5 ms·

Whilst this is interesting technology wise and may open impressive possibilities... so what? Without a solid install base why would I spend my time writing impr

Whilst this is interesting technology wise and may open impressive possibilities... so what? Without a solid install base why would I spend my time writing impressive software that is tied to this machine? Profit? No install base = no customers. Hubris? No install base = no one to impress. I'm fully aware this is a chicken and egg situation... (I say all of the above as a total Amiga fanboy from back in the day!)

Retric 17y ago
I think it's designed for the number crunchers out there. A GPU can crush a CPU for some applications but there is a huge middle ground between a GPU and a CPU.
- jteo 17y ago
  Then one would be better served buying a GPU from either Nvidia or ATI.
  - tsuraan 17y ago
    You would buy a GPU from AMD or nVidia to explore the middle ground between GPU and CPU? How does that work?
- vidarh 17y ago
  Why do you think it's designed for the number crunchers? I want one (I'm an ex-Amiga user, and still miss it...), but I certainly don't think a dual core PowerPC + a 400 MIPS co-processor is going to be cost/performance competitive with a suitable x86-64 based server at this stage. In fact, I'd be surprised if it was much faster than my current $600 laptop. The XCORE ("Xena") co-processor they've added is exciting, but because of the hard-realtime capabilities / low latency and IO lines, not for raw performance.
  - Retric 17y ago
    I was thinking of the people that crunch numbers not the programs. I don't think it's designed for a production environment; rather it's a dev box that lets you explore programmable processors. I think they avoided making an x86 with coprocessor because they want to own the dev environment for such things, but the innovation is all about that coprocessor. PS: 1x Xorro slot with a card that's 102400 MIPS is fast. An example of number crunching would be real time encoding of super? HD (4x 1080p). Or a custom high bandwidth router etc.
    - vidarh 17y ago
      I don't consider it a dev box as much as a box intended to satisfy Amiga hobbyists, some of which may use it for development, but a lot of which just want a non-x86 box running AmigaOS at decent speeds... Keep in mind that the people behind this are long time Amiga supporters - they're not in this for the sake of the XCore. For the people who are interested in the XCore, it's cheap to buy USB attached dev-board, so this only really matters if you want to play with a system that may have significantly tighter integration (we don't really know yet how tight). For the Amiga community, an x86 box would've been a non-starter. The people in the most likely buying segment for this box are people that have stuck to ancient "classic" Amigas, or at best upgraded to sub 1GHz single-core PowerPC boards running newer versions of AmigaOS, that they've paid more for than what you'd pay for a quad core 2GHz+ x86 box. Most of them would've been happy with "just" another PowerPC machine, but the XCore adds some extra excitement and may entice a few ex-Amigans like me to have a go for the fun too. > PS: 1x Xorro slot with a card that's 102400 MIPS is fast. An example of number crunching would be real time encoding of super? HD (4x 1080p). Or a custom high bandwidth router etc. The 102400 MIPS is an example based on a hypothetical card that doesn't exist, though. For comparison, XMOS themselves sells a roughly 25000 MIPS experimental board for $1500, so while you can make it fast, it'd also be fairly expensive and it does require a new product. I am excited about having the XCORE chip there, and I may buy an X1000 (who am I kidding, I will buy one unless the price is absolutely outrageous), but I am an ex-Amiga guy and would buy it in part because I still really love the whole Amiga experience, without any illusion that I couldn't do most of the same things on a PC much cheaper in most instances. When it comes to the XCORE what excites me about it, rather than performance, is that it will be standard. In other words, (the few) people who write Amiga software can soon reasonably rely on having it available in the computers of most users of their software. People are already thinking up all kinds of bizarre uses for it.
patrickgzill 17y ago
My take is that it is less for GPU-like number crunching and more for interfacing with real-world devices; perhaps writing relatively simple scripts in ARexx for working with serial, parallel, etc. devices.

Retric 17y ago

I think it's designed for the number crunchers out there. A GPU can crush a CPU for some applications but there is a huge middle ground between a GPU and a CPU.

jteo 17y ago
Then one would be better served buying a GPU from either Nvidia or ATI.
- tsuraan 17y ago
  You would buy a GPU from AMD or nVidia to explore the middle ground between GPU and CPU? How does that work?
vidarh 17y ago
Why do you think it's designed for the number crunchers? I want one (I'm an ex-Amiga user, and still miss it...), but I certainly don't think a dual core PowerPC + a 400 MIPS co-processor is going to be cost/performance competitive with a suitable x86-64 based server at this stage. In fact, I'd be surprised if it was much faster than my current $600 laptop. The XCORE ("Xena") co-processor they've added is exciting, but because of the hard-realtime capabilities / low latency and IO lines, not for raw performance.
- Retric 17y ago
  I was thinking of the people that crunch numbers not the programs. I don't think it's designed for a production environment; rather it's a dev box that lets you explore programmable processors. I think they avoided making an x86 with coprocessor because they want to own the dev environment for such things, but the innovation is all about that coprocessor. PS: 1x Xorro slot with a card that's 102400 MIPS is fast. An example of number crunching would be real time encoding of super? HD (4x 1080p). Or a custom high bandwidth router etc.
  - vidarh 17y ago
    I don't consider it a dev box as much as a box intended to satisfy Amiga hobbyists, some of which may use it for development, but a lot of which just want a non-x86 box running AmigaOS at decent speeds... Keep in mind that the people behind this are long time Amiga supporters - they're not in this for the sake of the XCore. For the people who are interested in the XCore, it's cheap to buy USB attached dev-board, so this only really matters if you want to play with a system that may have significantly tighter integration (we don't really know yet how tight). For the Amiga community, an x86 box would've been a non-starter. The people in the most likely buying segment for this box are people that have stuck to ancient "classic" Amigas, or at best upgraded to sub 1GHz single-core PowerPC boards running newer versions of AmigaOS, that they've paid more for than what you'd pay for a quad core 2GHz+ x86 box. Most of them would've been happy with "just" another PowerPC machine, but the XCore adds some extra excitement and may entice a few ex-Amigans like me to have a go for the fun too. > PS: 1x Xorro slot with a card that's 102400 MIPS is fast. An example of number crunching would be real time encoding of super? HD (4x 1080p). Or a custom high bandwidth router etc. The 102400 MIPS is an example based on a hypothetical card that doesn't exist, though. For comparison, XMOS themselves sells a roughly 25000 MIPS experimental board for $1500, so while you can make it fast, it'd also be fairly expensive and it does require a new product. I am excited about having the XCORE chip there, and I may buy an X1000 (who am I kidding, I will buy one unless the price is absolutely outrageous), but I am an ex-Amiga guy and would buy it in part because I still really love the whole Amiga experience, without any illusion that I couldn't do most of the same things on a PC much cheaper in most instances. When it comes to the XCORE what excites me about it, rather than performance, is that it will be standard. In other words, (the few) people who write Amiga software can soon reasonably rely on having it available in the computers of most users of their software. People are already thinking up all kinds of bizarre uses for it.

patrickgzill 17y ago

My take is that it is less for GPU-like number crunching and more for interfacing with real-world devices; perhaps writing relatively simple scripts in ARexx for working with serial, parallel, etc. devices.