Amiga to use Linux Kernel

Oh my dear God!!! by Anonymous Coward · 1999-07-08 20:10 · Score: 5

If you thought Linux Advocates were obnoxious now, wait 'til you see the new, improved Linux Amiga Advocate! ;-)

Re:Actually, you've got it backwards... by Adam+Wiggins · 1999-07-09 01:54 · Score: 2

By the way, the reason that Quake was slower was because Mesa was not as optimized as the 3Dfx miniGL drivers for Windows. They focus more on correctness and completeness than speed.

However, Mesa 3.1 is going to be MUCH faster, mostly due to heavy optimization work done by Keith W. and some others. I haven't tested Quake, but my own applications saw a speedup of 50%, and sometimes more. Keith reports that on both the Banshee and the Voodoo 2, Quake 2 timedemos are 10-15% *faster* than the running on 3Dfx's miniGL in Windows. Plus Mesa is a real OpenGL implementation, with all the bells and whistles!
I imagine that even Windows users will want to start using 3Dfx/Mesa with the release of this new version.

Don't quote me on those numbers, go look for Keith's timedemo posts on the Mesa-dev mailing list (www.mesa3d.org is the site). And Mesa 3.1 beta 2 is quite stable from what I've seen - download it and try it for yourself. I've been using it for Q3Test and it's quite speedy, certainly just as good as my Windows-using co-workers.

This is why I love open source/free software. In the time it takes you to say, " isn't as good as it could be", it's already better. In the meantime, 3Dfx has been working on their OpenGL driver for going on three years now and still have yet to make a "real" release.

Re:amiga is going to have a hard time by Eric+Green · 1999-07-08 21:52 · Score: 3

The nifty thing about AmigaOS (and BeOS, for that matter) was the fast message-based IPC mechanism. I suppose they could hack that into the Linux kernel to get them the speed they would need to compete with BeOS, or even hack MkLinux to get them that, but it's not a matter of just taking the normal Linux kernel and plunking it in place. At least if they want "classic" Amiga programs to be re-compilable on this architecture without running like crap.
Sockets just won't get it. I've been doing a lot of work with sockets lately, and they're too bloody slow for the kind of IPC work that BeOS and AmigaOS-Classic do.
But maybe they're just dumping the whole Amiga project and going to become just Yet Another Linux Distribution. Maybe this is how Gateway is going to maintain their relationship with Microsoft while jumping on the Linux bandwagon, i.e., by taking this defunct brand name they bought for near nothing and attaching it to their new Linux line. SIgh.

--
Send mail here if you want to reach me.

The Linux-hype claims another casuality! by jonr · 1999-07-08 20:03 · Score: 2

What can I say? This sucks. Linux is nice, but for the "new" Amiga? Gee, I guess I'll just put mkLinux on a iMac and call it iAmiga! :
QNX/Amiga would have been more exciting...
But I'll keep an eye on the development, as I have for the last YEARS!...

Jón

Re:What a waste of a year by Matts · 1999-07-08 20:37 · Score: 2

All your arguments about things coming to Linux (Flash, Director, etc) because of Amiga still don't create a compelling reason to use Linux instead of "Amiga-Linux". They do create a compelling reason to use Linux though.

My argument was about What, over and above what Linux will offer when Amiga-Linux comes out, will compell people to buy Amiga, rather than (say) Red Hat Linux? - not about what would make people choose Amiga-Linux over something other than Linux. Given that, your arguments about commercial support, multiple platforms, etc, are moot. Linux has that today.

And yes it was a waste of my time. I was Manager of the ICOA PMWG. But I'm not bitter :)

Finally, you say it's interesting because it's based on concepts newer than 2 decades. Which part? The Object model? Well, that remains to be seen - and what is an Object model if not an Object oriented API to the OS? We've had API's for decades. And Linux _is_ based on decades old technology.

I say it would have been interesting if it was going to be QNX based, instead it just looks like a Linux distro with a new commercial look and API. Woo hoo. Not.

perl -e 'print scalar reverse q(\)-: ,hacker Perl another Just)'

--

Matt. Want XML + Apache + Stylesheets? Get AxKit.

What a waste of a year by Matts · 1999-07-08 19:54 · Score: 3

A year ago Amiga made this big announcement about this secret chip that they were in talks with, that had 10 times the power of anything currently on the market. Later that changed into "We're evaluating different options, including the G4".

