either too expensive and power hungry for desktop systems (POWER) or extremely specialised and largely unsuitable for general purpose computing (ie general purpose apps perform poorly) like the chips used in gaming consoles where the code is optimized specifically for the cpu.
The G5 chip doesn't perform very well compared to an Athlon64, they are only available in dual core form at up to 2.3ghz (maybe 2.5?) which can no way compete with modern 4 and 6 core chips from intel/amd.
Mac OS X never got slower on PPC after Universal Binaries were introduced.
In fact x86 and PPC code is entirely seperate, just glued together afterwards by GCC. It's easily possible to delete x86 code using either Monolingual or TrimTheFat. The only gain is saved disk space, not performance.
You can easily check by compiling some open source app as single-arch app and Universal app. There is no performance difference.
Actually, it didn't come back to haunt them at all: Apple switched the line over to "now our machines will be cheaper and faster", and the Apple fans pretty much bought it without a second thought.
I think, if Apple started claiming that chewing broken glass had restorative properties (It cures ailments of the foot! Try it!), then probably about half of current Mac users would at least try it.
MacOS 10.4 initially launched without fat binaries on PPC...
Later versions shipped with fat binaries for x86...
10.5, which was the first full fat binary version actually runs faster than 10.4 on PPC, having fat binaries did not halve performance from the previous versions of MacOS which were PPC only.
Also, you can strip the unwanted sections from fat binaries. Doing so saves you some disk space, but doesn't improve performance.
It's not fat binaries that slow down PPC, its the fact that the PPC chips being used by Apple were far slower than x86 chips... The only PPC chips with any level of performance these days are either extremely expensive (POWER) or very specialised (Cell).
A PPC native snow leopard would be perfectly usable, but why would Apple spend so much effort on an architecture which they have fully migrated away from?
Claiming that a 1.75GHz PPC will compete with a 2.4GHz x86 doesn't mean much, modern x86 chips are available at considerably higher speeds than that, quad and even 6 core are widely available too.
Xena sounds like a solution in search of a problem.
If you need more processing power, why not use a better CPU (perhaps one with more cores) and create more threads instead of bothering with the pains of an heterogenous architecture?
Unless you need to do some massively parallel processing in which case what you want to use is GPGPU.
Anyway... Told you.
http://www.osnews.com/permalink?402070
Because its awesome. That's why.
Why is it that every time someone does something a little bit different, people start going on about how they should just use well established technologies. Pretty sure we'd still be using upgraded IBM 604 calculator machines if more people thought like that.
By the flip side, with computing technology, for every innovation that catches on, there's probably a hundred brilliant little innovations... that are solutions in search of problems, that don't catch on and fall flat. While it's good to have an open mind, it's also good to be a little discerning, and to wait for some real-world performance data and use cases before you start saying how awesome something is, and all the wonderful changes that it will bring.
Heck, why not use standard PC parts and a Core i7?
Because that wouldn't be an Amiga, that's why.
Before computers got all trendy and serious bussiness, geeks used them to make all kinds of crazy stuff that made no sense but was awesome.
I'd really like one of this things. Just for the sh!ts and giggles 
As long as Amigas were considered by the public just as game machines, they just weren't.
And you cant use Wii as desktop computer.
It is just a game console, so figure out you are really Mario and continue your tireless joking as Janitor on Wii. We are talking of computers here.
But be careful that nigga could stole your Yoshi and your Wii too...
The original Amiga, hell no. Hardware-wise it was years ahead of the competition, back in its time of course. This one is a clear wink to die-hard Amiga fans that want a modern platform for Amiga OS with a few interesting hardware bits that aren't common to other platforms.
I'm not talking about then, I'm talking about now.
Edited 2010-01-05 19:18 UTC
My most favorite gaming system has only half that MHz (the Super Nintendo). That doesn't mean that I want to edit photos and convert videos with it.
I don't knew any serious use of Super Nintendo, did you?
While the Amiga revolutioned the world of computing with just 7 MHz CPU.
It was a matter of facts.
If you are nostalgic of 3,5 MHz Super Nimbimbo then aim at your garage and search for the box in which you left your Nintendo getting dust.
I don't have a cat. Never had.
I'm bored and on top of that I had lots of fun making Commodore and Amiga jokes with friends when Escom http://en.wikipedia.org/wiki/Escom_%28computer_corp%29 owned the brands. I still have some shitty Commodore-branded CD tower somewhere!
Well, one can always troll when one is wrong.
You can say "I told you so" as much as you want to, but that doesn't make you right.
No matter what they would come up with, you would say "I told you so". If they came up with a port to x86-64, you'd say "I told you so". If they added multi-user support to that, and memory protection, .Net/mono, Java, Wine, NTFS read/write, coffee-making capabilities (which they did, sort of), and found a way to turn lead into gold, you'd still say "I told you so".
Why?
Because it it reconfigurable, and extreamly power efficient.
You want AES encryption? Ok, here you have 1 unit with pipelineing. Do you need more parallel AES encryption? You want, 2, or maybe 10 units? Why not. Which one, 128 or 256 bit? It will reconfigure to this you really need, saving some gates.
Maybe you want other cryptographical instructions? MD5, SHA1, RSA multiplication?
Or maybe FFT? Audio filtering? AI in silico? Own fast floating point precission? Or reconfigure it to perform raytracing in realtime? Financial computations? N-body problems? parallel Decoding/encoding of mp4 to ogg and flv? Build your own specialized pipeline. Etc, etc.
Sure you can perform many of this on CUDA with very big performance gains. But, graphic cards adds lots of latency, and are very power consuming.
Posibilities are endless. For sure there are some creative pepole which will use this power to amaze us.
So don't ask "why?".
The Xorro slot can host an accelerator card with many Xcores (which are very cheap by the way) and since the Xcore works like a transputer from a scalability point of view, you could build huge clusters with proportional speed increase without changing the software.
The news mentions that accelerator cards with many Xcores are major candidates for the Xorro slot.
Time will tell, but this is certainly fresher than other computer's mobos of late.
The computing market needs to get back to the 80s where you had many engineers working on different HW platforms instead of the homogenized BS we get today.
Besides the dual core PPC is but a fraction of this computer features (titpical x86 fan argument: CPU this, CPU that), on net books, or iBooks for that matter, the CPU and an ultra weak graphics processor is all there is to it.
You could build a Xorro Card for ultra HD Video editing (ie:iMax 3D) with real time effects on this thing (using a cluster of Xcores on the Xorro slot), of course, you can do the same on a net book ;-)
Edited 2010-01-05 16:46 UTC
You can just afford it in 1 year and a half if you decide to stop smoking (at least if you are smoke addicted) as a packet of cigarettes actually costs about 4 euro.
