Amiga.org
Amiga computer related discussion => Amiga Software Issues and Discussion => Topic started by: save2600 on May 10, 2009, 07:15:03 PM
-
Ever since the A1000, one of the features we were attracted to and were promised, was that we could easily change the personality of our Amigas. Effectively turning them into very different machines. Commodore pretty much totally ditched that concept and betrayed us all by marketing machines with Kickstart in ROM. So you save a few seconds bootstrapping your machine with ROMS - big deal. The way I see it, had Commodore stayed true with the Amiga teams original design, things could have gotten REAL interesting.
I guess what I am asking is this... besides some of the modern Kickstart patches and hacks, has anyone ever bothered to write a totally new and non-Commodore type "Kickstart"? Of course, this would mean using a totally different OS. Perhaps ST, Mac and PeeCee emulation would have been easier to swallow and more accessible to the masses.
And before I get too far ahead in my reasoning, isn't all of this possible? 256k-512k surely would have been enough room to play around with. At the time, all of the above mentioned platforms BIOS chips were typically even smaller than either of those figures.
So why hasn't the concept of a Kickstart on disk (or chips for that matter) been fully realised? Speed and ease of use emulating other platforms could have been so much better and Commodore could have marketed the Amiga as a true chameleon of sorts instead of wasting all that time and money marketing a PC-10, etc. :idea:
Licensing this stuff could not have been more expensive than the resources it took to create standalone clones, could it?
-
Speed and ease of use emulating other platforms
How is it any easier or faster to have the emulation booted off some disk first and only after that launching the actual target system? I'll take the method of launching the emulator directly, instead.
-
Purely for the sake of argument, lets say you inserted a Kickstart disk that told the Amiga it was really an AtariST. Kind of a TOS written for the custom Amiga chips/drives, and other I/O functions. You would then be able to insert a REAL Atari GEM disk and away you go. This scenario would save you the hassle of configuring an ST system AFTER Workbench was loaded and hypothetically speaking, grant you greater compatibility and speed since you are not converting ST functions to Amiga - at a Commodore Kickstart or Workbench level. Not to mention, the time saved loading Workbench and then the emulator. Perhaps my reasoning is flawed or impractical :-)
-
http://www.youtube.com/watch?v=K4ta4OYhsEY
well, the fact that the ROM is there doesn't mean that it can't be done. you can always ditch the ROM and do your stuff. not really sure why commodore is to blame for this ?
-
Early Commodore (or perhaps otherwise) marketing stated there was going to be much more made out of the Kickstart disk system. It certainly helped users make the transition from various Kickstarts a lot easier since they did not have to open up their computers and mess around with chips. Huge benefit for Commodore when dealing with the common end users.
Not trying to start a flame war here, just asking why more wasn't done to take advantage of Commodores awesomely flexible Kickstart architecture. 3rd party companies I would think, could have really exploited that, I would think.
Kind of like how you can turn your TI-99/4A into a Myarc computer system by installing a "simple" card into the PEB.
If it's true that Commodore only did the Kickstart on disk thing because the current Kickstart was so weak at launch and was all just a bad design, well... lol they must have digressed with the A3000 then :-)
-
Yes, C= created the WOM kludge to make up for the fact that Kickstart 1.0 and 1.1 weren't mature enough to be written in stone yet. A similar method was used with the A3k because 2.0 wasn't finished when the first machines were sold.
As Piru already posted, I'm missing the point why it'd be so much easier to port an entire OS to Amiga hardware that you're lacking the source code of (plus drivers for expansions) and then put up with incompatible software (due to 'no emulation') than to just run a much less complex emulation layer that lets you run the guest system on top of AmigaOS.
-
I guess my reasoning is flawed if you wouldn't have been allowed to have access to sourcecode Amiga<>Atari for their custom chips, etc. But that could have been ironed out through licensing. If not, how would original software be more incompatible due to no emulation? As an end user, I don't see how it's easier running emulation on top of and after Workbench. And we're all familiar with the incompatibilities of emulating. Especially when the MHZ (or lack of them) really count. I was just thinking emulation would/could have been more practical had it been written on and for more of a hardware level than software.
