26-Apr-2024 15:03 GMT.		 UNDER CONSTRUCTION
Anonymous, there are 126 items in your selection (but only 26 shown due to limitation) [1 - 50] [51 - 100] [101 - 126]
[News] Eyetech clarifies AmigaOne statementANN.lu
Posted on 31-Oct-2001 15:50 GMT by Teemu I. Yliselä126 comments
View flat
View list
Alan Redhouse of Eyetech clarified some of the confusion surrounding their latest AmigaOne statement in a comment here. I've posted it again below for those who don't read the comments section.

I try to use the clearest possible English in the status update but I have obviously failed to get the message across properly to everybody :(

Nowhere in my update does it say that OS4.0 had not started - and of course quite a lot of work *has* been done - but that the original plan required funding for OS4.0 to be *FINISHED*, and Amiga Inc had other priorities with their restricted funding. Bill has made no secret of the financial constraints that Amiga Inc were under at his public presentations at St Louis and Sacramento this year.

What I said is that WE decided to put development of the A1 on ice UNTIL we had a guarantee of a FINISH date of OS4.0. Of course development did not suddenly halt immediately, as we were all expecting Amiga Inc to obtain funding 'any day now'. But eventually other revenue-earning priorities took over.

I have absolutely no difficulties with any of Amiga's decisions or actions. They are exactly the same sort of business decisions that I would have made in their position.

We went into the AmigaOne project with our eyes wide open knowing the risks associated with events outside our control, and in no way hold Amiga Inc responsible for OUR decision to suspend development of the A1. In fact - again as I said in our update - we have all been trying very hard to obtain an all round satisfactory resolution to an OS4.0 completion date that would allow us to complete the A1 development and roll out into production. And no, there is no point in announcing that a potential crisis exists whilst there is still very real progress towards a resolution being made.

We obtained the agreement in principle earlier this week. This obtained, I wanted to set the record straight immediately to dispel some of the rumours and make sure that nobody made long journeys to WoA-SE on false expectations. As it is WoA-SE attendees will witness an historic contract signing.

Much of OS4.0 was planned to be implemented using the CSPPC in parallel with the development of the AmigaOne hardware. The decision to expand the market for OS4.0 was an integral part of the 'no upfront cost to Amiga' part of the agreement.

Hope this helps

Alan

Eyetech clarifies AmigaOne statement : Comment 101 of 126ANN.lu
Posted by Brecht [darklite] on 01-Nov-2001 20:12 GMT
In reply to Comment 92 (Emmanuel Lesueur):
>All this is possible to do, but the question is: has it already
be done ?
I think there are a few demos included on the Amithlon CD. I don't know how this works in detail, so you should mail Bernie if you want more info :)
Eyetech clarifies AmigaOne statement : Comment 102 of 126ANN.lu
Posted by Brecht [darklite] on 01-Nov-2001 20:25 GMT
In reply to Comment 99 (Alcemyst):
oh god, are we there with the stupid idea that people will buy amithlon and then use windows instead. Most Amiga users already have a windows machine, wake up!
Let's just see how it turn out instead of making these wild guesses.
Eyetech clarifies AmigaOne statement : Comment 103 of 126ANN.lu
Posted by Richie on 01-Nov-2001 20:30 GMT
In reply to Comment 54 (priest):
Upgrade to JIT, it's not slow.
Eyetech clarifies AmigaOne statement : Comment 104 of 126ANN.lu
Posted by David Scheibler on 01-Nov-2001 20:40 GMT
In reply to Comment 100 (Graham):
To point4:
The only emulator that actually exists (to my knowledge) is the old
H&P emulator which bases on WarpUP. You boot then the emulator resets
and takes over the hardware and starts WarpUp then you boot AmigaOS in
this "WarpUp-68k-emulation box". Of course for OS4 they speak about a
new ppc exec but this emulator is the only one they can use at the
moment.