>>The specs are only slightly better than a netbook
You don't know which type of processor it's gonna have....oh and try to expand your netbook to 102400 MIPS ;-)
This sounds like one hell of a machine! - Let it run AmigaOS and you're in for some serious speed in application/screens handling!
I already droooool at the bare thought
My preeeciooous!!
A netbook is a fraction of the size, completely different class of machine. That said, a netbook will more than compete with one of these machines in it's default config.
If i want a desktop system with thousands of mips, ohh maybe i'd go for a Core i7, which running at 3.4ghz can aparrently achieve 106182 MIPS (source: http://www.tomshardware.co.uk/forum/page-262886_10_0.html)
If i wanted an even faster system, i could go for a desktop system with 2 of those processors and almost double the performance.
This would be a standard, well understood system that doesn't require any custom programming to make use of.
The XCore chip used in the AmigaOne X1000 is the cheap 8 threaded model that sells on DigiKey.com for $5.60 USD a piece in less than 1000 unit quantities.
Add to that the toolchain is LLVM-based so it will be easy to port code to and from it for use on other similar designs.
Excellent bang-for-the-buck ratio.
RE[6]: $5.60 chip in 1000 unit quantities
How do YOU explain that Im sitting typing this on a AMD PC 2,8 ghz and have about 5-6 more PCs of different types around the house, and actually enjoys my newoldstock Amiga1200 with a measly 40 mhz 68030 and 18 meg ram the most ?
Sure I use the PCs for day to day work... But the Amiga gives me vastly more REAL fun ... !
Morb.
What are you talking about? Is it a crime that we happend to like AmigaOS? We are not saying that Xena is the best thing since sliced bread,it just happends to be a part of the MB,nothing else. All i care about is that we got more powerful hardware than before. And for your information,i do have a lot of computers in my home,and i am running different OS's on all of them.
And i do have Windows 7 on my i7 975(6GB DD3,and a ATI Radeon HD5970) My Debian distro runs on some older AMD 4800+. Most Amigans today own some kind of PC,MAC,or Linux box,so saying that we are not in touch with reality,is just stupid(yes,i do think you are a narrowminded,stupid clown) We just happens to like AmigaOS,that's all!
I guess the fact that AmigaOS is PowerPC and m68k code doesn't compute for you, does it? I assume you port operating systems across architectures on a weekly basis?
You're quite arrogant. I'm not even an Amigan and even I find that you are simply in the wrong place. If you dislike alternative software and hardware, then please leave OSNews. You are ruining it for the rest of us, and please do remember who owns this place.
you just forget one simple fact:
most of the amiga enthusiasts DO have a second or third computer at home, that is NOT an amiga.
and then it is mostly a pc, linux or mac box.
so most of the people DO know what it is, to run such a system. they still prefer and love their amiga.
Edited 2010-01-06 18:26 UTC
Nope... Simply we do not have seen any AmigaOS running on Intel processors. When we will see one version running on this architecture system based on Intel, then we could decide to buy even Intel hardware.
As long as AmigaOS remains on PPC architecture, it is normal that we Amigans pretend at least to have latest version of PPC running, with enough GHz, enough cores and enough FSB speed to avoid any bottlenecks.
You also never enjoyed the power of fully utilizing modern GPUs because you yet never had the hardware/drivers/whatever to use shaders on them. You therefore have never enjoyed the comfort and vast power that modern CPUs and GPUs offer (the later arguably being the most exciting and interesting chips that ever existed in any personal computer up to now, original amiga included)
Yes we had. Many of us have Intel PCe at work or at home with enough horsepower.
Personally I have Pegasos II G3 600 MHz and Pentium IV 3 Ghz socket 775 both at home.
And I think that if I could run AmigaOS on top of Pentium IV, it could had given me more performance than to run mindless Windows chaotical Operating System, In which there is an enormous caudron called C:\Windows (C:\Windows\System and System32 directories in which all things are installed and mixed up as in a soup, and sure it is a very crazy and ugy mess mindless solution Microsft had adopted...)
And also I think an AmigaOS Intel based had made my pentium IV more performant than any Desktop oriented Linux Distro and far better more friendly and responsive to final user.
But actually we must deal with PPC based AmigaOS. Point.
Yet you're welcoming the arrival of Xena, an obscure chip, probably much underpowered and of dubious interest in the context of a desktop machine as some kind of daring innovation. How do you explain such excitement towards that xena chip of dubious interest when much more interesting hardware is already commonly found in pretty much every proper desktop machine?
it is the fact that this Xena thingie is a processor aimed at high speed I/O and with extreme abilitiy in dealing lightspeed muxing/demuxing big amounts of data stream, at least as stated from PDF documents online and home site of the manufacturer.
This makes us very excited, beacuse it is a very interesting complement processor being added to Amiga that OTHERWISE could being perceived just as quite classic hardware architecture, with no difference from any normal PC except for the processor adopted on it.
If this processor keeps its promises sure it is a really amazing things to have one in an Amiga environment, and sure for the first time Amiga will be again on top of innovation, by giving for the first time on the market a moderboard based on this innovative solution.
it is sure new and something still unseen on any other Wintel or Mac architecture (nor PPC neither Intel)...
I have seen only some obscure evaluation motherboards spotting something like Hi-Speed higly configurable multi-purpose co-processors with high nubmner of I/O Channels.
These are barely supported by experimental Linux drivers and no one had used it for the common market, neither in any commercial OS.
Amiga had a great cofidence in the use of co-processors, keeping Amiga experience far more btter than any other platform in 1990 age, and amigan devlopers know ho to make very good use of such a XENA processor and tie and improve capabilities of this new co-processor with the long awaited availability into Amiga of 3D Extreme 16X PCI-e GPU based graphic cards.
If this solution will work flawlessly freeing the central CPU to deal with high speed Data streams (the future of HD-TV and Internet communications of tomorrow) then Amiga will lead again lead the innovation in computer world, and teach the path to any other system by showing with its performances, reliability and low consumption of resources, that the future architectures should integrate high speed multipurpose configurable multi I/O channel chips on any of the motherboards of next generation machines whatever it will be PPC based, Intel based, or any kind of future proceesor it could be seen in the market.
Also if you will be more open-minded you will see that without being backed by a big firm such as it was Commodore, the Amiga community and the little firms producing hardware for Amiga, had made a completely new and innovative product.