At the same time they were in talks with BeOS about using their kernel for the new OS (yes, I know this was never public knowledge, but some day someone official will leak this information to the press). Then Amiga pulled out of that deal 'cos Be asked for too much money - so QNX was picked, and announced as "the best technology available". Now even that is down the pan.

So what happened to the statements like "realtime is important to a multimedia OS". I guess it is'nt. And I guess neither is QNX Neutrino's transparent clustering technology (that would have made beowulf look hard to work with). In the meantime Amiga developers drop like flies, as the Classic update gets pushed further and further back, and the specs for it shrink.

So who is going to buy this new Linux box? Not current Linux users for sure - they're happy with what they've got (I know I am). Not current Amiga users - there's nothing to tempt them to buy the new Amiga over and above Linux. I guess they'll have a good shot at the embedded/palm/consumer market. Good luck to them.

Sad really. In all that time we haven't seen one iota of progress from them development-wise. What are they going to do with their current code? Bin it and re-write for X, instead of Photon? (yes, I know the POSIX stuff will be portable). We've seen constant changes in direction. There's no faith left. Today I read comp.sys.amiga.misc for the first time in a while and there's only about 2 people defending Amiga's actions (and only to the extent of "wait and see what they deliver"), compared to many more a year ago.

So, since the only revolutionary stuff is gone (the QNX Neutrino OS) - what's left? An unknown Object model running on top of Linux? I think I'd rather stick with Corba and what I've got right now. What a shame - I was very hopeful a year ago. What a waste of time.

Matt.

(all the above quotes are paraphrased)

perl -e 'print scalar reverse q(\)-: ,hacker Perl another Just)'

--

Matt. Want XML + Apache + Stylesheets? Get AxKit.

Re:What a waste of a year by deusx · 1999-07-08 20:16 · Score: 3

A year ago Amiga made this big announcement

That's their problem-- they should have never made announcements. But they had to do something to keep up interest. Things change though.

(stuff about BeOS and QNX)

Maybe BeOS asked for too much money, maybe they weren't quite what they wanted. Maybe the original announcement should have been, "We're investigating BeOS, and it looks good." As for QNX, maybe it was the best technology and better than Linux-- but compare the developer bases. I've never heard of QNX (tho granted, I could just be oblivious), but I've sure heard of Linux.

Maybe they have faith that Linux can *become* the best technology with their help.

So what happened to the statements like "realtime is important to a multimedia OS". I guess it is'nt.

Could just be PRspeak. Or, could be that they'll be incorporating realtime themselves or using realtime extensions by other people for Linux.

So who is going to buy this new Linux box?

Me.

Not current Linux users for sure - they're happy with what they've got (I know I am).

No I'm not. I can't run Macromedia Flash or Director, and any multimedia performance under Linux sucks rocks. Love the OS, use it at home, but I still have to boot into Win98 occasionally to do part of my job. (i.e. Flash on the web, CDROMs with Director, QTVR creation, etc, so forth...) I see this as what Amiga might bring to Linux.

Not current Amiga users - there's nothing to tempt them to buy the new Amiga over and above Linux.

Umm... maybe an OS that runs on multiple platforms? Maybe modern hardware? Maybe renewed contemporary commercial support? I've got an Amiga 1200 myself, and let me tell ya, tho it still rivals my current PC in a lot of ways, it's aging.

Sad really. In all that time we haven't seen one iota of progress from them development-wise.

That's what you get with a non-Open project. *Shrug*

What are they going to do with their current code? Bin it and re-write for X, instead of Photon? (yes, I know the POSIX stuff will be portable).

I would like to think that what they have is flexible enough to change easily. (i.e. the object model has low level classes to interface with the low level os and drivers, and so not everything up the line has to be rewritten)

Or maybe, they can use something else. LIke Berlin, or an Amiga windowing system with an X compatibility layer. But that's what I'd like to see, not what I think they might be doing.

We've seen constant changes in direction. There's no faith left.

I have no faith in Amiga. It's a machine. It was a religion, but I'd hope that people have learned. But it seems to have a promise to deliver multimedia to Linux now, and I think that's pretty significant.