Eyetech clarifies AmigaOne statement : Comment 105 of 126ANN.lu
Posted by Alcemyst on 01-Nov-2001 20:55 GMT
In reply to Comment 102 (Brecht [darklite]):
brecht you have a habhit of saying what you think ppl are saying
& not actualy what they did say. like with the tempory statement.
& tempory does not in all cases mean take away later. some tempory
things are built around or built on top of.
& your reply to my comment is stupid in its self as i know most PPL
who have a PC use windows infact most ppl asume that ppl who own a PC
use windows
& where did i say in my own word that ppl will use windows instead of
Amithlon ? i didnt . you have taken a very small part of my post &
twisted it all out of context. if you read it fully you will see it
was with the SW house not doing A Amithlon version if they already
have alread made Windows version as they will think as what most ppl
will think if you have PC HW YOU WILL MORE THAN LIKELY HAVE WINDOWS
ALREADY & YOU MITE AS WELL BOOT INTO WINDOWS & RUN THE WINDOWS VERSION
& SO WE WILL NOT BOTHER MAKEING A AMITHLON VERSION.
Eyetech clarifies AmigaOne statement : Comment 106 of 126ANN.lu
Posted by Emmanuel Lesueur on 01-Nov-2001 21:08 GMT
In reply to Comment 97 (Georg Steger):
Georg Steger writes:
> as far as AmigaOS and Scheduling is
> concerned, x86 tasks and 68k tasks are identical
Ok, then they have a 'mixed mode' system as MorphOS has.
> About compiling x86 native code: IIRC they said you need to compile
> them under Linux. But they probably do not (yet) have a special
> compiler which helps with the endianess issues.
Well, if they have the mixed mode, the main problem is solved.
Hacking gcc is not that complicated.
> And Bernd responded, that this
> will be explained/documented in the finished product, but that
> generally it does not really make sense to compile *that* kind of
> code (struct NewWindow nw; setup nw; winOpenWindowTags; x =
> win->LeftEdge) x86 natively.
To me it makes sense. If only to allow easy porting of stuff where
speed really matters. If one has to separate 68k and x86 code,
then it's the powerup/warpup story again (without context switches):
only few apps because porting is a lot of work.
Eyetech clarifies AmigaOne statement : Comment 107 of 126ANN.lu
Posted by Brecht [darklite] on 01-Nov-2001 21:14 GMT
In reply to Comment 105 (Alcemyst):
>brecht you have a habhit of saying what you think ppl are saying
¬ actualy what they did say. like with the tempory statement.
&tempory does not in all cases mean take away later. some tempory
things are built around or built on top of.
No, only in 99% of cases.
> & your reply to my comment is stupid in its self as i know most PPL
who have a PC use windows infact most ppl asume that ppl who own a PC
use windows
No, I said that most Amiga users also have a PC.
> & where did i say in my own word that ppl will use windows instead of
Amithlon ? i didnt . you have taken a very small part of my post &
twisted it all out of context. if you read it fully you will see it
was with the SW house not doing A Amithlon version if they already
have alread made Windows version as they will think as what most ppl
will think if you have PC HW YOU WILL MORE THAN LIKELY HAVE WINDOWS
ALREADY & YOU MITE AS WELL BOOT INTO WINDOWS & RUN THE WINDOWS VERSION
&SO WE WILL NOT BOTHER MAKEING A AMITHLON VERSION.
wow, and all that in one sentence.
Calm down, no need to get all excited. I'm sorry, I didn't read your comment very thoroughly. It's just that it's so hard to read without the use of caps at the start of sentences ;)
Eyetech clarifies AmigaOne statement : Comment 108 of 126ANN.lu
Posted by Alcemyst on 01-Nov-2001 21:24 GMT
In reply to Comment 107 (Brecht [darklite]):
brecht
yet another stupid reply as to run Amithlon you have to have a PC.
& if you read my full post again it is pritty clear that i talked abot
PPL who have an Amiga AS well as a PC.