They just not limited them in filling the gap with mainstream Intel solutions in hardware, and making just an obscure machine featuring nowadays common SATA, USB 2.0 and PCI-e.
No... They made something really innovative.
And you must admit it is very interesting and amazing news that such a little platform believed dead as Dodo since 16 years from the demise of mother firm Commodore is still alive and capable of such an innovation with very few amount of money than big $$$ hardware firms.
These are the facts that made us amigans so excited!
Edited 2010-01-06 19:04 UTC
While on the desktop Linux may have a small (but growing) market share, it does have the advantage in the networking world to all other O.S's ... Google's O.S can only increase the desktop share..
So I will go back to my Linux desktop thank you, it works great and most (non-games) software is released from this platform - unlike for the Amiga with it's 0.00000000000002 % desktop market share
btw : I love the amiga and always have.
The reason I love Linux is due to its similarities with the Amiga..
The reasons I hate Linux it is beacuse it is completely different.
Geeky, not-responsive desktop-GUIs, system for directories made for geeks only, installation and configuration may vary in quality and reliability thru various distros, by spotting some that are really low quality, medium quality and high quality in the well known public distros.
If it fails something you must find in manuals what of the dozens (literally dozens) script files that require Linux has the fault text string that must be edited by hand.
Hi people, you should read the article
It says that developers are currently using a 1.6Ghz clocked version of the cpu. This doesn't mean that the final version will be 1.6Ghz. I think they will offer a range of cpus, from 1.6Ghz to over 3Ghz ( power can do that too ).
Also you must consider the power of the xmos. 2010 will be a very interesting year 
Even without adding a cluster of Xcores, this thing has a PCI-E 16X slot that could hold, say, a Radeon HD5800.
Now I have tryed Devil may Cry 4, Street Fighter IV and Grid (Race Driver) on a entry level 2.1Ghz intel dual core + Radeon 4830 (quite entry level nowadays) and the above games did run at NATIVE (not upscaled) 1920x1080 with details set at max + 4xAA and 8x Anisotropic Filtering (8x-16X on Street Fighter IV).
X-Box 360 with its 3 3.2Ghz cores can only dream of those settings (talk about as the CPU alone doesn't mean anything nowadays).
By the same token an AmigaOne X1000 + powerful RadeonHD card would run circles around an X-Box 360 and as i said, that is even without taking advantage of its biggest innovation: the use of transputer like cluster computing via the Xorro slot.
Whose firmware needs to be modified to be PowerPC-compatible. "
Not necessarily. EFI is designed around a virtual machine, so in theory any EFI capable firmware should work on any EFI based platform. The original intention was to produce a single firmware for both IA32 & IA64, but in theory (and I say that without even investigating this at all) you could have a PPC EFI firmware that could work with standard EFI capable devices.
RE[3]: Less powerful than a Xbox360...
They don't. And stop trolling.
What the Amiga-fans usually find disgusting is the way standard PC parts are often used (e.g. non-coherent compared with platforms where parts are designed to fit together perfectly). And of course the x86 instruction set.
Personally I find the x1000 strangely intriguing, though I doubt the price will be reasonable, considering the price of the sam440ep.
RE[5]: Less powerful than a Xbox360...
What the hell is wrong with you? Are you forced to read Amiga news articles or talk to Amiga fans? NO. f--k off. Dear god. I've never even touched an Amiga in real life and I'm still interested in this. I'm interested in it because its something new that isn't a bog standard x86 machine.
And, no, Amiga users do not hate commodity hardware. Go look at all the upgrades/mods to classic Amigas that involve commodity hardware.
RadeonHD drivers (at least up the Radeon HD 4000 series) are being worked at HDrlabs and should be ready by the time this thing comes out, Still enough to run circles around an XBox 360...(but again that is only the tip of the iceberg, the X-Cores could act similarly to Intel Larrabee's concept and the amount of expandibility is quite endless there).
Who said it was 32-bit only. I think I recall somebody who works for Hyperion saying that there was quite a workaround to get the OS to not assume 32-bit values for pointers. That suggests that the new version of AmigaOS will be more than 32-bit.
The DSP will be 32-bit but not the main CPU.
"It comes with a dual-core PowerPC processor, conforming to the Power ISA 2.04 standard."
Compliant cores ( http://en.wikipedia.org/wiki/Power_Architecture#Power_ISA_v.2.04 ):
# The PA6T core from P.A. Semi
# Titan from AMCC
P.A. Semi was bought by Apple and does not ship any PPC CPUs any longer.
And "Titan is a family of 32-bit Power Architecture-based microprocessors designed by Applied Micro Circuits Corporation (AMCC)."
http://en.wikipedia.org/wiki/Titan_%28microprocessor%29
--Taken from A-Eon.com
Notes: "...and newer standards.", "...variety of CPUs..."
OK, then let's look which CPUs support newer PowerISA standards:
POWER6, POWER7, and e500mc.
I seriously doubt that a high-end server POWER6 or 7 CPU will be build into an ATX computer, not to mention that POWER6 starts at 3.6GHz, not 1.x.
That leaves the e500mc which "is a 32-bit Power Architecture-based microprocessor core from Freescale Semiconductor". http://en.wikipedia.org/wiki/PowerPC_e500
So yes, 32-bit.
Edited 2010-01-05 21:37 UTC
TheDax your naughty
your missing the first bullet point....
it also supports Cores that comply with the Power ISA v.2.03 stadards.
And these include:
Freescale Power PC
IBM PowerPC Power5 & Power6
IBM Cell which is in effect a 64bit chip
And for those of you talking about GHZ these processors range from between 3.2 to 5GHZ.
Another point to make is that one of the guys involved with A-eon is Ben Hermans, who used to be CEO/Director of Hyperion Entertainment, and he has always been a big fan of the Cell processor.
Edited 2010-01-06 00:17 UTC
HDRLab is working in the 2D part of Radeon x500+ support under OS 4. Although a massive achievement (and may very well be ready when this X1000 thing ships), this is still very far from having any sort of modern 3D support in the Amiga platforms.
Most likely, you'll never see any modern, competitive 3D hardware working in Amiga.
These customizable processors are really the future, I think. The most impressive effort I've seen is Microsoft's eMIPS project (http://research.microsoft.com/en-us/projects/emips/default.aspx), which is dynamically customizable - a running program can specify its own processor configuration.
Being able to add custom logic to a chip means programmers and compilers aren't limited to a very general set of instructions, but can instead make their own instruction set specifically tailored for the task at hand.