I'm sure I'm out of my league here and don't fully understand the politics and architecture of it all, just asking the "what ifs" and "why nots".
-
If not, how would original software be more incompatible due to no emulation?
Well, any software hitting the hardware directly is prone to fail as it's not really there. The systems you've got in mind have no protected access to hardware, so there's no control on who does what.
As an end user, I don't see how it's easier running emulation on top of and after Workbench.
What's easier than starting the emulation with a double click of the mouse? In addition, emulation usually enables you to run AmigaOS in parallel to the guest OS.
Especially when the MHZ (or lack of them) really count.
Performance-wise a ported system/OS may give better results than an emulated one, but this highly depends on the complexity of the emulation, esp. that of the CPU. When the CPU is no problem and the guest OS is somewhat hardware-friendly, there's little loss in performance (e.g. MacOS emulation).
I was just thinking emulation would/could have been more practical had it been written on and for more of a hardware level than software.
Sure. You just missed on the point that the only problem is lacking the source code it's impossible to port the OS in question to the hardware in question. Not to mention the applications in question. ;-)
-
I'm sure there might be obstacles. But the idea has benefits. As for Atari ST (or MAC) on Amiga. Without hardware emulation one could simple make a TOS API wich uses the Amiga hardware. Or even Amiga libraries.
And then there's the project to make a free Kickstart clone ;)
(to kick Amiga Inc leash ;) )
-
freqmax wrote:
And then there's the project to make a free Kickstart clone ;)
(to kick Amiga Inc leash ;) )
Now this is what I'm wondering about most. It's a fair assumption that most of the ROMs nowadays are burnt EPROMs, but why hasn't anyone started designing a new kickstart replacement?? That, right there would be the most of use "classic" item.
-
Methuselas wrote:
freqmax wrote:
And then there's the project to make a free Kickstart clone ;)
(to kick Amiga Inc leash ;) )
Now this is what I'm wondering about most. It's a fair assumption that most of the ROMs nowadays are burnt EPROMs, but why hasn't anyone started designing a new kickstart replacement?? That, right there would be the most of use "classic" item.
Because the ROMs are copyright to AMIGA inc and no one is allowed to copy, sell or reverse engineer them without their permission. It is why you have to buy AMIGA forever to get the emulation with the ROM images. Clonato have a license to sell them.
-
No need really to reverse engineer the Kickstart ROM. Application software expects certain functions in each library. Make a replacement function for each of them, viola!
As the Kickstart is heavily based on inheritance may not necessarily need to be a large programming task after all?
The v1.3 ROM ought to be a measure of the minimum amount of code reqiured to run. So how much of the 256 KB (ROM) is really used?
-
It is perfectly legal to clean room reverse engineer them, the patents are long gone, there are no trade secrets anymore. It is not legal to copy them or directly use their code since that would break copyright, which lasts 50 years. It is not legal to call them Amiga since the trademark belongs to somebody else.
-
Doesn't the AmigaOS license explicitly prohibit even clean reverse engineering?
As for name.. I found that "best friend" is el mejor amigo, which maybe is to close. But "friend" is compañero, most likely sufficiently different to be in the (legal) clear.
Otoh, library interfaces are published. Making replacements doesn't likely count as reverse engineering.
Seems a huge mystery as to the actual details of who has the copyright on AmigaOS presently ;)
-
There was a different set of Diagnostic Kickstart roms made for the A1200 I believe. I have seen them for sale before. I almost bought a set just out of curiosity but I sold my A1200.
-
Diagnostic Kickstart ROM ..?
Diagnosing specifically the A1200 hardware I suppose?
-
Doesn't the AmigaOS license explicitly prohibit even clean reverse engineering?
It doesn't matter what M$ lawyers says, a contract enacted when you open the box where the install media of an operative system is packaged is not a binding contract.
Also, I rememeber the #$%& DMCA has a clause that allows reverse engeniering of hardware that is no longer produced and guess what, it has been more than a decade since an amiga with a kickstart rom was last manufactured.