Eyetech clarifies AmigaOne statement : Comment 109 of 126ANN.lu
Posted by Alcemyst on 01-Nov-2001 21:31 GMT
In reply to Comment 107 (Brecht [darklite]):
also you are to young to know how often tempory things become perm
Eyetech clarifies AmigaOne statement : Comment 110 of 126ANN.lu
Posted by Brecht [darklite] on 01-Nov-2001 21:53 GMT
In reply to Comment 108 (Alcemyst):
>yet another stupid reply as to run Amithlon you have to have a PC.
& if you read my full post again it is pritty clear that i talked abot
PPL who have an Amiga AS well as a PC.
Nothing you type is 'pritty' clear because of the way you write.
>also you are to young to know how often tempory things become perm
& whatever you say, gramps.
Eyetech clarifies AmigaOne statement : Comment 111 of 126ANN.lu
Posted by Alcemyst on 01-Nov-2001 22:11 GMT
In reply to Comment 110 (Brecht [darklite]):
you should try posting with V3 beta & you will see how badly your text
comes out.
& NO in no where near that age to be called that but the fact is you
are to young to know many things & them things can olny be lernt with
time. there is no shortcuts & no one can show you, you have to go
through it your self.
& i dont see any one eles complaning about my poor typing skills
Eyetech clarifies AmigaOne statement : Comment 112 of 126ANN.lu
Posted by Anonymous on 01-Nov-2001 22:36 GMT
In reply to Comment 100 (Graham):
> 3) Zico is a definition of the hardware that the next generation Amiga would contain. I said it was out of date and not concise enough, yet it is clearly a definition of what would be in the next Amiga. A pretty obvious definition, yes. But people here were arguing that an Amiga was defined as some 68k machine with AGA (in essence) and I was pointing out that the definition of a computer's hardware changed over time.
Are you sure? AFAIK the Zico spec was just for the developer box and has nothing to do with the new Amiga hardware.
Eyetech clarifies AmigaOne statement : Comment 113 of 126ANN.lu
Posted by Anonymous on 02-Nov-2001 00:52 GMT
In reply to Comment 112 (Anonymous):
In fact Zico was for AmigaDE (with different host processors).
Eyetech clarifies AmigaOne statement : Comment 114 of 126ANN.lu
Posted by Graham on 02-Nov-2001 01:26 GMT
In reply to Comment 113 (Anonymous):
Yes, that is true, I remember it now. That was when AmigaDE was going to run native on hardware though - this feature disappeared to be replaced by AmigaOS 4 as a PPC host platform running on the same hardware featureset.
At least we are not in the same position as the Acorn crowd. That is nearly dead, the new OS doesn't support all the hardware apparently, and there are issues and it is further behind than AmigaOS. There is no IDE, for example, and the platform is erratic to program for and has architectural issues that limit it today, much in the same way that lack of MP/VM limits AmigaOS until you go to the emulation (or whatever you want to call it) stage for classic programs and just restart from scratch again otherwise.
Eyetech clarifies AmigaOne statement : Comment 115 of 126ANN.lu
Posted by TheArrogantSarny on 02-Nov-2001 04:14 GMT
In reply to Comment 58 (Brecht [darklite]):
>> No, if it runs all AmigaOS software 100%, and it has a licence from Amiga
>> Inc. to be labelled "AmigaOS-compatible" then it is "AmigaOS-compatible".
>That's a political issue, not a technical one.
There's no such thing as 100% compatible. Although you're referring
specifically to the OS it's the equivalent of saying an A1200 is not a REAL
Amiga because it can't run ALL A500 software.
Now, baring the fact that there is no such thing as 100% compatibility in
mind, using compatibility as an indicator of whether something is an Amiga or
not is a red herring. Is an A1200 an Amiga or not? No-one questioned
whether the A1200 was an Amiga or not when it was launched simply because it
carried a five letter label - "AMIGA". If Atari had release an Amiga
compatible at the same time labelled something like "OMEGA" would anyone have
questioned it was an Amiga? HELL YEAH!
Applying this to operating systems. As far as I'm concerned, there can only
EVER be ONE AmigaOS and that is the one LABELLED "AmigaOS". Sure, there may
be compatibility problems between versions. But I've already established
above that 100% compatibility is a red herring so as far as I'm concerned
it's not an issue - AmigaOS 4.x IS AmigaOS.
