To Vampire or not

[Iggy]

Adding to what Djole just wrote, there was a comment in the now deleted and restarted FPU thread on the Apollo-Core forum that modern software for floating point seldom uses raw FPU functions anyway but instead usually uses SSE2 on the Intel as a minimum so that 4 times the computation can take place.  I doubt there is room on the current generation of FPGAs to do a full floating point vector unit so it is likely that Gunnar is just finishing up what he can do with the current generation before he even thinks about starting the next generation.  Sometimes I wish I had his discipline to finish one thing before starting the next.

Thanks Sam, that's all I've ever said (or whined depending on your POV :angry:).

Okay.  I have my answer and really appreciate those who were able to stay on target.

However, this thread has become just another fanboy argument so it's time to ask the moderators to CLOSE THIS THREAD.

PLEASE!!!!!!!!!!!!!!!!!!!

"so it's time to ask the moderators to CLOSE THIS THREAD"

I think I mentioned that awhile ago (that you'd already had your answer), so you have a point.
But I don't let character assassination go easily.

And since apps under AROS68K often require an fpu...

Anyway, I guess we aren't supposed to ask for it, even if its been mentioned as an intended feature.
So, for now (since the guy that started the thread DID get his answer), I'll let it drop.
And those that want to remind Kolla, SamCrow, myself (or anyone else that's brought it up) what b*stards we are for mentioning it, go for it (he whined ;-) ).

[wawrzon]

And since apps under AROS68K often require an fpu...

interesting. how have you discovered that? own experience or assumptions as customary among aros complainers?

[Iggy]

interesting. how have you discovered that? own experience or assumptions as customary among aros complainers?

Fascinating...doing your Vulcan impression, are you?
Perhaps requires is too strong a phrase, would you prefer can use (and often runs better with) an FPU?

Funny, I didn't think it was much of a complaint, asking for support for instructions that have been with us for decades.

Are you sure that isn't a matter of you being an apologist vs my being a "complainer"?

[talybont]

Yes, scary is definitely the word.
I might go as far as crazy, but that would just bring on the flames.

Hmm...the question becomes, can I rely on someone with this level of stability?

But I don't let character assassination go easily.

And to quote the good book: "Do to others as you would have them do to you."

I'd also rather see this thread get locked.

[Iggy]

And to quote the good book: "Do to others as you would have them do to you."

I'd also rather see this thread get locked.

Oddly enough, so would I, as I (myself) seem to be the ping-pong ball.
And to quote another passage "An eye for an eye, and a tooth for a tooth".

I can do that just as well as you...
Now, can you explain Isaiah 45:5 to me, since it seems to credit God with disaster and the creation of evil.

[wawrzon]

Fascinating...doing your Vulcan impression, are you?
Perhaps requires is too strong a phrase, would you prefer can use (and often runs better with) an FPU?

aros software requires fpu as much as amiga software. that is usually then when its sensible. it doesnt require fpu by design. on aros 68k you would likely mostly want to use genuine amiga software, simply because it is a much larger choice. in this case requirements would simply be the same.

Funny, I didn't think it was much of a complaint, asking for support for instructions that have been with us for decades.

Are you sure that isn't a matter of you being an apologist vs my being a "complainer"?

no. at this point im simply informing you about the state of affairs and correcting your wrong assumptions according to the subject i have obviously more experience with than you.

[Djole]

yeah the ones who have run out of arguments (give us FPU now) would like to get this thread closed now.  Give us FPU or we will hate you is the main argument. If you dont, you are a dork and gay of course, liar, amiga hater, cheat and a AROS lover. BUrn mf....