XCore is not a customizable processor. It's just an 8-threaded embedded DSP with 8k of ROM and 64k of shared local-store SRAM built into it.
https://www.xmos.com/products/xs1-l-family/l1lq64
Edited 2010-01-05 19:48 UTC
From that website:
An energy-efficient, general-purpose device ideally suited to a wide range of embedded applications and systems including audio DSP, USB peripherals, networked LED displays and robotic motor control. Integrate any of these functions without compromise to cost and power requirements.
It is customizable. It is a processor. And as such is a customizable processor. This is not invalidated by the fact it is optimized for digital signal processing (a poorly chosen word, since any treatment of data processing in a digital computer (and any kind of processing of a digital signal) qualifies as digital signal processing).
general-purpose != customizable
General purpose simply means it is suitable for a wide range of tasks, kinda like a regular CPU. PPC and x86 chips are general purpose, but not customizable after manufacture.
Customizable generally means either a FPGA, or the design is customized prior to manufacturing.
Nobody claims that Xena is a FPGA. It is clearly stated as a "programmable CPU". Programmable also means customizable, though not necessarily at hardware level as you claim it has to.
It is the functionality/usage that is customizable, and not the physical layout. It is essentially a general-purpose co-processor, rather than a specialized co-processor.
You'd known that if you had read what Thom wrote, and if you had read what it says at the a-eon website 
Hardware like this is already in processors like Tilera (http://www.tilera.com/) and the Zii StemCell (http://www.ziilabs.com/). The latest Zii device I read of has an array of 64 programmable floating point elements (http://www.linuxfordevices.com/c/a/News/ZiiLabs-Zii-SiVo-Digital-Ho...). Personally, as someone who BOINC's, I'd be interested to see what one of the Zii StemCell co-processors could do for something like the docking modeling programs used by different projects. (For example, the protease inhibitors for HIV [FightAIDS@Home] or dengue fever / West Nile virus [Discover Dengue Drugs], or for receptor blockers [Fight Childhood Cancer].) If this Amiga reincarnation makes heterogeneous co-processors on desktop systems look appealing for serious design, then I think the whole Aeon project will be well worth it. One group tried with an IBM Cell BBE card to accelerate video encoding, but the price of that was... "prohibitive".
Would I buy a Zii / XMOS PCIe card to make BOINC projects run faster? If I had the money, oh, yes.
Let's take a look at this XCore processor. It appears to be a very weak processor (400 MIPS) without floating point capability. It's instruction set appears more dense than MIPS, but not so dense as ARM. Probably it has computer power comparable to an ARM920, which is not a lot, so forget about powerfull accelerators or doing GPGPU things on it.
The nice thing about this Xcore processor is that it has hardware mutltithreading with Quality of Service. This makes the XCore awfully suitable for realtime applications.
Todays computers are extremely weak at real time tasks. There are a lot of chips in todays PC's that have the task to buffer data until the CPU has time to process them. For example, while a Commodore 64 can do RS232 rather easy in software, a PC needs UARTs to buffer the data until the CPU has time. Basically all I/O controllers in current PC's, from ethernet, SATA, USB whatever, they just buffer data until the CPU has time to process it.
I/O would be one strength of this chip, it could process data on many wires in realtime without the need for a complex I/O controller.
Another use would be in emulators: Current emulators of C64, Amiga, whatever, work asynchronously: The emulated video controller first generates frames, and when the time is right those are shown. A chip like this could do this in real time.
Sound is another topic that would be very well suited for this chip. Wavetable synthesis is almost impossible to do right in software, because you need large buffers to prevent sound interruptions, while you want to have those buffers as small as possible to respond quickly to sound events (if you push the trigger on your joystick, you want to hear the explosion immedeately, not half a second later). This Xcore would be capable to do software wavetable synthesis just as well as in hardware.
So looking at the big picture: You get very little extra compute power, but this Xcore solves something todays computers are not good at. In fact, while the classic Commodore Amigas didn't have a proper solution, they were quite good at realtime tasks.
I think Hyperion made an interresting choice, rather than focussing on raw computer power they are targetting an area that prevent modern computers from giving the "Amiga experience".
Wether they will succeed is another story, but for anyone interrested in computer architecture, this is at least an interresting development.
Old hardware is tied to TV refresh rate. You'll need extra sync/output step anyway.
Lets launch 2-3 emulators or emulator and something what utilizes this hardware. They need to virtualize/ multitask the chip. That means limited direct access and/or complicated job/task schemes.
This is OS problem only. ~5ms sound buffers are good enough and should not cause any extra CPU load.
It doesn't have floating point. Dealing with fixed point quirks/precision problems is an 80s style of doing things.
For Amiga now it is better to focus on software, but not niche exotic hardware.
This is OS problem only. ~5ms sound buffers are good enough and should not cause any extra CPU load.
"
Are you denying this device will help with solving this issue?
I consider it a flaw of modern hardware that can at the best be reasonably fixed, a perfect fix on the OS level is hardly possible. There aren't many operating systems that'll get you a 100% guaranteed 5ms switching latency. For example Linux has a context switch granularity of 4ms with HZ=250, which is usually the default nowadays. One scheduler choice to anything else than the sound mixer and you have a buffer underflow... Configure some app to a 5ms buffer and try dragging some windows and you'll hear the effect.
Also, seems you have never experience real hardware wavetable as in Gravis Ultrasound..., you would talk different how much better hardware implementation is. It's not easy to describe the experience, but it is a plus that you always get the high quality hardware implementation rather than mixed quality software implementation, and it is a plus that you have no latency at all while never ever hearing a sound interruption.
While never have been an actual Amiga user, I own an Amiga 500 and the experience of a GUS and an Amiga regarding sound is very similar... I think it's this feeling that many users are missing on current computers.
Wether this device can really make a huge difference, I can't tell. It's just an interresting architectural choice, and Hyperion have to prove its usefullness.
Edited 2010-01-09 21:34 UTC
Everybody seems to get totally fixated on the XCore chip. It is just an auxilliary hardware resource which will hopefully differentiate X1000 from more generic, mainstream systems. This is good, I believe the market actually needs something like that, it will generate a lot of interest and is a good selling point.
However the yet to be announced PPC CPU (or most likely Soc, judging from the suspicious lack of chips on the mobo) and the appropriate PCIe gfx card are the main performance setters here.
Which reminds me... when Amiga came out in 1985 its 1979 CPU was really not all that revolutionary any more. MC68000 running at 7Mhz was seen more of a popular choice back then. A solid, affordable chip that still managed to outperform most x86 processors of the time but nothing that special or unique.