So, should we allow competitors to label their products as AmigaOS Compatible?
No. Why? Look at the Linux market with all it's variations. There ARE
compatibility problems between various implementations and assuming Linux
ever makes it onto the 'average' users desktop this will lead to problems.
We should not make the mistake of encouraging Amiga inc. to offer competitors
a licence to label their efforts as "AmigsOS Compatible" quite simply
because it is not the OS you should be certifying as being AmigaOS compatible
- it's the software. i.e. "MyApp2000(TM) is AmigaOS 4.0 compatible". If I see
that, I will be confident it will run on AmigaOS 4.0 but that is no gaurantee
it will work on a rival like MorphOS, AROS, etc. and if it doesn't work on
them then it is no fault of Amiga or the application developer as they will
have met their claim of being AmigaOS compatible not "never stands a chance
in the mainstream" MorphOS/AROS compatible. Which is not to imply either
MorphOS or AROS are worse than AmigaOS. Whether or not MorphOS or AROS are
superior I do not know and I do not care as in the mainstream it is the label
"AmigaOS" that carries the weight not some unheard of label like "MorphOS"
and "AROS"
Of course Amiga could make the mistake of doing what you suggest which is
allowing rivals to label their efforts as "AmigaOS compatible". But what does
this mean?
Quite simple really. One organization would have to act as the driving force
to ensure there was consistency between all the variations and because Amiga
are the OFFICIAL and LEGAL owners of AmigaOS it means they are the ideal
choice. I don't know for a fact, but the impression I get from these forums
is that MorphOS and AROS developers/followers would not like this as it would
imply Amiga would ultimately have the final say over which way they took the
OS. Tough titty. If they don't like it then "Get the hell out of our galaxy."
However, this approach would ultimately lead to failure quite simply because
of the issues raised above: There would inevitably be incompatibility
problems between the various implementations even with a single organization
with ultimate responsibility over development and like I say... it is the
software you should be certifying not the 'variations'.
I for one would rather be able to go into a shop one day and pick a product
off a shelf that says "AmigaOS Compatible" safe in the knowledge that it WILL
run on my OS than pick it up thinking "Oh, I hope it runs on MorphOS/AROS,
they claim to be AmigaOS compatible".
Back in the real world. The thing to remember at the end of the day is that
Amiga have NO say and don't look like they ever will have any say over
MorphOS/AROS development and inevitably there WILL be compatibility issues
between them so offering licenses to these organizations to label their
efforts as "AmigaOS compatible" would be futile and only bad for the Amiga
industry as a whole.
TheArrogantSarny
Eyetech clarifies AmigaOne statement : Comment 116 of 126ANN.lu
Posted by Menthos on 02-Nov-2001 06:28 GMT
In reply to Comment 52 (Brecht [darklite]):
"No, but it would be an x86 CPU, that's my point. "
Bad example... Take this one:
IBM's OS/2 (or whatever the name was) could run some Win-apps (if I don't remember it wrong), should that make it Windows? NO, IT IS OS/2!!!! =P
But hey, I can be reasonable... You can call anything you want AmigaOS but that just won't make it right... =)
Eyetech clarifies AmigaOne statement : Comment 117 of 126ANN.lu
Posted by Harry Kok on 02-Nov-2001 09:46 GMT
In reply to Comment 110 (Brecht [darklite]):
Alchemyst, 10 points for number 99!
Unfortunately, you seem to have missed:
- Sorry, I'll read your post better next time - (107)
..And reply with something like:
- Yet another stupid reply -
Which makes it no suprise Brecht then goes like:
- Whatever you say, gramps. -
*sigh* There's a flipside to every coin I guess.
Number 99 is still very good though.
Eyetech clarifies AmigaOne statement : Comment 118 of 126ANN.lu
Posted by Bernd Meyer on 02-Nov-2001 10:32 GMT
In reply to Comment 42 (Graham):
>Amithlon does not have native hardware support - that is provided by Linux, and abstracted by
>Amithlon. Amithlon does not contain the drivers, the hosting OS does.