-
Is there an assumption here that the legal rights to OSx.x are holding back anything? Cannot the ROM images/software be purchased very cheaply "per machine" via Ainc (via Cloanto?) -- for far less than a PC manufacturer purchases Windoze? (Disclaimer: Maybe I've been staring at my computer screen far, far too long today.)
-
@save2600
I was just thinking emulation would/could have been more practical had it been written on and for more of a hardware level than software.
Putting something to KS ROM doesn't make it any more "hardware level" than having it as a separate program to run. ROM is just read only memory to store software to.
-
I would like to see a kickstart replacement that would accelerate the booting of AOS3.9 machines & classic PPC machines. (my 68k A4000 does two resets to patch the kickstart before it loads up, it makes the booting 3x slower)
-
Right - like a Kickstart with the latest patches burned into a neat and tidy ROM. Has anyone actually burned their own EEPROM like this or is everyone content with the annoying patch and reset scenario?
(http://www.coinopdreams.com/kick.JPG)
-
save2600 wrote:
Right - like a Kickstart with the latest patches burned into a neat and tidy ROM. Has anyone actually burned their own EEPROM like this or is everyone content with the annoying patch and reset scenario?
Burning an EPROM with the BB2 updates (and other fun things) is fairly straightforward, check doobreynet (http://www.doobreynet.co.uk/amiga.html) for the tools and docs.
-
>by Piru on 2009/5/11 0:53:51
>@save2600
>Quote:
> I was just thinking emulation would/could have been more practical had it been written on and for more of a hardware level than software.
>Putting something to KS ROM doesn't make it any more "hardware level" than having it as a separate program to run. ROM is just read only memory to store software to.
For Atari ST, it would make it more hardware level since the TOS would be residing at same memory location as original TOS on Atari ST ($FC0000) so software accessing OS functions wouldn't require any address translation. They should be able to run full-throttle (7.16Mhz instead of 8Mhz). And if they re-compile the ST executables and filter out the direct I/O calls like $FFFF9201 to vector to Amiga software equivalents, they should also be able to run stuff that directly accesses hardware ports. However, that would be a more challenging program to write.
-
There's an alternative to filter direct I/O calls inside software. Scan the memory location residing at the same location as the I/O port and update any hardware with it.
-
@save2600
Another way could be a very simble BIOS ROM that kicks the AOS/AROS running from the HDD. That way we could have the latest kickstart on the HDD & faster boot. And AROS only machines would not be dependant on ROM that AmigaInc owns.
-
>by freqmax on 2009/5/12 9:02:42
>There's an alternative to filter direct I/O calls inside software. Scan the memory location residing at the same location as the I/O port and update any hardware with it.
But then you have to determine the maximum frequency, freqmax, for updating those i/o ports. For example, joystick ports and paddles can be read at any time; others may not require an update that frequently, for which your algorithm is good.
I was thinking scanning the graphics memory area to fix up the image. I.e., map the graphics memory to the same area as STs so ST writes full throttle to the graphics memory and then update/fix the image during VBI before making it visible.
-
Apple went from ROM chips to files on the CD/DVD that are installed onto the hard drive.
Tandy/Radio Shack once promoted DOS in ROM as a way of preventing virus problems;they dropped it by the time DOS 4 came along.
So it's no giant deal either way.
I think perhaps the thought of having universal hardware ,and then plugging in a personality module is very attractive but perhaps only the hobbyist or rearcher will ever see one.
-
amigaksi, I know my update-I/O suggestion have serious drawbacks. But so has other approaches to. The only really clean path is FPGA ;)
VBI is however a suitable opportunity to update at.
recidivist, Personality module.. well have a look at the FPGA systems (Minimig, C-One, Fpgaarade etc..). There a simple binary can change the actual hardware and software instantly.
-
>besides some of the modern Kickstart patches and hacks, has anyone ever bothered to write a totally new and non-Commodore type "Kickstart"?
Check out Dragon's Lair.
-
I don't get it. What's the game got to do with Kickstart?