It was that unique blend of custom hardware and software that gave the original Amiga the edge. Perhaps, just perhaps a slightly oddball hardware of X1000 may do it once again, but it all depends on what people will do with it.
You don't really need a revolutionary hardware to do something amazing - you just need to do amazing things with it!
Edited 2010-01-05 20:58 UTC
Amazing!
I will buy this new AmigaOne X1000.
Im not Happy with WinOS X Crap, MacOS X Crap,
LinuXOS Crap, QNX CRAP, HAIKU X CRAP and bla bla bla all the others shitOS out there! Ive try them all and they all are CRAPS.....
I WANT TO BELIEVE! AMIGA HAVE THE TRUE X and will always have!!!
What is the true X? 
You're right! Who needs 64-bit, memory protection, IPv6, modern software, advanced 3D graphics, Blu-Ray, or powerful multi-core hardware?
Just imagine how fast those 20 year old Amiga programs will run, not to mention all 10 or so of the new Amiga programs will will come out in the next couple of years!
After all, there is no more noble use of a computer than nostalgia!
Well, I do. Oh well, not the Blu-ray part, but the rest of it is essential. Oh wait... the question was rhetorical... I see :p
The co-processor could be useful in turning your Amiga X1000 into a 3D-capable coffee maker - this way you could automagically have coffee ready by the press on a button - all while you're playing Quake III (or whatever you are doing). Not bad at all :p
Hum. Seems to me your attempt at trolling is more pitiful than irrirtating, as you are no even informed about Amiga availability of software and commodities, but mainly for the fact you are unaware that Amiga world is slowly advancing but steady evolving.
- 64 BIT: Actually 64 bit is already supported in AmigaDOS (certainly is in the DOS of MorphOS), and kernel64 seems to be work in progress as stated by some innuendo "preview" rumors coming directly from Frieden btothers, the programmers of AmigaOS kernel.
- Memory protection: Memory protection is already partially implemented, and it still remains partially to make elder AmigaOS application still to run in actual Amiga hardware.
It will be fully implemented the moment old AmigaOS legacy will be deprecated.
Let us make things step by step, at least we have few money and few programmers to make giant steps like as in very supported and very $$$ rich Operating System development teams.
- IVP6: Hummm... Let me think...
Was not IVP6 implemented in MorphOS Internet TCP-IP Stack?
It seems to me it is IVP6 ready...
- Modern software: We already have all modern software we need at the moment , for example the latest version of mPlayer, Blender, Python language, our own Internet browser called OWB (based on Safari Webkit), mail programs, our own Jabberwocky, MSN (SabreMSN) and Twitter programs, etcetera, etcetera, etcetera.
Forgive poor Amiga Software for not using same FAMOUS AND WELL KNOWN names of software of very famous $$$ brand $$$ Operating System.
But I know that it is diffcult for the only neuron in your brain to understand that an Amiga software runs the same as software in other OSes even if is not called "INSERT_NAME_OF_FAMOUS_SOFTWARE_HERE".
Amiga software can compete flawlessly with famous software in other Operating Systems, but just the fact it has a different name does not prevent you being a fool of yourself by not knowing anything of Amiga software scene.
(And perhaps you look just a fool of yourself by being intolerant and desperately trying of get joke of people thinking different from you)
- 3D graphics: We have BLENDER running on Amiga that is a very superb Open Source piece of software, it is enough for any usage.
More software like Lightwave will sure return on Amiga as soon as it will be available new AX1000 motherboards and Accelerated 16x PCIe graphic cards.
We lacked modern hardware on which to run modern accelerated 3D grapahci cards. Now we have it.
So what?
- Blu Ray: Actually I use it as standalone tied to my HD-Ready LCD TV.
I found Blu Ray readers for PCs still too much buggy and they not perform the same as dedicated stand alone Blu Ray Set boxes.
I read in some forums that some recent new Blue Ray readers are now out of any defects that I found previously in other readers actually on the market.
So I will buy one and perhaps I will connect it in my next brand new AX1000.
Time will see.
- Powerful multi-core hardware: TOday we achieved dual core, tomorrow we will achieve multi-core.
Amigans learned how to advance in little steps, and learned to be patient.
At least Amiga manufactuer are very poor compared to big $$$ names manifacturing PeeCees, so we Amigans as any poor family we live on compromises and must save money by saving any cent.
But savings hundreds of cents make real money and then we will buy the facilities we need.
There is no any nostalgia in being poor and little (not in real life neither in a big market such as computer market), but just sacrifice and honest tireless labour in order to emerge!
Edited 2010-01-06 07:07 UTC
RE[3]: Comment by Plexus
Do not refer to wikipedia only. there are plenty of documents online.
q]Memory protection: Doesn't have it yet, but it's being worked on.
It has... Even Thom Holwerda of OSNews wrote an article about it when trying OS 4.1 on the road...
http://www.osnews.com/story/21826/sam440ep_AmigaOS_4_1/page4/
Software: Cool. You got web browser, and you got Blender, and a load of other opensource software that runs on everything. Whoop-di-doo.
It is the modern age Honey...
There are no more killer applications making an OS more amazing than others.
And there arte no more killer apps that are not available in other systems in less than 6 months... (just the rime enough to be ported).
Maybe perhaps some little pearls are still available in Amiga that aren't in other systems, for example the YAMM mailer or WookieChat, or the image viewer "Show Girls" in MorphOS that could circle around any other piece of software on any other operating System.
3d graphics: I meant 3d graphics acceleration. You say modern hardware is supported now. I've found no indication of anything later than RV280 running with 3d acceleration. That chip is 7 years old. Only OpenGL 1.4 is supported, enabling Dot3 bump mapping, and other similarly hip features.[7q]
Dunno what you what to point here.
Both AmigaOS and MorphOS got hardware acceleration on their Desktop GUIs.
AmigaOS Workbench has Cairo 2D composition engine tied with 3D graphic acceleration features, and morphOS has Ambient 4.0 that is fully 3D hardware accelerated.
Both system can disable this feature on the fly just by chooising a yesn/no button in preferences.
You can find lots of YOUTUBE movies showing Amiga accelated features.
It is impressing that machines with just only processor clocked at 600 MHz with AGP 3D cards Radeon 2560 could perform 3D dirty tricks and Desktop Zooming smoother than modern Intel 2.0 GHz dual cores.
Perhaps you are confusing the lack of accelerated features with fact hardware acceleration graphics in Amiga runs only on Radeon R1xx and R2xx family that are 7 years old.
It is a matter of lacking of drivers, not a matter of 3D acceleration feature missing.