[Zooz]

The work on the core is just going well, everyone who connects to the IRC knows on what the guys are working at moment, and no it is not a 'club', like if it was 'closed'. They are just some amiga enthousiasts, like we all are. There are IRCs for plenty amiga stuff, the apollo-team is just one among others, and is not as close as some seems to think, many vampire users also owns another amiga config. I myself have a morphos machine, a 1200 + blizz, and yes a vampire, all 3 are just excellents systems in there areas. FPU is tiring discussion as some here have no idea of what are the current investigations / brainstormings / core remaining spaces / ... FPU exists but do not fits in current shape, some choice are done, not because team is unskilled or mad or stupid or frightening but because she had to do with constraints; all developers knows of what can be constraints. So please, be patient, be smart and just nice. One other thing, it seems people still thinks that the core is a "one-man core" made, public should change their mind about this. Some members actually works on the core, at VHDL level, i would like also thanks these guys who are doing great work silentely and contributes to bring visible progresses and open-minded spirit inside core team. 4 or 5 brains are always better than 1, be sure the team is aware of this.

About the initial question, A2000 and Vamp, i advise anyone interested in to watch ShK videos, that use a V500 daily in an A2000, aside some real amiga hardware (Delfina, X-Surf, Prisma, PIV, Oktagon, ...). Looking at them can convince or not, dependings of what is wished at end. But at least it proves it works and well - with some clean system (as for any amiga system/configuration). Core is much more compatible than people seems to think.

For example (and sorry if already pasted) :

https://vimeo.com/200954816

https://vimeo.com/185116764

https://vimeo.com/206998554

https://vimeo.com/195123926

Looks carefully at the videos descriptions.

[Iggy]

aros software requires fpu as much as amiga software. that is usually then when its sensible. it doesnt require fpu by design. on aros 68k you would likely mostly want to use genuine amiga software, simply because it is a much larger choice. in this case requirements would simply be the same.



no. at this point im simply informing you about the state of affairs and correcting your wrong assumptions according to the subject i have obviously more experience with than you.

Your grasp of the "obvious" misses a few points.
A developer targeting AROS specifically, may very well be porting code from X86 AROS to AROS 68K, so floating point routines would be likely.
AROS while it is an AmigaOS analogue, should have to potential to surpass AmigaOS in capability.
And as about 20% of all Amiga software supports an FPU, its entirely possible that software specifically designed with AROS in mind could exceed that percentage.

SO, before you write off at least 1/5th of all of our software...

Also, as to your experience level, with AROS 68K I have no doubt, with 68K platforms in general, its really unlikely.
My company was building its own hardware when most of you were focused on playing games with your A500s.
Which appears to be the same focus many of you still have, since those are the demos you show on YouTube.
And since games rarely require a FPU, you're not really likely to be too worried about it. Are you?

Now as for something like Lightwave or Aladdin4D...
(Or anything that might require heavy mathematics)...
Obviously the Vampire (running AROS OR AmigaOS) is still going to be at a disadvantage.

SO, instead of trying to distract others from the point I've made, why not stop being an apologist, and admit that floating point support would be useful?

This isn't a "wah, you're not being fair" issue, or a "uh, you just don't like AROS" issue. Its a simple matter of a feature set being missing from your cpu. Its a HARDWARE issue (btw - that's the topic, hardware).

When you're ready to stop pointing fingers or pontificating...

[kolla]

It's not just "heavy mathematics", it's random software compiled for 040 and 060, for which the compiler used assumes FPU to be available and hence may use it for optimization. It doesn't matter if the usage is heavy or not, the use of FPU instructions can be quite arbitrary when there is software meant to run on 040 and 060 in use. To avoid the 8000 000B crashes, users must be very aware what they install and run, with AC68EC080 most often being detected as 040 or 060 by installers, the user must override what is detected and insist on 68000/020/030 if they exist.

The most worrisome 8000 000B I have seen yet was on file system validation, on a system that afaik was "clean" for all FPU software (duplicated straight from my MIST), had to work around it by disabling partition in early startup and fix from hdtoolbox...

But whatever, these are all worries that AmigaKit etc will have to deal with now.

