Amiga.org
Amiga computer related discussion => Amiga Software Issues and Discussion => Topic started by: DrValkyer on December 12, 2007, 05:19:09 AM
-
Amiga is today in the situation where it cannot be used as a main computer even for simple general stuff, like browsing Youtube, using some non-amiga applications and so. Of course I understand this; Classic Amiga HW is too slow for some modern applications or porting them to AOS is impossible/difficult.
So I got this little idea, dunno if you guys have already talked about it but here it is anyway, why not use x86 hw as a accelerator for classic Amiga? And no, I haven't smoked pot recently or drunk alcohol, thank you! :-D
[color=800000]"WTF? A new x86 turboboard for classic? Insane! :pissed: "[/color]
No, I mean that we could use old/new x86-PC as a server for x86 programs and then only the GUI+graphics were drawn in AmigaOS like in x-windows system. So for example we could use PC with linux + x-windows-server and modified x-windows-client on Amiga.
[color=800000]"Huh, why not use VNC / RemoteDesktop?"[/color]
Because it sucks! I want feel programs like they were native applications, not some mirrored mess in one little window. This could be achieved by using special "launch"-icons on workbench and special x-windowing-system.
For example if I launched a special-icon (pointing to a x86-program on PC) on my workbench then it would launch a x-window-client on Amiga which itself would tell the PC to run this program and mirror it's GUI+graphics back to the Amiga by using x-windows. When we add a twoway-networkfilesystem to this I could use Amiga files from PC program just like the PC program were running on Amiga natively.
One usage scenario could be for example to use Gimp on Linux. At first I just clicked my Gimp-icon on workbench and then the linux-PC would start running the Gimp and I could see the Gimp GUI on my Amiga. For next I could save the image manipulated in Gimp and save it to the AmigaHD via NFS and voi'la I would see a new image file on my Amiga machine!
In this way one could run virtually any application which exits on Linux/Unix. And think about possibility to use small ITX-PC-boards as an embedded "turbo-board" on your AmigaTower! The small ITX board could be even clued into your empty Zorro / PCI slot :-D and you would feel like it were a real accelerator board! :-)
[color=800000]"Ok, so what about the connection and it's bandwith problems?"[/color]
At first the x-windowing + PC-controlling could be done via normal ethernet network but this may cause problems. I am not sure but I think that the networking on Amiga is not fast enough to be able to carry realtime-graphics even on local-network (this true?), so this would make the graphics to slowdown and jam and for example browsing internet wouldn't be so smooth anymore. Of course a special HW-link (Amiga-PC) could be build for this and all x-windows + stuff would be transfered via this fast link.
[color=800000]"Ok, so what if it were possible? Why not just use directly the damn PC?"[/color]
Hey! I am making here the assumption that one wants to use Amiga as a main workstation at any cost, also slaving CoreDuo-PCs would be funny. What do you think? :-? :-D
-
Sorry, I completely fail to see the point.
Why not use WinUAE as a start, so in addition to the fantastic speed you'd get a great RTG card for free - and wouldn't have to rewrite all your software to enable the graphical client/server idea.
The more you develop this idea of yours, the more you end up with an Amiga on a PCI board sitting in a Wintel PC. Actually, most / all Amiga hardware can be emulated by software at higher speed, so you'd optimize the PCI card to nothing. Then you end up with WinUAE.
The point is, there's no rational in using physical Amiga hardware. It's the fun of it.
-
Sorry for running off-topic, but your headline ignited my X86 accelerator idea again:
In short, imagine an Amiga accelerator. But instead of a, say, 68060, there's an X86 emulating it.
Bloody hell, I'll build it myself one day, if no-one else does.
-
Hi, sorry to be patronizing but you have heard of AROS right?
-
X-Window system and linux apps won't give any any "amiga feeling" anyway, so why not just run linux on a PC directly?
Heck use Keyrah and you can even use your amiga keyboard... ;-)
-
Sorry, I completely fail to see the point. Why not use WinUAE ...
Hell yes this is unrational idea :-D , but arent't you with the WinUAE/AROS in the same situation?
The point is (if there is any) to be able to run x86 software on Amiga "smoothly" like they were native applications, can you do this inside WinUAE?
If windowing system were build in right way I dont think that you should need to rewrite target-programs at all only heavily modified x-server + x-client but like piru said the whole x-windowing probably wouldn't look AOS anymore.