But again without a modern moherboard in Amiga we could not mount new graphic cards cards neither developing any drivers to run these graphic cards.
Now there is new hardware to run PCI-e graphic cards and so we escaped the Snake-biting-its-own-tail problem, and it will be possible to develop new drivers for modern Graphic cards.
And there are also third part projects ongoing:
http://hdrlab.org.nz/radeonhd-driver/
" Blu-Ray: i don't care about blu-ray either. HD can be frightening. I agree with this guy: http://www.xkcd.com/598/ However, blu-ray means more than a drive capable of reading the format, but also playing protected video. Personally, I won't let that protected content shit anywhere near my system. I don't watch enough movies to care.
If you don't care of Blue Ray, why you ported it as exaple of diminishing of Amiga?
Was it a nice try in trolling?
Multi-Core: No, you don't have it yet. You have a product that's been announced. Like many other Amiga products announced in the past couple of years, it isn't on the market yet.
Actually MorphOS system Quark microkernel already supports multicore Simmetrical Multi Processing capabilities. It was programmed having already in mind this feature.
Unfortunately ABOX Amiga compatibility sandbox that runs into MorphoOS doesn't allow morphOS to perform multicore facility as main hardware drivers are hosted into it.
It is just time to develop better the QBOX part of the MorphOS and run the Amiga subsystem just as emulated and deprecate old legacy, so SMP could be featured in its full beauty.
[quote]
Little steps aren't enough when Linux, MacOS, and Windows are all taking big steps.
[/quote]
All I saw as big steps in Windows world is letting the OS booting faster.
it is just a dirty trick by booting the OS without loading processor stressing Win-Services that are loaded later, during normal desktop activity.
Apple already make its big step by deprecating its whole old MacOS and making EX-NOVO its whole nes OS... a Mix of Mach Kernel, FreeBSD, Darwin and a clone of former MacOS interface, and calling it MacOS-X.
There were no real big steps improvements since then.
Linux is currently trying to deprecate whole X11 graphical engine and rewritting it more small and performant in X-ORG to make at least became Linux performant as Amiga in Desktop activities. But still they need almost two or three years to cacth the Amiga tail and being finally a Desktop Operating System performant almost as Amiga is nowadays...
And perhaps main big steps from other OSes its just develop Desktops with 3D eyecandies.
You can call it REAL development...
It was just baroque trompe-l'oeil to make astonshing ingenuous users saying "OOOH..:" looking at 3D new graphic features in these Desktops GUI...
http://en.wikipedia.org/wiki/Trompe-l'%C5%93il
Really, this is what happens when you get emotionally attached to operating systems. "
Actually we are meotionally attached to an operating system that is really configurable by the user, obeys flawlessly the user commands making no hidden services dirty tricks (windows). It does not require a comuter science degree to be used in its deep as does Linux and does not threat the user as such a little child who must be stopped by its on Operating System Desktop barrers to dig the whole operating system as does MacOS.
Macintosh desktop it is like momma saying the child "NOT TOUCHY THERE YOU BIMBO USER.. IT IS DANGEROUS!"
In MacOS only those who dare to dig inner the OS, and bypass the desktop could choose to activate/deactivate all its hidden features and make new and interesting things, but again as being a UNIX-Like BSD based, you need to have a computer-science degree to unveil its real potential.
Amiga it is far better beacuse it is easy, not resource consuming, leaves processor free for heavy usage, it is easy inconfigurating and upgrading. It is maore than emotional approach.
It is just the best OS ever.
Whether this will ever materialize in real-life (or spread to the wild) is still to be seen.
Nevertheless, this is the first bit of really exciting hardware development I've read about in the last little while. Furthermore, there is an useable operating system available which can leverage it in ways not yet imagined.
And, by the way, a reality and success here might prompt Apple to return to the PPC! Half joking here yet this could happen if the capabilities of the combination leap-frog what can be obtained within the X86-64 environment.
What was the speed of the last PPC Mac? I believe it was a 64bit Dual Core G5 2.7Ghz, found in the last PowerMac, just before the switch to Intel.
So, we all know at what speed a G5 could run, and the heat it generated. The change to Intel was for the better, both for speed and heat.
So, this new Amiga Dual Core CPU, based on the PPC, lets hope it's faster than 1.6Ghz. Yeah yeah, I know that Amiga OS 4 is lighter than OS X, but Mac OS X included much more technologies not found in Amiga OS.
Most of all, let's hope that all this is not Vaporware. For the sake of Amiga Fans...
Hyperion didn't let us down so far. It's the only company that really stood up for Amiga,since the fall of Commodore. Sure there have been other companies that tried,but with no luck. It's not easy to rise from the "dead" when there is'nt money to back it up.
To those that critize the spec of AmigaOne X1000. Do you really think it's easy to just build the ultimate machine,and expect a community of not more than a few 1000 members pay 10th of thousand of $ for that kind of computer? If Hyperion and A-eon had the financial backing of the likes of Microsoft and Intel,and a userbase of millions,it would not be hard to come back into the limelight,but this is not the reality! AmigaOne X1000 is not the most powerful computer in the world,but it's a good enough computer so our community can see some growth.
I am infact amazed of what Hyperion,and partners has achived,with so little resources! And you should be too!
Exactly Caveman2.
You also have to remember that there is a move to "good enough computing" as witnessed by the sales of Netbooks. All recent computers can really accomplish anything any other recent computer can do, with only moderate differences in the time they take to do it.
For instance, I am finding my netbook surprisingly usable for things I used to rely solely on my desktop to do. Others might not, but I don't believe people are returning their netbooks to stores in their droves which some people(probably with vested interests) like to portray. For crying out loud, my eeePC netbook is happily running two instances of Visual Studio 2008 and SQL Server 2008 Management Studio, with Firefox, IE and Lunascape, along with a few of my .NET applications and the program I am debugging. And while I could probably notice a launch time of a second or so longer, actual usage to my Core 2 Desktop(which is by no means a speed demon) is negligible.
It really comes down to, how good is good enough. Yeah, the latest PC might rip an MP3 faster. Big freakin deal. I don't notice such things. I just want the thing to work the way I want it to. Amigans want the computer to work the way THEY want it to. You may want it to work another way, which is fine. There is enough room for everyone. It ain't all about GHz any more. It is about the user experience. Getting the speed up around the latest technology opens up a whole lot more possibilities, but it doesn't have to beat everything out there.
The big apps will come, and there are substitutes for practically everything. In terms of software development the Amiga is the wild west. Which makes it all the more fun.
