The vampire incompatiblity discussion thread

[Overflow]

+1 Marlon_

[grelbfarlk]

+2 Kev +.5 Chucky +1.6667 Don Adan
It was at least a year and a half ago that Gunnar said that the FPU was "Done" and it was just disabled because it had a few bugs. But then again AGA was "Done two years ago" as well.

I bought my V500 on the basis that the FPU was going to be working soon. And it was "just waiting on people to write testcases for him" as apparently it was too much effort to download an FPU enabled application and troubleshoot it. Or maybe the FPU code might infect the rest of the core with those confusing and too difficult instructions.

At any rate as others have said the 68EC080 and RTG do work for what they are.

Some in the IRC channel suggested various theories why the FPU isn't there yet, in no particular order here they are:
- There is no room left in the FPGA for a fully compatible FPU in hardware
- It was working enough but wasn't fast enough to satisfy someone's ego
- A fully working FPU while doing FPU work will cause some terrible or just some slight speed penalty to the Integer core or the RTG
- It's too hard
- It's being withheld to only be enabled on the standalone board(forced obsolescence)
- Working on the FPU gives someone dreams of waking up in math class naked


I have no idea if any of these theories have any basis in reality and I don't really care as long as the FPU finally comes around.

[kolla]

Current status seems to be that FPU will only be available, if at all, for new boards based on new and faster FPGAs. Or, if you have some other FPGA project you need it for, you can supposedly buy a license to use it, or something - but of course, you can not buy for your Vampire card. For FPU to be "worth it", someone must write a "killer app" in assembler, that showcases the superiority of the Apollo Core FPU.

[TrashyMG]

@Grelbfarlk - Small leaked information always seems to get taken out of context, people take snippets of information they heard and then run with it. Perhaps in the case of AGA claiming to be done two years ago, the initial design was done, but not the actual FPGA implementation. Right now Gold 3 is around the corner and AGA is on it's way... This is a project ran on free time and passionate people.

Perhaps a full FPU will eventually come, some people may have to wait an iteration or two of vampire hardware. This project has a long plan ahead of itself, however if support and people testing current hardware/software don't happen now it may never happen. Honestly the current vampire boards can do amazing things even sans-FPU and it's at a quarter of the price of prior accelerators.

[BSzili]

Originally Posted by kolla

Current status seems to be that FPU will only be available, if at all, for new boards based on new and faster FPGAs. Or, if you have some other FPGA project you need it for, you can supposedly buy a license to use it, or something - but of course, you can not buy for your Vampire card. For FPU to be "worth it", someone must write a "killer app" in assembler, that showcases the superiority of the Apollo Core FPU.

Isn't it odd that only the FPU has this requirement? As others have said in this thread, there have been a lot of mixed messages regarding this topic. They should just say they don't want to make an FPU, and that's it. At least my work on Hexen 2 didn't go completely to waste now that Cowcat is doing a WarpOS version

[TrashyMG]

@BSzili, They never said they didn't ever want to make an FPU. They may have joked on IRC because Gunnar gets harasses all the time about it. Current hardware is a Cyclone 3. the Stand-alone Vampire and the V1200 is supposed to have beefier FPGAs and may be better targets. But I'm not apart of the Apollo team, so it's just assumptions on my part.

Right now they're focusing on things that can fit the current hardware and things that could make a big impact. AGA, 16-bit audio and such.

[AMike]

I do not really understand the discussion - the Apollo team has a clear vision - that vision is no secret - a smart evolution of the 68k CPU and the Amiga. So if some one don't like this vision - do not buy the Vampire, simply as that. Stuck to the existing hardware or hope that Jens or phase5 develop new 060 turboboards. There is enough space for everybody.

[ptyerman]

Jens stuff doesn't have a FPU either in general so no point waiting for anything from there!

[kolla]

Originally Posted by AMike

I do not really understand the discussion - the Apollo team has a clear vision - that vision is no secret - a smart evolution of the 68k CPU and the Amiga.

I wonder what they would do if someone starts a competing project.

[BSzili]

Originally Posted by TrashyMG

@BSzili, They never said they didn't ever want to make an FPU. They may have joked on IRC because Gunnar gets harasses all the time about it. Current hardware is a Cyclone 3. the Stand-alone Vampire and the V1200 is supposed to have beefier FPGAs and may be better targets. But I'm not apart of the Apollo team, so it's just assumptions on my part.

Right now they're focusing on things that can fit the current hardware and things that could make a big impact. AGA, 16-bit audio and such.

I'm not talking about IRC logs taken out of context, not did I claim that they are never going to do an FPU. I said there were a lot of mixed messages regarding the FPU up to this point.

Originally Posted by AMike

I do not really understand the discussion - the Apollo team has a clear vision - that vision is no secret - a smart evolution of the 68k CPU and the Amiga. So if some one don't like this vision - do not buy the Vampire, simply as that. Stuck to the existing hardware or hope that Jens or phase5 develop new 060 turboboards. There is enough space for everybody.

Oh no, people are discussing things on a forum, help!

[kolla]

Originally Posted by ptyerman

Jens stuff doesn't have a FPU either in general so no point waiting for anything from there!

Well, at least on the ACA1230 cards, I can add one myself fairly easily.

[AMike]

Originally Posted by kolla

I wonder what they would do if someone starts a competing project.

Theoretical question - they have a massive head start. Do you see a company or a group of developer with plans to implement a 68k compatible CPU in a FPGA? The only advanced 68k core I'm aware is the one from tobiflex (tg68) - but tg68 can't compete with the Apollo core.

[ShK]

TG68.C Core
Vampire600 schematics
retro.68k Vampire 600 V1

[kev]

snippets? u want more logs i have them

[kev]

Originally Posted by AMike

Theoretical question - they have a massive head start. Do you see a company or a group of developer with plans to implement a 68k compatible CPU in a FPGA? The only advanced 68k core I'm aware is the one from tobiflex (tg68) - but tg68 can't compete with the Apollo core.

yes i know people decapping 68060 as we speak with FPU

[kolla]

Originally Posted by AMike

Theoretical question - they have a massive head start. Do you see a company or a group of developer with plans to implement a 68k compatible CPU in a FPGA?

Sure, there are several commercial 68k softcore alternatives in the embedded market, and older than the Apollo Core too. But they do not target Amiga.

[AMike]

Originally Posted by kolla

But they do not target Amiga.

Will they change their mind?

[wawa]

Originally Posted by kolla

Sure, there are several commercial 68k softcore alternatives in the embedded market, and older than the Apollo Core too. But they do not target Amiga.

none of those can compete with apollo for what i know.otehrwise there wouldnt be any problem, youcould simply switch to a faster core.

[wawa]

Originally Posted by ptyerman

Jens stuff doesn't have a FPU either in general so no point waiting for anything from there!

id say it is in general much more limited, yet where is that whole discussion about it?

[wawa]

Originally Posted by TrashyMG

Perhaps a full FPU will eventually come, some people may have to wait an iteration or two of vampire hardware. This project has a long plan ahead of itself,

yeah. and there are people who want everything immediately. even if it wanst promissed, rather than waiting for the good stuff and trying to get along with what they have. i have a number of genuine amigas put to task of testing an alternative compatible system to accompany the prespective of what apollo or similar may open ahead of us. i dont need to have vampire for it. i can wait. but those who want to get vampire only to resell it for a premium price, while the cue lasts, may become nrevous about the timeframes.

im not fully in accordance with apollo plan, but with the most of it as far as i can judge. im pretty relaxed about it as it seems to rest in competent hands and i have no investments to lose. id advise to see the project as what it is, a chance, not a duty.