And yes I have tried AROS a long time ago when it still were buggy (nowadays different story).
I just don't see a point using WinUAE + task switching because then I would prefer to use two desktops at same time (heh like I am doing it with Amiga right now). :lol:
-
In short, imagine an Amiga accelerator. But instead of a, say, 68060, there's an X86 emulating it.
Sounds interesting :-)
Do you mean that you would attach PC somehow to the Expansion Port of Amiga?
-
No, he means a turbocard like the blizzard or apollo cards, where the CPU is a cheap x86 chip. All you'd need is a boot-rom on the board that loads a 68k emulator and then boots the amiga and you'd have the fastest classic amiga ever.
The downside is that the glue logic to attach an x86 chip to the amiga would be really hard to do (as far as i'm aware; i'm no hardware designer).
-
Ah, you beat me to it. But yeah, that's the idea.
One could of course use some FPGA logic to interface the whole thing.
You could also have all the fastmem you want on the accelerator board, as there are no 24-bit address limitations in that jungle. The RAM speed could also exceed the traditional by far.
Someone build it, I'll buy it - or else I'll have to start experimenting with an old A500 and my Xilinx II dev board.
And that ain't going to be pretty. ;)
-
You've gotta love the fantasies that people have on this board.
-
Hey, a little daydreaming never hurt anyone. ;-)
-
Oh, what about for starters to just connect expansion port to the PCI-bus of PC (emulate 68k) and so make a protype and dev-platform. For example by using PCI-FPGA board, like Dragon (http://www.knjn.com/board_dragon.html) (pretty expensive 300$,hmm heard that name somewhere... the most likely Dragon named turbocard we are ever going to get :-P)
-
@skurk
sounds like a very interesting idea, start your own topic and many strange things can happen :)
-
Nice idea! But it's far from original. It's already been done on A500, not as an accelerator, but as a hardware emulator;
http://www.amiga-hardware.com/showhardware.cgi?HARDID=333
The idea was to load MS-DOS and Windows using the NEC version of an 8080 on board, thus effectively having a fully fledged PC inside the Amiga. A friend of mine at school had one and it worked beautifully. Of course, there was no GUI integration as you envisaged.
Still, with the low power cheap Geode and a custom northbridge I believe it's duable!
-
Mmmnnnooo, that's not what I were thinking. I'm thinking of an accel that replaces the existing CPU - not a software-activated/passive unit. You are right, they both are mounted inside the computer, but that's about where the similarities end, IMO.
Anyway, nevermind my rambling. I'm not too good with words. :)
-
@ DrValkyer,
You have an idea, then why not go for it.
Just because others don`t see it as a good thing because you can use WinUAE or AROS or AMiKit instead.
Its this sort of thing that stops ideas from coming into action.
They do have a valid point of course and I do agree I use AmiKit for a FAST Amiga.
My Tower system (signature) is to slow for real applications if I am truly honest.
But all the same its an idea that may work.
There are lots of ideas out there. Some good some not so good, don`t be put off.
I think some times people want to go for the more easy option, I am one of them but having said that, as I say don`t be put off, just because some one else cant see the point of it, if we all done that then we would get nothing done.
Its ok I suppose to sit back and say why bother?
I say why not?
Its people with ideas that get things done, example AmiKit, AROS, WinUAE. If it was not for those people then we would have virtually nothing.
Mike.
-
Well, you lost me. You want a X86 accelerator in a amiga with pci slots. You can buy them cheap at any computer store. Along with WinUae you can run your amiga apps. Putting an accelerator with x86 in amiga would allow the same thing but unless you rewrite all the amiga apps, games, etc for x86 you would be stuck with a expensive wintel box. I think you would be best off buying a PC and using WINUAE. You can still make claim to having the fastest amiga if you want to.
-
What's the deal with amiga-people and custom hw anyway? What about just porting the os for x86 based machines? Apple did it... :idea:
-
I think someone on a1k.org already is going to experiment with a slave x86 system connected to his A1000 as a number cruncher.
-
@Skurk >> sounds great. go for it.
I would love to my part to such a card, but it has been far to long since I have done any HW design.
Also, I would not know how to do the logick, that is, how to interface with, any excisting CPU even regular RAM.
-
Well, you lost me. You want a X86 accelerator in a amiga with pci slots.