I might be tempted to splurge and buy such a system provided that they sell more than 10,000 to others first and prove that the mfg's are in it for the long haul. After my A4000 Tower died for the last time I promised myself to stick with more ubiquitous hardware - but I do miss the Amiga OS, Linux can be so convoluted at times in the way it does things (does every application HAVE to open a socket to the internet? I don't think so!).
Look, rather than making pointless comparisons between this hardware and that -- with useless anecdotes about how this thing will never be faster than your waffle-maker or that it won't run XYZ, why can't we geeks just appreciate this for the cool, unique hardware that it has?
Where else can you get a new PPC-based desktop that was manufactured in the past 3 years? Dual-core? 8 threads? 20 watts? With dual PCIe ports? I'm betting this is using a PowerFicient core based on some slides I've seen and how that compares to what they've release or people have discovered.
As for the XCore, who cares that it's "only" 400 MIPS, "only" 32bits, or that its not an FPGA -- These Xcores, while they are general-purpose processors, are the new ASIC -- designed to be loaded with firmware that would have previously required custom silicon, and they're much cheaper and hacker friendly than FPGAs: Anyone that knows C can program them to do something different, no HDLs required. Had the original amiga had such a chip at its disposal back in the day, you can be sure they'd have opted for this instead of custom silicon. Check out the XCore exchange for lots of cool projects -- on of them impliments a SNES-like GPU running on only 2 of the possible 8 threads (apparently once you go beyond 4, you loose some response resolution) but that still leaves half the chip free -- and that's generating an NTSC signal in software -- you could do a lot better offloading that to a cheap encoder IC.
My only complaint, seeing as how this motherboard is probably going to have a not insignificant cost, is that it's only opting for the single-core XMos chip, when they sell 2 and 4-core versions for not much more in quantity, and that would be all kinds of fun to play with.
If you can't appreciate something so unique, and (by the sound of it) just plain fun, being too old and crotchety to forget more practical matters for just a little while, then maybe its time to turn in your geek cards 
If they really want grow the Amiga community they need to hit a $200 price point. Raw speed is not as important as perceived speed. An ARM11 is a $20 part, along with the XCore which is a $5 part, with these the new Amiga could make serious inroads. But an expensive PowerPC chip that pushes the price over $1000, the new Amiga will never grow beyond the usual cult following.
If you are THAT Rich, then invest in amiga just one or two million dollars for the design of the new motherboard and pay the Frieden brothers full fee to allow them working full speed to port AmigaOS to ARM processor.
Then in more or less than a year you will have the new computer and AmigaOS ported to ARM...
What? No money? So we will be stuck to PPC architecture.
Well, at least even if we are stuck to PPC architecture, now we have modern PCI-e BUS and modern set of peripherals, so at least we filled one of the three gaps that left Amiga behind the mainstream OSes.
Next gap to fill is to improve the AmigaOS.
The biggest problems in Amiga are the lack of a brand like Commodore backing it, and the lack of inverstors who gavemoney to develop Amiga.
When two months ago Hyperion was entitled by Amiga Inc. to have full rights in te usage of the Amiga name, brand and IP of AmigaOS 4.x, Then Amiga was again lead in the market by a firm who believe in Amiga and has right projects and a clear plain roadmap for it.
And sure it is more than expected.
I hope that the new solid management from Hyperion, the new hardware project, and the clear plains that Hyperion just demonstrated about Amiga and AmigaOS could attract new investors in Amiga.
Edited 2010-01-06 07:28 UTC
... think about power consumption?
Why does this board need to be faster than the fastest PC? The Intel Atom processor is totally useless? If this boards uses less than 20W, which is less than an Atom board, what computer can compete to this?
We already have fast computers. An Amiga would only need to be good at some area of computing to become attractive, and that area is not necessarily number crunching.
I also think another inroad to make with this is with the distributed community, particularly BOINC. Umm... "those" people are quite obsessive as well. I'm sure a SETI client will appear quickly, as SETI gets ported to *everything*. (OS/2? HURD? And old SGI workstation? Yep.) There are DC people who love getting this to run on obscure hardware. (See above. Einstein and SIMAP are also diverse.) If the price is reasonable, BOINC would give you and instant niche market. Also, if a hard science or life science app gets ported, or even a math app, that's good advertising. SIMAP is a project that catalogs protein homologs (proteins with similar amino acid strings, likely being a close evolutionary relation and similar function). As I understand it, it's light on floating point math. It's also been ported to PPC and SPARC clients. Something like that would be a good "Here's how the X1000 is helping the sciences" add.
PS: the fact that I chose my current video card based on price and how well it runs Folding @ Home GPU does not mean I'm obsessive.
A few comments (ok, one...):
First thing to understand is the difference between "general-purpose" and "domain-specific". Any chip made where the domain is known in advance can be designed to do in 1 instruction what general-purpose chips need 100 to do (to put it into perspective). Therefore, the 400 MIPS specs only make the processor "old-spec" if it were a general purpose such, but...
Since it is re-programmable, it in fact represents a paradigm shift: The processor can be rebuilt in software, meaning it can change its architecture to suite each specific domain, in effect making it "multi-domain-specific".
So, any comparison would come down to "oh...it can only do 400 million things a second", which I can certainly live with.
Also, these days it's more about performance unit/watt than absolute performance per hardware unit. If it can run a dual-core cpu with performance like, say, a dual-core Athlon, at 20 watts max, that's equivalent to the best quad-cores AMD has to offer.
Even the price/performance ratio, which I doubt will be very competitive (excluding the prospects of the coprocessor), aren't very important, since it will likely remain a niche-product.
And not everybody has to strive for world domination: Sell 100K units at $100 earnings and you have 10 mill. dollars, which you can run a smaller company on for a while (if you can live without the Learjet).
Since it is re-programmable
The XCore is a general purpose CPU and *not* reprogrammable. It's only special feature is that it can do multithreading in hardware ((context switch every clock cycle) with Quality of Service which makes it terrible usefull for real time tasks. Something which modern computers are very poor at.
Skimmed the spec docs and you're right -- my bad. The merits of "go to the source" instead of just reading the sum-up...
It seems it divides its clock cycles between each core as well, which somewhat lessens the usefulness of a multi-core design.
In other words, a general purpose CPU at a nominal 2.5 GHz clock frequency will have 1x nominal freq per core but the XCore only 1/n per core, where n is the number of cores.
By real-time tasks I suppose you refer to hard real-time. All general-purpose processors do soft real-time well enough -- well, actually, the operating systems do. You certainly are able to watch your videos and do your software-mixing of 128 audio tracks without hickups -- all inheritly soft real-time tasks (dead-line with escape strategy).