That's not correct.
* Linux provides display and IDE/SCSI drivers for Amithlon. AmigaOS, however,
provides both the P96 system as well as the filesystems and/or SCSI-Direct
apps to work on those abstracted devices. In both cases, we are just using
linux for the very lowest abstraction layer, leaving things to AmigaOS side
things as soon as possible, for maximum Amiga-feeling.
* Amithlon itself provides PS/2 keyboard and mouse support (linux has nothing
to do with it), which can easily be disabled in favour of letting drivers
loaded under AmigaOS play with the hardware. But of course, we needed some basic
support while AmigaOS isn't up and running yet. The included program "ps2test"
shows how to do this.
* The network driver is an AmigaOS SANAII driver directly talking to the NE2000 hardware
* The sound drivers are AHI drivers talking directly to the AC97 and SB128 hardware
* The serial and parallel port drivers are derived from the drivers for the HyperCom
line of multi-I/O cards, and access the hardware directly.
* The included BT878 TV card demo program runs in 68k code under AmigaOS, directly setting
the TV card's DMA engine.
* The Picasso96 team's Voodoo3 driver runs almost without change on Amithlon,
accessing the Voodoo3 card directly.
* Except for the memory needed by Amithlon itself, all of the PCs memory is
invisible to the linux kernel, and directly under AmigaOS control.
* In Amithlon, memory space as seen by the 68k is the same as memory space seen
in x86 mode, as well as hardware memory space. So if you use the direct hardware
access to set up some PCI DMA transfers to address X, they will show up at
address X in the Amiga side of things.
* In Amithlon, PC hardware interrupts can be mapped to 68k interrupts;
Effectively, one can say "whenever PC interrupt X occurs, please send me
68k interrupt level Y"
Also, even though we are driving most video hardware through the linux-supplied
framebuffer devices, there still isn't any extra layer between the CPU and video
memory (as is the case in just about any other emulator solution I am aware of).
The video memory the 68k writes to is the actual video memory of the card.
The philosophy behind Amithlon is not to emulate an Amiga on another OS, but
to turn an x86 into an Amiga. A tremendous amount of work and hacking (and
debugging :) went into making exactly those things I listed above possible.
Heck, if you wanted to, you could write 68k SCSI drivers for your favourite
card, and just remove the support for that card from the Amithlon linux kernel
(which you can do --- that kernel is GPL-ed, and you get full source with your
copy of Amithlon). No problem.
Eyetech clarifies AmigaOne statement : Comment 119 of 126ANN.lu
Posted by Bernd Meyer on 02-Nov-2001 11:07 GMT
In reply to Comment 60 (Emmanuel Lesueur):
>and I want to have the fastest hello world program ever,
>running under AmigaOS. How do I compile it for x86 ?
Let's use a slightly more useful example (please note that I'll leave
out #includes and other necessary candy, for the sake of readability,
and brevity):
---------------------- 68k only program ---------------------------
unsigned long fib(int x) {
return x>2?fib(x-1)+fib(x-2):1;
}
int main(int argc, char** argv) {
int n=argc>1?atoi(argv[1]):10;
printf("Fibonacci number %d is %lu\n",n,fib(n));
}
--------------------- end 68k only program -------------------------
What you do is to split this off into parts that should run on the 68k side
(here: the main() function, handling user parameters and output), and parts
that should run on the x86 side (here: The fib() function), and add a little
glue to the 68k side to put them together:
-------------------- 68k side program -----------------------------------
int main(int argc, char** argv)
{
int n=argc>1?atoi(argv[1]):10;
elfObject eo=open_elf("fibonacci.x86");
call_native fib=find_slowcall(eo,"fib");
printf("Fibonacci number %d is %lu\n",n,fib(n,0,0));
}
------------------- end 68k side program --------------------------------
(Note that in real life, you need to check the return values of open_elf
and find_slowcall, of course, and really need to make sure you are running
on Amithlon before doing any of that stuff --- a simple "check_emulation()"
will do that, or "assert_emulation()" will even handle printing an error message
and quitting on non-Amithlon systems). Call_native type functions *always*
take three parameters, that's where the extra ",0,0" comes from.
