Poster | Thread |
fishy_fis
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 6:02:06
| | [ #101 ] |
|
|
|
Elite Member |
Joined: 29-Mar-2004 Posts: 2160
From: Australia | | |
|
| @Hypex
With Amithlon amiga os has direct access to the machines hardware. Write to RAM and that's exactly where you're writing. Write to VRAM, ditto. Want to access a pci card, you do so by addressing the bare metal and so on and so forth. Just because its using an emulated cpu a lot of the time doesn't mean its not directly accessing hardware. As for updated pirated copies, I never bothered. Its pretty simple to update the kernel and install a few os side updates. |
|
Status: Offline |
|
|
Hypex
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 8:13:32
| | [ #102 ] |
|
|
|
Elite Member |
Joined: 6-May-2007 Posts: 11230
From: Greensborough, Australia | | |
|
| @umisef
Quote:
Windows 9x. And I believe you also found that starting Amithlon would silently repair the RDB :) |
I also recall other fixes were to offset the RDB two blocks so the MBR bootblocks didn't trash it. But given what it stored in the RDB how were you able to restore this easily?
Quote:
Huh? How can a reset look fake? |
When it put the messages about the 68040 reset phase 1 then 2 or whatever it was on screen.
Quote:
I believe you are "wrong" about that --- the early startup screen is, in fact, as un-fake as they come. |
What I meant was it was obvious it wasn't the real thing. It's only used when needed but it didn't look or feel exactly like a bog standard A1200. I recall some jerkiness.about it. I know that wasn't the point of getting it to work, it just hwo I observed it.
Yes I know, the only jerk is me.
Quote:
Had I known how much of a pain that would turn out to be, I would have taken the time to implement my own startup menu, using native code and talking directly to the real gfx memory. |
How hard is to to setup up the VGA card with a basic bitmap mode and a cursor? So that it works in colour with a hardware sprite for a pointer. Can VESA mode do this?
Quote:
But at first, it looked like it would be easy, just a few hours. And it was "just a few hours more" for quite a while.... |
You know how it goes! You'd think by using the original routines it would save work but now you know better. |
|
Status: Offline |
|
|
Hypex
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 8:16:48
| | [ #103 ] |
|
|
|
Elite Member |
Joined: 6-May-2007 Posts: 11230
From: Greensborough, Australia | | |
|
| @michalsc
Quote:
How about switching to x86 camp and using a motherboard with EFI instead of BIOS? ;) |
I think that would be better!
I haven't seen it live but don't they have a GUI with a pointer in EFI? That means the PC has caught up with the Mac.
And the Intel MacBook I saw went backwards when it put up a PC text screen after the disc got stuck in the DVD drive. |
|
Status: Offline |
|
|
Hypex
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 9:02:59
| | [ #104 ] |
|
|
|
Elite Member |
Joined: 6-May-2007 Posts: 11230
From: Greensborough, Australia | | |
|
| @bison
Quote:
I'm not sure the hardware has to be special anymore - even the cheap stuff is fast and capable. |
It had good features that should have been updated and continued. And they could have after reading about the BoXeR. A machine that has an AGA chipset and PPC support. This should have been the AmigaOne!
With projects like Natami even now there is interest in futhering the hardware with designs like SuperAGA.
But what I see lacking is things like the copper. Sure now you tell the VGA card to render the framebuffer and switch to it in the next blank instead of doing live register changes mid-screen. But I still see merit in copper ideas. I think it would be cool if the AmigaOne could hack the VGA registers in real time. And being able to display multiple playfields overlayed would be cool. Being reduced to raster interrupts if that is done at all is just a poor mans copper at best.
The thing is that OS4 lacks good hardware drivers. AHI still uses soft mixing. It should have a hardware driver. RadeonHD drvier lacks overlay with no replacement yet AFAIK and MPLayer breaks. So in a way OS4 has gone backwards. It lacks custom hardware doing cool things but itself doesn't have proper drivers to program modern hardware like it should. OS4 almost ends up as being like UAE, there's this software emulation in it that shouldn't be there. That is emulating things the hardware can do for it.
I thought we'd moved on from one channel stereo 16-bit sound cards. Which sounds lame in itself. Should have been at least two track stereo! But that's right where we are still stuck at, one track stereo. Unless AHI has proper hardware drivers for the sound cards. And AHI is a mess, not being able to mix music clients with instant sound clients. It's all software, should have been fixed years ago!
Quote:
But software - it's still mostly pretty bad, so there's real room for improvement in that area. |
I think OS4 needs a better system to manage software. It just complcates things with paths, assigns and whatnot. We really need a drop in place software system or some type of packager to package applications so they can be easily installed. But we also need developers to support these methods. Too many times there's been some good software let down by forcing the user to install it. It shouldn't be that hard!