Nope! My first post was about running x86 apps on x86 and then showing the results on Miggy's screen.
My later post was a reply to skurk. And what I meant was that if someone is going to build an turbocard using a x86 processor which software emulates mk68 processor it would be better to start with a PCI-FPGA card which is connected to the normal PC which itself is running mk68-emulator and using PCI-FPGA card as mk68-busboard. This way you virtually dont have to do anything than connect PCI-FPGA card in right way to Amiga's expansion port and program the FPGA.
Huh, getting complex... :-D
-
What's the deal with amiga-people and custom hw anyway? What about just porting the os for x86 based machines? Apple did it...
That has already been done (http://www.aros.org).
Dammy
-
dammy wrote:
What's the deal with amiga-people and custom hw anyway? What about just porting the os for x86 based machines? Apple did it...
That has already been done (http://www.aros.org).
Dammy
AROS would be my #1 choice if it could seamlessly (and flawlessly) run the good old 68k games and demos out-of-the-box.
-
DrValkyer wrote:
Huh, getting complex... :-D
He he, indeed ;-)
If no-one else does it, I will - one day. I'm trying my best to wrap my head around the mysterious world of electronics these days. I'm starting to get the hang of the basics, but I'm far, far, far, far, far away from accomplishing something like an x86 accelerator.
:-)
-
AROS would be my #1 choice if it could seamlessly (and flawlessly) run the good old 68k games and demos out-of-the-box.
EvilRich is working on that very project (http://thenostromo.com/teamaros2/?number=7). Donations toward his work is always welcomed. :-D
Dammy
-
in regards to building an x86 Amiga accelerator...
You could also have all the fastmem you want on the accelerator board, as there are no 24-bit address limitations in that jungle. The RAM speed could also exceed the traditional by far.
Very true.... Of course, you'd exceed the speed of the whole old Amiga by far. At that point, why not virtualize (a skeleton) of it. Then you've ended up back at Amithlon. Take away the copyright problems from that, and you're at AROS.
I just don't see the classic hardware having a future beyond that of retro/collectible.
-
Ilwrath wrote:
in regards to building an x86 Amiga accelerator...
You could also have all the fastmem you want on the accelerator board, as there are no 24-bit address limitations in that jungle. The RAM speed could also exceed the traditional by far.
Very true.... Of course, you'd exceed the speed of the whole old Amiga by far. At that point, why not virtualize (a skeleton) of it. Then you've ended up back at Amithlon. Take away the copyright problems from that, and you're at AROS.
I just don't see the classic hardware having a future beyond that of retro/collectible.
Very very true...
-
< The idea .. Sorry its in german > (http://www.a1k.org/forum/showthread.php?t=1897)
< First letters on Amiga.org > (http://www.amiga.org/forums/showthread.php?t=29072)
< Schematics > (http://de.geocities.com/a_bandel/Babylon.pdf)
< read the second comment > (http://www.amiga.org/gallery/index.php?n=2210=6)
< actual overview .. now with 1Ghz Athlon instead P3 700 > (http://www.a1k.org/forum/showthread.php?p=128056)
< reserved for the pci-bridge > (http://www.a1k.org/forum/showthread.php?p=127303)
:hammer:
-
How high can you clock a 68K FPGA implementation? That to me seems far more interesting than an x86 accelerator or remote execution / remote procedure call enviornment. The ColdFire is cool, but can you make a faster, more compatible processor in an FPGA? Sorry, I haven't been following the FPGA threads.
The idea of using an x86--or any other host system--for remote execution certainly isn't anything new. Rather than move GUI data back and forth, you're better off having a local GUI and just transporting data and procedure calls. Or something like that. High-speed interconnects would be prefferable--maybe GbE on an accelerator card with an appropriate buffer ("instructions" and data) and custom (or open) protocol? It would then at least be interoperable with most of what's out there today.
-
AFAIK the framework needed for a modern x86 CPU is a major pain in the *ss. You'll need lots of stuff like a north bridge (RAM controller, interfaces) and high performance power regulation. To make the north bridge work, you'll have to license a complete PC BIOS, which in turn requires more peripherals, so you end up with 3/4 of a PC mainboard.
The easiest way would be to take a small (µATX, mini ITX) mainboard and build a PCI <-> Zorro III or CPU-fast-slot bridge and just connect the mobo to the Amiga, just like in Ratte's link - would even give you an upgrade path, so you might consider using PCIe instead in the first place.