One could argue that all dead-lines are merely a matter of resource availability/use (excepting execution strategies that include prediction)...
An interesting property is the ability to quickly scale down the frequency automatically or as explicitly directed to 1/100th the nominal power, making 256 processors use 2.56x the power of 1 running full steam.
Try that with a modern C/GPU.
They're probably using an 8-stage pipeline in the processor. In this case switching between the threads on each clock cycle is used to avoid collisions between the data of the independent threads. It's a cheap trick but it gets the job done.
Adding cores in parallel only increases the number of I/O lines. To get added speed you have to add processor chips in parallel rather than using more cores per chip with this one. It's probably because the multicore chips have only one local-store bus to fetch from. Maybe it's a good thing that A-Eon is using the cheap chips.
It sounded like you were thinking that the XCore is an FPGA. I agree, an FPGA as a standard part of a desktop computer would be brilliant. I'd love to see hardware acceleration on-the-fly for more tasks. I would think that an FPGA would have been a natural choice, especially given the success of the MiniMig in emulating a classic Amiga.
My understanding is that the XCore divides time among it's threads, not cores. In this case, however, the point is moot because the XCore chip on this board is only a single-core unit.
Each core can handle up to 8 threads, but from other discussions elsewhere I've read, it can only look at/prioritize 4 threads on each cycle, so if you use the full 8 threads, then you handle 4 on one cycle, and the other 4 on the next, repeating. So yes, you do divide that 400 MIPS by the number of threads, but you can have 8 threads consuming 50 MIPS each (with minimum 2-cycle latency), 4 threads consuming 100 MIPS each (with minimum 1-cycle latency) or 1 thread consuming all 400 MIPS -- The split doesn't have to be even either, you could have 1 "hot" thread consuming most of the resources, and several "cold" threads consuming the remainder.
Each processor core, in the 2 and 4-core models, are their own domain though, with their own RAM and ROM, and their own MIPS to go around.
There are two things that are yet unknown about this platform that need to be addressed before any of us can really make a sensible choice in deciding if it is a good product or not:
1) How does the XMOS chip connect to the memory and I/O?
If it does not have sufficient supporting circuitry to give it fast and near-instantaneous access to the memory and I/O buses as well as interrupts and DMA channels then it will be of little use.
2) Are there XMOS development tools available for AmigaOS 4?
My gut feeling here indicates that the answer here might be "no".
About I/O XENA has its own indipendent interface slot that A-EON called Xorro in honour of ancient smart Amiga Zorro bus.
As stated in A-eon site:
http://www.a-eon.com/6.html
Thanks to Xena and the Xorro interface, the X1000 offers extraordinary flexibility. We believe that with this easy gateway to the world of 'Software Defined Silicon' and a path to massive parallelism, the X1000 will once more make the AmigaOS platform the best choice for truly creative and unique applications. For custom hardware control from robotics to theatrical lighting, for hobbyist creativity, for hardware hacking and for a multitude of applications we haven't even imagined yet, the X1000 is a dream platform - and therein lies another meaning of 'X', the unknown. It is you, not us, who will define the future.
AND
To accompany 'Xena', we have 'Xorro', a new slot using an industry-standard PCIe x8 form factor to give access to the 'Xena' IO. This will be the route to Xena's 64 IO lines, which are dynamically configurable as input, output, or bidirectional. 'Xorro' will allow bridging Xena to external hardware for control purposes, to internal systems, or to other Xcore processors. This last point is worth more exploration; XCore is a parallel processing architecture, and if you want more power, you can simply chain more XCores together. Reference boards have been made with up to 256 cores, offering a theoretical 102400 MIPS. Those of you interested in high-end imaging or scientific applications, for example, take note.
My gut feeling here indicates that the answer here might be "no".
The tools are open-source and written in C++. Furthermore they have LLVM underpinnings so porting code to XCore will be absolutely no problem.
See: https://www.xmos.com/technology/design-tools-source
At last we can see a little of what is to come to the world of Amiga. Lets hope that it is at a reasonable price that we all can afford. I know I would love to own a "modern" Amiga as an Amiga owner of 23 years. I still have a working A1000 and a A500 connected to a big screen TV and still love to power both of them up.
What gets me though are the negative comments on this thread, why do people have to be like this. Let people enjoy their computers whatever they may be. I hate the whole "mine is better than yours" crap that people dribble on with. I have many computers here from quad core machines running Linux to 8 bit machines like Vic 20's that all work. At the end of the day if people enjoy doing what they do then that is all that matters. I started using computers in the day of the Apple ][ and used a great many platforms and they all have the pluses and minuses whether it be a C64. Windows 7 or Linux. Enjoy it for what it is !
Personally what gets me is that so many people out there have only ever known a Windows/x86 environment yet they are experts on any thing and everything computer related.
As to where Amiga will head from here no one will really know until the future comes, but the Amiga community has lived through what seems to be bad dream of leading the world to almost obscurity (to the majority). So finally a ray of sunshine emerges and it is a great moment for this community, either embrace it or leave them alone!
Why not be happy for them, as an alternative in hardware is a great thing as it has inspired evolution or innovation. Innovation and excellence is what makes the whole industry move forward. If there is no competition on innovation what would we be using now?.
I agree. I owned an Amiga 500 back in the day, and in the end finally had to jump ship for OS/2 and Linux. It was a step back in terms of multimedia, only mitigated by the cheap CPUs and hardware. The faster processors could brute-force the effect of hardware acceleration... kinda. It wasn't as sophisticated. My interests went into C++ and MUD programming as a result; it's what it was easy to get gratifying results from.
Now, with MorphOS, AROS, Anubis and all that, the Amiga OS has lived on, though I've been wondering to what effect they do so. This announcement is cool. It actually has a claim to recreating what was so cool about the Amiga: innovative use of hardware acceleration. Having a nice, lean multitasking OS on top was just icing on that fancy cake.
I still think they need an FPGA in there as a standard! It works to recreate the A500 in the MiniMig, so the backwards compatibility could be greatly assisted - but then, you have an FPGA, and you could hardware-accelerate all kinds of things right on the fly. I hope they do it! If you have a general purpose CPU, a GPU, a grid of XCore processors, and an FPGA right there, you have a serious swiss army knife of hardware-accelerated computing that could do cutting edge computing. Come now, they already have LLVM to help leverage it all. That kind of spec would be worthy of the Amiga name.
Edited 2010-01-07 19:36 UTC
As long as it has a game line-up to match this.....
http://www.warblade.as/feature01.html
Everything else is imaterial 