So, since the only revolutionary stuff is gone (the QNX Neutrino OS) - what's left? An unknown Object model running on top of Linux? I think I'd rather stick with Corba and what I've got right now.

Hmm, I'm thinking the AmigaObject architecture will be the revolutionary stuff. Corba's just a framework. The AmigaObject environment will be the flesh in the framework. You can already see the seeds of it in the old AmigaOS, with datatypes for multimedia, etc...

Imagine-- a multimedia OS based on objects. Oh, wait, that's BeOS. :) Well, maybe Amiga can do better for Linux.

What a shame - I was very hopeful a year ago. What a waste of time.

*shrug* Let 'em waste their time. Or wait, was that a waste of your time? Ignore 'em and stop worrying. :)

My thing is that my ears perk up whenever I hear of someone trying to make a new operating environment that has concepts newer than 2 decades. Granted, those are tried and true concepts in Linux, but we've got better ideas by now one would hope.

Possible RTOS extensions for Linux? by tgd · 1999-07-08 21:27 · Score: 2

I wonder if this means they'll spend some time working on RTOS extensions for the Linux kernel. They've said all along that you need a RTOS for high-end multimedia, Be has said the same thing, if in different words.

Any kernel guru's know how much it'd change the compatibility of the Linux kernel if a more RTOS-type scheduler were present in the kernel instead of the stock one? I remember seeing a while back a group that was working on a replacement for the scheduler to provide better networking performance, how about multimedia performance?

It seems that high-bandwidth multimedia is where the convergence technologies (gak, marketingspeak!) are going. Lots of places are modifying Linux to allow the sort of real-time operations that are needed for such high-bandwidth data processing (Tivo or Replay, whichever is the one using Linux says they made changes to the LinuxPPC kernel to make it more realtime). I don't keep up with the 2.3 kernel discussions, but has this been talked about? It would seem to me that part of the redesign for even better SMP performance should be a look at the needs of a real-time scheduler so Linux can be used in a stock form to provide the sort of eye-popping multimedia demonstrations we see on BeOS, the old Amiga's (for their time), and to some extend modern workstations like SGI's. (Although SGI seems to be a more brute-force approach, largely based on their hardware architecture)

I think it'd be cool if Amiga approached the consumer market with a Linux-cored operating system with a non-X gui. X is awfully resource intensive for a consumer OS where a defined break between the GUI server and application clients are not needed. It'd be cool to see that GUI working with a linux kernel that made Linux THE platform for high-bandwidth multimedia applications.

It'd also be cool to see those extensions work under X for people who want that added power.

Wishful thinking? Probably. But I'm still curious about the RTOS issues with the Linux kernel.

Linux has good real time support already by roystgnr · 1999-07-09 01:12 · Score: 2

Do a "man sched_setscheduler" if you want to know how to include it in your own programs, or hunt down the (years old) "rt" utility if you want to be able to set any arbitrary program to be scheduled realtime. It's a nice thing to do with your MP3 player, for instance: in my experience, any system that can play MP3s at all can play them with zero skips if the player is running realtime.

The biggest problem here is security on multiuser systems, and safety against broken programs. Basically, if you run a program with SCHED_FIFO or another realtime scheduling strategy, no non-realtime programs get to run until the realtime one is in a blocking system call. In other words, if your realtime program infinite loops, you've just locked up one CPU in your system. For this reason, only root (or a suid root program) can set a process to run with realtime scheduling; otherwise running "rt cat /dev/zero > /dev/null" is a pretty sweet denial of service attack, freezing the X server and console by starving them of CPU cycles. You can still ping the system, and you can still open TCP connections... but inetd will never get a timeslice to spend answering those connections.

Re:Maybe a port of BLAZEMONGER to Linux? by ewhac · 1999-07-09 02:59 · Score: 2

For those who are unfamiliar with the phenomonon of BLAZEMONGER ("If you don't shout when you say it, you suck when you play it."), you can read the complete archive of BLAZEMONGER "press releases." Still funny after all these years...

Schwab

--
Editor, A1-AAA AmeriCaptions

Re:about time for a decent AmigaOS by eponymous+cohort · 1999-07-09 00:06 · Score: 2

Well, Amiga had Multitasking, but as for User interface -- Ugh! Atari's GEM interface, as bad as it was, beat the crap out of it.

--