Look how Linux does it. All the files organised into folders where they belong. If we take anything from Linux it should be this! Everything is copied to the system and it just works.
I think if OS4 migrated from rigid config files to splitting it all up into seperate files in there it would be easier. Imagine if adding a path, assign or startup script was as easy as to copy a few files across. The developer would just have to organise his software and it coud be installed like the "MacOSX" way of dragging an application icon to a desination!
This idea also follows where Commodore was going. On OS2 they split the mountlist into DOSDriver files and added a User-Startup. Moving asigns and paths into DEVS: as files is the next step. The old way can stil be supported. Right now a programmer needs a script to tell the legacy Installer to add an assign at startup and not many do it. Imagine how much easier it would be just to copy a file. The user could do it easily without shaking their fist!
Sure, it's slower reading a directory full of files than to have it all "cached" inside a file. But these days systems are fast and I think the freedom it would allow would far outweigh the "cost" involved. Also, assigns only need adding when accessed.
How did my reply become so long? Well, so long! |
|
Status: Offline |
|
|
Hypex
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 9:46:24
| | [ #105 ] |
|
|
|
Elite Member |
Joined: 6-May-2007 Posts: 11230
From: Greensborough, Australia | | |
|
| @fishy_fis
Quote:
With Amithlon amiga os has direct access to the machines hardware. |
That sounds risky! Since it could poke addresses that have other meanings.
I do see your point, but the OS would still need drivers going in-between. As AmigaOS knows nothing about VRAM or PCI spaces. Still, by having the OS hand it to over to a hardwarew driver, would vastly optimse the operation.
Quote:
As for updated pirated copies, I never bothered. Its pretty simple to update the kernel and install a few os side updates. |
I'm not talking about pirated copies. Although as I was asking about your setup that's how you replied.
I'm talking about legal copies which my friends have. Most of the Amiga people I knew back then bought it. But this should be moot as according to this Amithlon went openn source: http://sourceforge.net/projects/amithlon/
So all we need now is for someone to update the files for the latest Linux, build a package installer and create an ISO that can boot and install it to a PC.
Alternatively, make an install guide. That would be a start. For those who know what they are doing. |
|
Status: Offline |
|
|
fishy_fis
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 9:58:31
| | [ #106 ] |
|
|
|
Elite Member |
Joined: 29-Mar-2004 Posts: 2160
From: Australia | | |
|
| @Hypex
Ah, my mistake. There are "updated" isos out there in torrent land, which is what I was refering to. There's a few people I know of who attempted to use them, only to find them more of a chore to setup than updating the original from scratch.
As for pci access, that's what openpci.lib is for :) Last edited by fishy_fis on 24-Oct-2013 at 12:47 PM.
|
|
Status: Offline |
|
|
olegil
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 10:06:07
| | [ #107 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @bison
Quote:
bison wrote: @michalsc
Quote:
How about switching to x86 camp and using a motherboard with EFI instead of BIOS? |
I'd start with Linux or NetBSD on a PC and work top-down, identifying and fixing the top five things that aren't very Amiga-like. The problem with working bottom-up is that you're always five years away from having competitive application software.
|
If your fixes are deemed useful it might even be possible to get them included upstream (as compile time options) if you do it that way, which is always nice._________________ This weeks pet peeve: Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean. |
|
Status: Offline |
|
|
Arko
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 10:52:37
| | [ #108 ] |
|
|
|
Super Member |
Joined: 17-Jan-2007 Posts: 1989
From: Unknown | | |
|
| @Hypex
Quote:
Hypex wrote: @fishy_fis
I do see your point, but the OS would still need drivers going in-between. As AmigaOS knows nothing about VRAM or PCI spaces. Still, by having the OS hand it to over to a hardwarew driver, would vastly optimse the operation. are doing.
|
On UAE you could do this by using P96._________________ AmigaONE. Haha. Just because you can put label on it does not make it Amiga.
I borrowed this comments from here (#27 & #28): http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=38873&forum=2&start=20&order=0 |
|
Status: Offline |
|
|
fishy_fis
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 12:44:59
| | [ #109 ] |
|
|
|
Elite Member |
Joined: 29-Mar-2004 Posts: 2160
From: Australia | | |
|
| @Arko
Funny enough, and unless Im mistaken you cant do this with p96 under UAE, but can, in a hacky way with cgx assigning a pci slot at a low level so its seen as a slot on a mediator (using mediator drivers). You can use p96 of course, but uaegfx I dont believe gives direct hardware access, but instead works as a wrapper of sorts with the host oses gfx. While the faster gfx cards available and better drivers of a host can help with speed vs the older cards available to 68k amigas and Amithlon it does restrict flexibility (inability to create custom screen modes with p96mode for example). |
|
Status: Offline |
|
|
drstrangelove
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 13:57:59
| | [ #110 ] |
|
|
|
Member |
Joined: 16-Aug-2005 Posts: 93
From: Unknown | | |
|
| @Hondo, all
Reading this thread has me curious about that initiative, so I called my friend, the systems architecture department who was responsable of that initiative and I have been asked questions that I share with you:
The last answer, leave me out of game ...