-
If I remember correctly Dragon's Lair for the Amiga 1000 had it's own kickstart as a mean to save precious ram.
-
Well I'll be dipped. That sounds really cool, but awfully strange. What does it do... re-write Kickstart with a very streamlined one that doesn't use as many resources? This would mean it patches Kickstart, right? Erasing what is there and replacing it. If so, how come no reset? Game seems to simply fly off floppies. Figured they had their own proprietary crunching/de-crunching scheme, but not a Kickstart hack.
-
Twinkick "3.9" (http://www.amiga.org/gallery/images/4308/1_3212.jpg)
KickWork3.1 (http://www.a1k.org/download/area51/OLD/ftp/KICKWORK31.MPG)
KickTOS (http://www.a1k.org/download/area51/OLD/KICKTOS.MPG)
:-)
-
1. You could use a ROM switcher to change kickstart versions, and create your own if you wanted.
2. You can sotboot a kickstart image such as with softboot, blizkick, etc.
3. You can patch/change system library code during boot.
4. AmigaOne has kickstart on hard drive. I assume SAM does as well. OS4 classic certainly does not come with a ROM chip...
-
@Ratte:
KickTos... now we're talking :-) How does one get a hold of that brilliant piece of coolness? Twinkick I'm familiar with, but Kickwork3.1... what was that all about? Or should I fire up the German translation program? lol
-
@save2600
Rembeber that the amiga 1000 (as well as early 3000s) loaded the kickstart into ram, they did not had roms built in the motherboard.
-
Kickwork is a complete 512KB 3.1ROM crunched with LZMA down under the 256KB WOM limit.
The 256KB includes the crunched rom and some decrunching code.
The Kickstart itself is modified with an enhanced version of Pirus Kickwork-Patch.
... Its a Kick3.1 + WB3.1 on one disk for the A1000 ...
save2600 wrote:
And about the e-mail... are you xxxxxxx_xxxxx@xxx.xx? If so, yes - I got that, but haven't had time to play around with 'em yet. I did reply with a thanks. If that's not you, then no, I have not received it.
-
@billt:
I'm aware of the first three suggestions, but have no experience (yet) with the other machines and OS4.x
@little:
Thanks, yes I knew that :-)
@Ratte:
Didn't know Piru had an all in one solution. Guess I'll hound him for a copy :-) I've been using Andre's Twinkick, which is brilliant. BUT... on a 2.5mb A1000, you lose 256k (boo-hoo). lol
If I am reading this right, with 3.1 squished to fit into 256k, I should have even more ram :pint:
-
save2600 wrote:
@Ratte:
Didn't know Piru had an all in one solution. Guess I'll hound him for a copy :-) I've been using Andre's Twinkick, which is brilliant. BUT... on a 2.5mb A1000, you lose 256k (boo-hoo). lol
If I am reading this right, with 3.1 squished to fit into 256k, I should have even more ram :pint:
Unpacked it will still be 512k :-)
-
Oh - and duh. He did that in able to fit the whole bloody lot into an 880kb disk I suppose. :crazy: But perhaps a routine was written to throw the extra 256k in Fast Ram then? And I'm sure that's got to be within a certain memory confine (my autoconfig expansion ram). Which since TwinKick works, Piru's should still work :-)
-
details about kickwork (http://www.amiga.org/forums/showthread.php?t=32883)
the crunched 256kb are loaded to the wom
decrunching a 512kb (executable) rom-image to fastmem
same as twinkick, but without kick1.3
but it is able to repair the romimage in fastmem if some kind of code or a systemcrash destroys the rom
the "kickwork"-patch is added for fun only and has nothing to do with the lzma-compressed kick3.1
save2600:
did you read your complete emails?
-
@ratte:
a little OT, but have you noticed this thread:
http://www.amiga.org/forums/showthread.php?t=41497
maybe u could help them out with the aros kickstart replacement project? docs like the kernel reference manual are needed.
-
>by freqmax on 2009/5/12 13:53:41
>amigaksi, I know my update-I/O suggestion have serious drawbacks. But so has other approaches to. The only really clean path is FPGA ;)
FPGA, you mean "Forget Programming to Get an Atari"? I thought it was "Forget Programming to Get an Amiga" since Atari ST was more emulatable.
