‘Silly-SMP’ is a project to determine “What are the minimal changes needed to AROS to support ‘full’ SMP? Is it even possible?”
[…]
This is Research with a capital ‘R’.
But that ‘25% of the time’ _does_ show that a full SMP system on AROS is possible.
This is quite a big deal for the Amiga world – and proves that, in my view, AROS has more of a future than AmigaOS will ever have: it’s portable, and it’s open source, so experiments like this are more likely to happen.
This is not “Research,” it’s yet another instance of “Rediscovering” the wheel.
That being said, people working on alternative OSes just for the hell of it is a cool thing IMO.
Linus Torvalds started in the same way.
He didn’t start with SMP, but it is still just development, or having fun as I would call it, but it is not research.
For the AmigaOS world this is research!
The AmigaOS is a strange beast, and many design decisions made in the early 80’s to implement preemptive multitasking, never took into account of more than one CPU. Big locks, shared OS structures and a very fine grained strict task priority scheduler are just some of the problems that need to be solved without breaking compatibility with older software
Fun, fun,fun!
This would bring AmigaOS experience to be able to fully utilize modern multiprocessor hardware.
I remember a long time ago, the IT department where I worked was doing an evaluation with a side-by-side comparison of Windows 3.11WFW, OS/2 Warp 3.0, and AmigaOS. The only reason AmigaOS was on this short list was that the VP in charge was using it at home and preferred it over the office stuff! Anyways, the winner was somewhat predictable (Windows).
This is a key milestone for this project.
I ran this on a VM. Nice little OS, however I unfortunately don’t have a practical use for it as yet.
With Java and Qt ported to it its useness would be increased!
I think one of the former Amiga OS projects did have Qt3 ported.
True in some respects, but do we need another OS just to run Java and Qt based apps?
I think AROS will find it’s own niche…
You mean like Haiku?
I do master-slave multicore. Core Zero runs the application and it tells each other core what task to do. It’s optimal. I want to run one application faster, not two!
Pray that processor cores stop multiplying like pest over time, then. Otherwise, the master core could easily become a performance bottleneck.
Edited 2013-08-25 07:39 UTC