Q) What happened to the project? A) He was just a student project, when they got their degree, the project is over
Q)Why AROS? A)I was excited my computer Amiga, I think it was the best machine for your time, I like the idea of seeing her alive, embodied in the magnificent hardware we have now; AROS is as close to that
Q)What hardware did you use? A)the best a student could have: Asus Sabertooth motherboard 32GB RAM AMD 8-core FX Nvidia GTX 690
Q)Why the MIT exokernel and why not linux? A)linux is a monolithic kernel, too heavy, not real time and because we needed a genuine challenge. Moreover, the use of exokernel allows several fundamental things given the short time we had asigned to the project:
1 - the exokernel left expose, for the developer, hardware resources. 2 - this allowed us to very quickly change the AROS core and have it running with the hardware in just one quarter. 3 - make linux drivers to exokernel is very very easy
Q)What did you mean with this project? A)Winning the game ... :) no, out of jokes, the goal was to build the best gaming machine ever designed.
Sorry for my bad english ... Last edited by drstrangelove on 24-Oct-2013 at 01:58 PM.
|
|
Status: Offline |
|
|
Leo
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 15:04:28
| | [ #111 ] |
|
|
|
Super Member |
Joined: 21-Aug-2003 Posts: 1597
From: Unknown | | |
|
| Quote:
A)linux is a monolithic kernel, too heavy, not real time and because we needed a genuine challenge
|
Too heavy ? Explain ? What do you need Real Time for ? There are Linux variants that are real time, but I don't see your point here (should I mention that AmigaOS is not realtime ?)._________________ http://www.warpdesign.fr/ |
|
Status: Offline |
|
|
OlafS25
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 15:49:35
| | [ #112 ] |
|
|
|
Elite Member |
Joined: 12-May-2010 Posts: 6363
From: Unknown | | |
|
| @drstrangelove
I do not understand. Is that real? A AROS system based on exokernal? |
|
Status: Offline |
|
|
bison
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 16:00:17
| | [ #113 ] |
|
|
|
Elite Member |
Joined: 18-Dec-2007 Posts: 2112
From: N-Space | | |
|
| @olegil
Quote:
If your fixes are deemed useful it might even be possible to get them included upstream (as compile time options) if you do it that way, which is always nice. |
Maybe, but I'm quite pessimistic about getting changes in upstream. I had a short and turbulent stint as a Gnome developer many years ago that has left me somewhat cynical. The Amiga way of doing things is so far from the zeitgeist of software design that it's hard to get people outside the community on-board.
But that's probably not too much of an issue — I don't think too many actual code changes would be required to get a more Amiga-like Linux system. Most attempts so far have started by replacing the wallpaper and icons, which is pretty low on my list of priorities. I don't care that much about looking more like Amiga; I would rather have a system that responds more like an Amiga.
I would start by building a reference system and stripping down the kernel to support just that system. The next thing would be to speed up the boot process by disabling unnecessary services, streamlining script execution, etc. Third would be put together a lightweight desktop environment from existing components. Fourth and fifth I'm not sure about — I'd have to get to step three and see what needs to be done, or get feedback from other users on what seems bad to them.
Last edited by bison on 24-Oct-2013 at 04:05 PM.
_________________ "Unix is supposed to fix that." -- Jay Miner |
|
Status: Offline |
|
|
bison
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 16:04:18
| | [ #114 ] |
|
|
|
Elite Member |
Joined: 18-Dec-2007 Posts: 2112
From: N-Space | | |
|
| @Hypex
Quote:
But what I see lacking is things like the copper. |
Yes, I see what you mean. The commodity platform that "won" the hardware "war" is certainly less than what it could have been. There's this area of inertia where something is not as good as it could be, but not bad enough to replace. A lot of programming languages fall into this category.
_________________ "Unix is supposed to fix that." -- Jay Miner |
|
Status: Offline |
|
|
DC_Edge
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 17:19:59
| | [ #115 ] |
|
|
|
Regular Member |
Joined: 1-Oct-2003 Posts: 190
From: France | | |
|
| The day AmigaOS is based on a linux kernel will be the last time I would consider AmigaOS as an alternative operating system, and certainly the last time I would think about butying a new AOS machine.