[kolla]

yeah the ones who have run out of arguments (give us FPU now) would like to get this thread closed now.  Give us FPU or we will hate you is the main argument. If you dont, you are a dork and gay of course, liar, amiga hater, cheat and a AROS lover. BUrn mf....

You have no concept of what hate is, you can rest assured that I have zero hate for any people involved with Amiga. Stop being so damn polarized, and rather spend time actually using Amiga software and be creative.

[wawrzon]

Your grasp of the "obvious" misses a few points.
A developer targeting AROS specifically, may very well be porting code from X86 AROS to AROS 68K, so floating point routines would be likely.

you dont need to hypothetize about all that. aros has a pool of software in contribs and ports that compile for different targets a the same time. x86, arm or plain mc68000. check yourself.

AROS while it is an AmigaOS analogue, should have to potential to surpass AmigaOS in capability.

it does.

And as about 20% of all Amiga software supports an FPU, its entirely possible that software specifically designed with AROS in mind could exceed that percentage.

as i said. if necessary.

SO, before you write off at least 1/5th of all of our software...

Also, as to your experience level, with AROS 68K I have no doubt, with 68K platforms in general, its really unlikely.
My company was building its own hardware when most of you were focused on playing games with your A500s.
Which appears to be the same focus many of you still have, since those are the demos you show on YouTube.
And since games rarely require a FPU, you're not really likely to be too worried about it. Are you?

i habe not put any videos on youtube..

Now as for something like Lightwave or Aladdin4D...
(Or anything that might require heavy mathematics)...
Obviously the Vampire (running AROS OR AmigaOS) is still going to be at a disadvantage.

cant you think of some more original application example then raytracers? that btw usally even had integer binaries on amiga. web browser? or maybe fractal generator?

SO, instead of trying to distract others from the point I've made, why not stop being an apologist, and admit that floating point support would be useful?

certainly. i dont discuss that, just your other mistakes.

This isn't a "wah, you're not being fair" issue, or a "uh, you just don't like AROS" issue. Its a simple matter of a feature set being missing from your cpu. Its a HARDWARE issue (btw - that's the topic, hardware).

When you're ready to stop pointing fingers or pontificating...

as soon as you stop spreading wrong assumptions about aros 68k as facts, you obviously know nothing of.

[psxphill]

And as about 20% of all Amiga software supports an FPU, its entirely possible that software specifically designed with AROS in mind could exceed that percentage.

Using floating point variables is very common because (at least at some point) some x86 fpu operations were faster than integer & it is essentially lossy compression, which helps cache hits.

Obviously you can compile them to use software routines, or you could trap the errors and then patch the code. But the current line is that FPU will come eventually, so doing anything other than waiting seems a waste of resources. If the line is just a diversion tactic then it does mean that we lose out in the long term, but you can't control people.

[trixster]

Anybody who is feeling "hatred" over some aspect of the vampire project needs to get a grip.

[grond]

Do you guys really believe it takes your input to make a professional CPU developer and long-time Amiga user understand the value of a compatible FPU or compatibility with as much of the Amiga software catalogue as possible?

If so, you must have a strange idea of the level of intelligence, dedication and hard work it takes to develop a CPU as complex as some post-Pentium Intel CPU which usually were developed by a team of more than thousand engineers. And the FPU is STILL not done, what is Gunnar THINKING!!!

[kolla]

Do you guys really believe it takes your input to make a professional CPU developer and long-time Amiga user understand the value of a compatible FPU or compatibility with as much of the Amiga software catalogue as possible?

It has worked before, so why not - the trick is to somehow make him think that it is his idea.

And the FPU is STILL not done, what is Gunnar THINKING!!!

It is clearly done, he has said so many times now, and besides, 3 years ago, it was even available...

http://web.archive.org/web/20140516224526/http://apollo-core.com/

Optionally, a fully pipelined, double precision FPU is available to be included in the Core.

So the questions is - is this professional CPU developer and long-time Amiga user able to do it, or is he not?

What happens with the already finished FPU implementation? Will it be available for sale? If so, how much?