>VBI is however a suitable opportunity to update at.
For joystick i/o, Atari st uses 7812.5bps instead of 50/60hz.
-
>by Ratte on 2009/5/13 6:39:23
>details about kickwork
>the crunched 256kb are loaded to the wom
decrunching a 512kb (executable) rom-image to fastmem
same as twinkick, but without kick1.3
Another useful item for A1000 users would be to compress the 256K kickstart into the bootstrap ROM space and decompress that into the WOM upon power-up. Using a pair of 27512 EPROMs to replace the bootstrap ROMs would give you 128KB so compress the 256K Kickstart into less than 128KB (and decompressor also would have to fit into that space).
Perhaps, someone who has the copyrights to those bootstrap ROMs and Kickstarts could do it.
-
amigaksi wrote:
Another useful item for A1000 users would be to compress the 256K kickstart into the bootstrap ROM space and decompress that into the WOM upon power-up. Using a pair of 27512 EPROMs to replace the bootstrap ROMs would give you 128KB so compress the 256K Kickstart into less than 128KB (and decompressor also would have to fit into that space).
Perhaps, someone who has the copyrights to those bootstrap ROMs and Kickstarts could do it.
its easier to add an rom-socket to the a1000
< link > (http://www.a1k.org/forum/showthread.php?p=100270)
-
amigaksi, It's not necessary to provide the bits. But rather one can provide a script that will automatically copy the existing ROMs into a file. Which is then compressed and processed to files (.hex?) ready for burning EPROMs with.
-
R
save2600:
did you read your complete emails?
Hi Ratte:
Had to do a little investigating, but yeah - finally tracked it down. Kickwork - WOW, you pulled another rabbit out of your hat!
-seems to only work with my MiniMegs external 2MB expansion
-does not work with my A2091 and its 2mb of memory
...does the A2091 map its memory different or what? When plugged in,
the screen flashes red and constantly resets itself. Even after about 10
minutes. When I replace the A2091 with the other ram card, the screen
never turns red. Instead, flashes different colours for about 5 mins and
then finishes booting.
Not a complaint, but just wondering: why so long for it to boot? Is it 'cuz
of all the decrunching?
I'd really like a 3.1 software solution to work with my A2091 if at all
possible. As I mentioned before, TwinKick will go into constant reset mode too when the A2091 is plugged in and after I've tried disabling autoboot, slower drive, etc. :-(
LMK if you need some beer zapped your way or maybe something
along the lines of a PayPal donation to help in the quest of getting it to work! :pint:
-
A2091 (http://www.amiga-hardware.com/showhardware.cgi?HARDID=1161)
"Full-length Zorro-II DMA SCSI hard drive controller that was originally introduced with the A2000HD."
-
So... does DMA mean that it doesn't place or map the RAM where Kickwork, Twinkick and the like expect it to be?
-
Its related to the configchain.
Twinkick (& Kickwork) "needs" a memory-expansion as first expansion.
The A2091 is configured as a "hdd-controller" first and then as a "memory-expansion" in a second step (and thats to late for Twinkick).
You can try a workaround, but i cant promise that it works.
(Its just an idea ...)
Set the a2091 memory-jumper to "0kb" and give it a try.
-
Its related to the configchain.
Twinkick (& Kickwork) "needs" a memory-expansion as first expansion.
The A2091 is configured as a "hdd-controller" first and then as a "memory-expansion" in a second step (and thats to late for Twinkick).
You can try a workaround, but i cant promise that it works.
(Its just an idea ...)
Set the a2091 memory-jumper to "0kb" and give it a try.
Thanks Ratte. Good idea, but I just tried it with no luck. Was even thinking I could build a switch and in-between resets, switch her to the full 2mb, but all the A1000 wants to do is constantly reset after loading KS 1.3. That's without a HD connected. If a HD is connected, it wants to boot straight away after loading KS 1.3. Looks like I am DJBase bound again for a KS adapter :-)