Linux is completely different than aOS(ses), this is a true nightmare for me, where every piece of software in absolutely un-understandable by any tech guy who has not spent 500 hours playing with the su command and any other non senses (editors with so strange shortcuts).
Last time I had to touch a linux system, that was for an IT compagny, they offered some king of scrappy linux server to customers, with some CRM, mail, intranet. The solution was sold more expensive than a microsoft windows server and all the tools provided by krosoft that acheve the same thing better;)
With a big and rather well documented procedure, It tooks us (me, I was noob in ths compagny plus the manager) a whole week to master the server, then setup all things, that was on a debian distrib base.
Then the customer had to phone us many times, because some ip routes were bad, then then server hanged (yes completely), then mails were not able anymore to be sent by anyone, well, a big "it is ok, if you have the time to play with" thing IMHO.
Last time I had to deploy such things with a 2008 server, it took 2 days alone to have the whole solution up and in production, and without any proper manual for setup. I haven't had to come back. This is a "It works almost out of the box, it's simple to manage, it doesn't consume a lot of time, but this is more expensive".
Please let linux to server guys who congratulates themselves for being able to talk to an autist machine, and be serious, amigaos is nice due to a completely different philosophy.
Linux is almost mainstream, and thus, is really boring. Android has linux, mac products are linux ones....everything is nowaday either linux either krosoft.
Last edited by DC_Edge on 24-Oct-2013 at 05:22 PM. Last edited by DC_Edge on 24-Oct-2013 at 05:21 PM. Last edited by DC_Edge on 24-Oct-2013 at 05:21 PM.
|
|
Status: Offline |
|
|
damocles
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 17:39:00
| | [ #116 ] |
|
|
|
Super Member |
Joined: 22-Dec-2007 Posts: 1719
From: Unknown | | |
|
| @DC_Edge
There is a huge difference between Linux kernel and the GNU distros that use Linux kernel.
_________________ Dammy |
|
Status: Offline |
|
|
drstrangelove
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 19:25:30
| | [ #117 ] |
|
|
|
Member |
Joined: 16-Aug-2005 Posts: 93
From: Unknown | | |
|
| @Leo
I can not answer that, just share the answers from my friend, but I can tell you that that person is a big fan of computer amiga from the late 80 ...
@olaff25
in short: Yes
I'll try to explain what I know about it:
What I have seen work is a system AROS, I know EXEC has been modified and is supported by the MIT exokernel, has a system of additional libraries. Some are strictly for MIT exokernel.
I do not know much about operating systems, but when I saw it, I thought and felt that this was really the way to go.
The thing that most impressed me was that, my friend said it was the best gaming machine ever built
Sorry for my bad english
|
|
Status: Offline |
|
|
IntuitionAmiga
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 19:55:34
| | [ #118 ] |
|
|
|
Regular Member |
Joined: 5-Sep-2013 Posts: 118
From: Unknown | | |
|
| Wow, this thread is full of comments by people who don't know the difference between a kernel and an operating system yet bizarrely believe themselves to be qualified to waffle on about both.
Mac OSX is not Linux either btw |
|
Status: Offline |
|
|
wawa
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 20:02:21
| | [ #119 ] |
|
|
|
Elite Member |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| @drstrangelove
perhaps the work of your friend would be interesting alternative to look into by aros/arix developers? could you forward them to it? |
|
Status: Offline |
|
|
Signal
| |
Re: Should Linux kernel power future AOS solution ? Posted on 24-Oct-2013 20:40:10
| | [ #120 ] |
|
|
|
Cult Member |
Joined: 1-Jun-2013 Posts: 664
From: USA | | |
|
| @DC_Edge
Quote:
DC_Edge wrote: The day AmigaOS is based on a linux kernel will be the last time I would consider AmigaOS as an alternative operating system, and certainly the last time I would think about butying a new AOS machine. |
I feel the same way, and I do use Linux with comfort every day.
If people want to use Linux kernels or run on x86 then they should either run some Amiga look alike Linux distro or one of the x86 Amiga like systems.
I have tried most of them.... something is either missing or just does not feel right. Yeah, 'feel'. No, I can't explain it, and for those that have forgotten or have never used AmigaOS, there is no hope of ever expressing that man AND machine feeling.
As for moving to a different architecture, ARM is the way to go. But, not for the desktop, leave that PPC. There are plenty of small ARM devices that could be integrated into the Amiga environment with a cut down OS, enhancing the Amiga experience.
_________________ Tinkering with computers. |
|
Status: Offline |
|
|