Next, the x86 side:
------------------ x86 side program --------------------------------------
unsigned long fib(int x) ELFFUN;
unsigned long fib(int x) {
return x>2?fib(x-1)+fib(x-2):1;
}
------------------ end x86 side program ----------------------------------
That's it. Compile each on the respective architecture (yes, this is clumsy
right now; Ideally, one needs a second machine running linux/x86 for software
development and have them networked --- which is my setup :). Run. Done.
Now, how does this differ from using x86 code under UAE:
* While the x86 code is running (and I chose this horribly inefficient
implementation of fib() for a reason --- it might run for several
seconds for large n, without ever returning into 68k mode), the AmigaOS
scheduler is still fully in control. There is *no* difference, from
the scheduler's point of view, between a task in x86 mode and a task in
68k mode.
* The "fib" function pointer (which is what "call_native" is tyepdef'ed to)
is transparent --- a *good* implementation of the above would provide a
68k side version of "fib()", and assign *that* function to the function
pointer if the .x86 file couldn't be opened or didn't contain the required
function. The x86 part only comes into play at initialization time, the rest
of the program just dereferences a function pointer
* You can pass pointers between x86 and 68k side of things, and they actually
*mean* something (because the address spaces are identical)
* Once you are in x86 code, you can in turn call 68k code, which can call
x86 code, which can call 68k code, which.... you get the idea. In other
words, if you want to allocate 1kB of Amiga memory in Amithlon, in x86 code,
all you do is "mybuf=AllocMem(1024,0);". And it doesn't really matter whether
AllocMem has in turn been patched to use x86 code, which might use calls
to 68k routines, which....
Trust me, this interface has been thought through a lot. It is also the second
one. I learned what was wrong with the first one, and then did a proper one the
second time around. And I am using it myself. A LOT :)
Eyetech clarifies AmigaOne statement : Comment 120 of 126ANN.lu
Posted by Bernd Meyer on 02-Nov-2001 11:30 GMT
In reply to Comment 79 (Graham):
>Yes, an AmigaOS driver that interfaces with the Linux driver underneath, not directly with the
>hardware. Hence support for any ethernet or audio hardware.
Wrong, sorry. I don't like too many layers of abstraction, especially for
sound. Thus, the above drivers *do* speak directly to the hardware --- with
DMA, interrupts and all. No linux drivers in sight. Same for parallel and serial
ports. PS/2 mouse and keyboard is not handled by linux drivers, either, but
instead directly by Amithlon.
Which is exactly why lumping in Amithlon with UAE or AmigaXL (which is UAE for QNX :) is such a fallacy --- those emulators run as applications on bog-standard kernels.
Amithlon's kernel is full of decidedly custom hacks, which make a lot of Amithlon's cool features possible. It also makes it a very very bad idea to run anything else on that kernel, but who cares --- nothing else is meant to run on it.
With UAE, one needs to play nice with the underlying OS. With Amithlon, I have absolute control oever every aspect of the software that is running. Which means that so much cooler stuff is possible.
And maybe to dispel another popular misconception: Amithlon is not fast because it got rid of the custom chip emulation, or because it doesn't have to deal with Windows. Either of those only brings about 5% speedup even on moderate hardware, less on really fast stuff.
Amithlon is as fast as it is because of some very complicated and intricate changes to the JIT compiler. Those changes only became *possible* due to getting rid of custom chip emu and using a custom kernel --- but there is a big difference between making something possible and making something happen!
Eyetech clarifies AmigaOne statement : Comment 121 of 126ANN.lu
Posted by Ralph Schmidt on 02-Nov-2001 11:38 GMT
In reply to Comment 119 (Bernd Meyer):
@bernd
That`s basicly how powerup/warpup worked. It`s not simular to the mixed
mode stuff in MorphOS where there is actually only a PPC exec which
only runs PPC tasks...68k programs==code are just handled like
subroutines inside the PPC tasks contexts.