Again, this is pretty much an Amiga on a PCI card. If you also use the x86 side's graphics hardware through RTG and the sound through AHI, there's actually not much left of the original hardware, so you could also use UAE with Keyrah and Catweazle.
-
dammy wrote:
What's the deal with amiga-people and custom hw anyway? What about just porting the os for x86 based machines? Apple did it...
That has already been done (http://www.aros.org).
Dammy
"If you came here because you thought AROS was a finished, complete and fully usable operating system, you will most likely be disappointed." <- In my book, that's a no.
-
AMD processors have all that stuff (North Bridge) on-chip.
While reading this I was thinking MiniMig-type of accelerator, but the device uses a real 68000 CPU. I think there was another post which mentioned 68k emulation in an FPGA, then clocking the hell out of it.
How would an FPGA emulated 68040 run on an accelerator? Could an FPGA emulate an '040 with an equivalent speed of 80MHz or more? Could it work in a standard Amiga accelerator, or would it require more electronics?
With the way we're moving along with technology, I would expect that we could build a 68000-compatible accelerator without a real 68k on-board.
But then, the question is who's going to front the money. Engineering Master's project, anyone? Seem like some of the best stuff came out of uni... :-)
-
Ratte wrote:
< The idea .. Sorry its in german > (http://www.a1k.org/forum/showthread.php?t=1897)
< First letters on Amiga.org > (http://www.amiga.org/forums/showthread.php?t=29072)
< Schematics > (http://de.geocities.com/a_bandel/Babylon.pdf)
< read the second comment > (http://www.amiga.org/gallery/index.php?n=2210=6)
< actual overview .. now with 1Ghz Athlon instead P3 700 > (http://www.a1k.org/forum/showthread.php?p=128056)
< reserved for the pci-bridge > (http://www.a1k.org/forum/showthread.php?p=127303)
:hammer:
Can we get any of this translated? My German is rusty, dating back to 80's info files for BBS downloads...
-
How about Coldfire?
:flame:
he,he....sorry folks, I just couldn't resist.
-
"If you came here because you thought AROS was a finished, complete and fully usable operating system, you will most likely be disappointed." <- In my book, that's a no.
Or outdated documentation on AROS.org. There is only one major "gotcha" left with AROS and that can be neutralized during a fresh install by creating a very small AFFS for booting and make work partition as SFS. I have AROS running natively on a laptop and a desktop without issues.
Dammy
-
I actually had an idea of usin DosUAE with dos 6.22 and booting strait into WB from the batch file. True, its not real Amiga hardware but basically its the OS we want to use. Remember when you had to have Dos to use Windows? This would basically be the same idea just OS 3.1 and Dos underneath the whole thing like W98 and such.
Its fun to play around with this stuff and it always sparks new ideas that could lead to something different. Anything besides Vista is fine by me!! LOL
-
This is an interesting idea. I've been thinking along similar lines myself.
However, I'd like to have some FPGA between the x86 and the Amiga to translate and to handle SMP, so the Amiga thinks it sees a ridiculously fast single-core 68k chip rather than a multi-core x86.
This would involve some looking ahead and translating in parallel as the FPGA would be somewhat slower than the CPU.
-
This idea gets brought up on a regular basis.
Personally, I think it would be easier to create an FPGA based board that would plug into an X86 PC and emulate an advanced Amiga.
-
@Iggy
Yes. That's even better.
-
@DrValkyer
Yeah mon, I totally see it.....
(http://new.assets.thequietus.com/images/articles/4603/AMOORE_STOOLP_001AB_1278868344_crop_550x450.jpg)
-
This idea gets brought up on a regular basis.
Personally, I think it would be easier to create an FPGA based board that would plug into an X86 PC and emulate an advanced Amiga.
I'd be amazed if an FPGA that has to communicate through a "slow" PCI-e bus could run faster than pure software.
On a low end PC you'd still have 3 to 7 multi-GHz x86 cores plus the GPU to act like a custom chipset.
-
2 things would help the problem....
1. A way to directly access the floppy drive from within UAE in read and write form (because OS4/MOS/AROS/Amithlon all use UAE for chipset emulation)
2. Some affordably priced replica Amiga cases to house modern X86 motherboards for the above.
Nobody has done this.