The OS or the applications? Almost anything works better than Windows on x86, and Linux on x86 is pretty damn fast for its purposes.
It's interesting that some of AMDs SPEC marks are different depending on the OS used, it's not a big difference (<10%) but this is on a test suite in which your OS will only take around 1% of the processing power.
No way. I've been wanting a CPU independent programming language ever since I saw how fast AMOS was compared to pure assembly
How did you measure this?
Good pure assembly is very difficult to write these days.
There are plenty of CPU independant languages out there BTW: Perl, Python, Squeak etc.
I base my conclusion that PowerPC is a bad idea not because it's technically inferior and x86 is just better, it's because x86 is a more stable market. Windows machines can't defect to PowerPC overnight, so you have to think about what 95% of the industry is going to do when x86 goes belly-up.
The problem is if you try to get into the x86 market with Eyetech or Genesi's volumes you'll have zero sales due to the price difference.
So, you drop the hardware and do an OS only. This puts you up against Microsoft:
Game over.
Intel may have used AMD's documentation to develop the EM64T for compatibility reasons but that doesn't mean they "adopted" anything.
Intel have their own 64 bit CPUs in the Itanium, they could have designed their own 64 bit extensions to the x86 ISA but didn't, they used AMDs instead.
Amiga is dead, there is no room for innovation any more and no possible way it would ever compete.
Sure there is, you just have the imagination to do it and know not to target the existing desktop market.
Read up on Sony's Cell architecture, there's nothing like that anywhere right now. It is truly revolutionary.
It's not just a new chip either, it's an entire dristributed parallel processing architecture for both software and hardware.
The industry was ripe for such a venture at that time. It is not now, not without billions of dollars available.
Depends what you build, nobody would build a completely custom system from scratch in this day an age.
All the leading players however are doing what Hi-Torro did, designing custom chips (for example, ATi, NVidia, Sony). Those that don't fall behind very quickly (Nintendo for example).
I wouldn't say ATI or Nvidia make custom chips as custom means it's done for one customer. ATI and Nvidia do commodity graphics parts but they do design them themselves. That wasn't so difficult in the A1000's days, these days an average custom chip costs $15,000,000 to develop - ATI & Nvidia's chips are costing something like $400,000,000 to develop.
Nintendo and Microsoft didn't even attempt to do their own chips, both are using modified parts from ATI.
Sony can do what they want and not only develop their own chips (actally co-develop with Toshiba and IBM) but are building their own fabs to make them in - at $2 billion each.
PPC's only hope is IBM, and will need Apple to help get PPC CPUs into computers in a wide enough scale.
What about Nintendo, Sony and Microsoft? They are all planning to use PowerPC.
BTW, to those who don't believe there's any innovation left go read "The future of computing" series I wrote:
links here - Warning, long!
My next article has a similar theme but describes how to build a new platform using technology which is either already available or will be soon. The idea is to combine multiple technologies in a single box to create something completely new.
The A1000 did pretty much the same, the custom chips were an evolution from Jay Miner's previous work at Atari, multitasking existed before, so did the GUI, so did the 68K. Nobody had put them all together before the A1000 and it took the rest of the industry years to catch up.