This whole ppc a-box enviroment is a process under the quark kernel...
think of it as a big fat driver "process" with its own virtual 32bit
addressspace.
Therefore there is no real difference how software needs to be written
for 68k or ppc as it`s basicly the same.
Eyetech clarifies AmigaOne statement : Comment 122 of 126ANN.lu
Posted by Graham on 02-Nov-2001 12:05 GMT
In reply to Comment 118 (Bernd Meyer):
Thank you Bernd for your interesting and informative posts.
Eyetech clarifies AmigaOne statement : Comment 123 of 126ANN.lu
Posted by Bernd Meyer on 02-Nov-2001 12:12 GMT
In reply to Comment 121 (Ralph Schmidt):
@ralph
I agree, the way MorphOS handles things is rather different. Considering the different angles and circumstances, that is not surprising. You guys essentially have PPC code as your "normal" environment, and drop into 68k mode when required; Amithlon has 68k "normally", and can optionally drop into x86 mode.
I chose the "two files, explicit load at runtime" approach for a number of reasons. The main one being that I most certainly do not want to encourage anyone to write programs for Amithlon, and Amithlon only. If I tried that, and was successful, the oft-quoted splitting of the market would have happened, and everybody loses. BAD idea!
Instead, I made sure that it is as easy as possible to *optionally* add support for special Amithlon-optimized versions of routines to existing and future AmigaOS programs. The important part being that these programs should, without modification, also work on 68k Amigas, or under MorphOS, or OS 4.x, or anything else that runs AmigaOS software.
The way Amithlon calls x86-native routines from 68k should be transferable to calling PPC-native routines from within OS 4.x's 68k emulator, or (if you would like to) calling PPC native routines from within MorphOS's 68k emulator. Which, if it came to pass (and I will provide any and all support required to make that happen, if the other parties are interested) would mean that it would be trivial to provide not only optional x86 routines, but also optional PPC routines.
Of course, I can't close this comment without pointing out a *major* difference between Amithlon's solution and WarpUP/PowerUP --- in Amithlon, there is *one* task, which switches modes. Under WarpUP/PowerUP, there are *two* tasks (one on the 68k, one on the PPC), which need to do all sorts of nasty synchronization stuff (including cache flushes, due to lack of cache coherency) each time they want to communicate with each other.
On my Athlon, calling the x86 fib() function with a zero argument would take about 200 nanoseconds. This is orders of magnitudes faster than anything you could hope to achieve with the dual processor setup.
Eyetech clarifies AmigaOne statement : Comment 124 of 126ANN.lu
Posted by Ralph Schmidt on 02-Nov-2001 12:27 GMT
In reply to Comment 123 (Bernd Meyer):
@bernd
Ok..to me the previous explaination looked a lot like the
powerup/warpup design of 2 different enviroments.
Eyetech clarifies AmigaOne statement : Comment 125 of 126ANN.lu
Posted by Alcemyst on 02-Nov-2001 15:27 GMT
In reply to Comment 117 (Harry Kok):
thanks we all make mistakes
but the H missing in my nick in not, as many ppl have Alchemyst &
no one else would be stupid enought to leave the H out.
but there is always an exception :)
Eyetech clarifies AmigaOne statement : Comment 126 of 126ANN.lu
Posted by Steffen Haeuser on 02-Nov-2001 17:44 GMT
In reply to Comment 104 (David Scheibler):
Hi!
> The only emulator that actually exists (to my knowledge) is the old
> H&P emulator which bases on WarpUP. You boot then the emulator resets
The old H&P Emulator is not the only 68k Emulator existing.
OS 4's 68k Emulator will *not* run as a Blackbox.
Ralph Schmidt is not the person first coming to your mind when
you have questions about OS 4. He is the first person coming to your
mind when you have questions about MorphOS.
Steffen
Anonymous, there are 126 items in your selection (but only 26 shown due to limitation) [1 - 50] [51 - 100] [101 - 126]
Back to Top