Apollo Team announces the Vampire V4

[Iggy]

It would be nice to know if the core used in this will have an mmu and an fpu.

[Iggy]

Okay, I GREATLY appreciate the work going into the Vampires.  I have a 500+v2 or v2+ of my own and it's been pretty damn awesome.  HOWEVER, why are we still sticking with IDE and not going SATA?  Or both?  Am I in the minority for my desire to use SSDs instead of CFs or even IDE DoM?

In any case, a Vampire for the 1200 will be sweet!

IDE to SATA adapters exist.

[Iggy]

Apollo Core already includes a special MMU, as far as I know there are no plans for a Motorola compatible MMU. As for hardware FPU I don't know, it may or may not eventually have one so if that's a deal breaker either wait and see or stick with Motorola hardware.

'Special' MMU, sound SO useful.
And yes, the lack of floating point functions is a deal breaker (for that matter, no matter how 'special' the MMU is, if it isn't Motorola standard there is software I want to run that isn't compatible).

So, as before, unless all you want to do with your Amiga is play games, you're out of luck.

[Iggy]

Before people jump off a bridge due to the FPU/MMU issue, there is always the chance there will be library support for missing compability.

But in true Amiga tradition, we go nuts instead.

No, the Amiga tradition is making excuses for persistent flaws in hardware and software and fawning over said flawed products like infantile schoolboys with a crush. :hammer:

But, if all you're doing is playing games, hey you're all set (except for the few games that need an fpu).  :laugh1:

[Iggy]

The V4 standalone looks like a fantastic hobby computer. This is exactly what I was suggesting when I posted negatively about the useless PPCLinux laptop idea.

I can't wait to see what the pricing structure is like

Yes, because a device that can run some legacy software has a relationship to a much more modern device that could actually run a modern Linux distro, right?
You opinions are duly noted and dismissed.

You really need to spend less time with that sock, and more time in the real world.

[Iggy]

It has an MMU, but it is an evolved design from previous chips. This has good/bad points. Bad point is it is not backwards compatible. Good point is it is a forward-looking design supporting MMU features (such as non-executable memory) that are not in previous MMU designs. Also has other bonuses that contribute to overall system performance.

Gunnar has taken a long-term view on where the design should be going forward and I agree completely with the decisions being made. I am no fanboy but this guy does know what he is doing. Some short-term pain for sure as the tools/utilities catch up with the new MMU.

The FPU side - short-term the team has recently delivered the "femu" software that allows *any* FPU-less Amiga to run programs requiring an FPU. Pretty cool. Is it as fast as a real FPU - heck no! Does it let a lot of new programs run - yes. How much of a hit the software FPU makes on performance depends a lot on the program. Also this software is very new (version 0.10 recently released) so in theory the speed could be improved going forward.

Long-term the plan is for an FPU to be there in the core. It is just a resource/priority issue (and maybe an available LE issue). At least for now there is a short-term solution.

For some people a hardware FPU is very important. I can understand that completely. For example I think Quake ran at like 2 or 3 FPS (instead of not running at all). In that case you just need to keep waiting for the feature to be implemented... only 2 more weeks...

Cheers!

Well, yes Greg, you are a fanboy.
And the new device has capacity to spare, so where is the f'ing fpu?

As far as 'completely agreeing' with Gunnar, see comment one.

On another posters comment about the Atari ST, you guys are aware that an Apollo core has been run on a Vampire installed in one of those, right?
It would be a waste of Gunnar's time and energy to develop a core specifically for that kind of a downgrade.
If you want a stanalone fpga AtariST, buy a MiST board (or a ColdFire based Firebee).
But then, this isn't an Atari forum, so...

Finally, more than a few of you are likely to be rubbed the wrong way by my post. If so, that's a good thing, as I feel its part of my obligation to counter the occasional really stupid comment.
Like Bloodline's slight of the PPC laptop project I'm an advocate of that has NO relationship to this project what so ever (neither is a substitute for the other, and they are in no way competing devices, so its just plain dumb to go there).

I am glad to see the Apollo core and the Vampire project evolving.
It has a great deal of potential, and I'd like to experiment with the standard Apollo core in applications outside of the Amiga.
One constraint I've run into is that the Vampire is not easily adapted to the processor socket used in an Atari MegaST.
That cpu could give the MegaST a boost above TT performance levels.

Anyway, as I've mentioned before, I'm a big 68K fans, and not necessarily limited to just Amiga (because frankly, you guys get more than a little weird when it comes to your platform).
So...this development is great news, and a credit to Gunnar and the rest of the developers.

It took an idiotically long time to get here, but lets face it, this new device is pretty much a Natami.

[Iggy]

Right, and like I said they're bulk and a PITA but I also acquiesced over the issue.

All the drives in my PowerMacs that aren't directly connected to an SATA controller have one of these attached.
After all, why would you want to have to deal with the expense and lower capacity/features inherent with PATA devices?

[Iggy]

i'll take 3 Vampire4's for my 1200's
and probably another 2 to replace my existing vamps!

erk! this is getting expensive! :lol:

Compared to what?
Could be worse, you could be coughing up the funds for a legacy '060/PPC card.

[Iggy]

@Pat the Cat
you know about femu right? FPU emulator?
http://forum.apollo-accelerators.com/viewtopic.php?f=9&t=1797

its not the fastest, but it allows a lot of FPU software to work without an FPU being present.
IIRC there are ec040, ec060 and vampire binary's.

i'm still figuring all this out myself, so i'm not going to pretend to have all the answers. I have seen it report as having a live FPU in "whichamiga" though after running the executable - and no FPU in "whichamiga" prior to running the exe in a yoochoob video

I never considered this as a solution for an EC based system. Interesting idea. The EC '060s are great over clockers running anywhere between 75 and 100 MHz, so this might be useful beyond Vampire applications.
Most people aren't aware of the fact that the mythical 'FE133' '060 CPUs coming out of China are merely later mask EC or LC CPUs that have been relabeled.
Freescale was emphatic with me when I inquired about those ID numbers, they are NOT Freescale part numbers.

But the availability of cheap EC '060s, when combined with this floating point package, could allow for some interesting dirt cheap accelerators.

[Iggy]

FPU is extremely important as is MMU for running real legacy apps. Not all of us just have amigas to play whdload and .mods! This is the main problem for me with vampire it cant really replace real accelerators for using serious applications that need MMU and /or FPU

Thanks Mag,
I've been saying that for awhile, but the users focused on gaming don't seem to think its important.
One application I can think of is NetBSD, which simply won't run without an MMU (a Motorola compatible MMU, guys).
And a lot of applications make use of floating point calls.
This is not a minor issue, its a serious flaw for anyone that wants to do more than play around, and its freakin' irritating that the fanboys keep fluffing it off.

[Iggy]

iggy most ppl here that use classics basically just use them for gaming not much else.... i have classics more for video stuff (genlocks) and music (midi) and weird gfx. So i need mmu and fpu

Yeah, I kind of gathered that when their primary basis for comparisons between  
PPC NG and the Vampire was how well each platform supported gaming software.
But I don't play a lot of games, and I never have (not even in the '90s).
And I'd actually like to get in on this, but all I get are excuses as to why I don't really need these features, or claims that they'll be implemented down the road.
And if I bring it up, people pout, as if I'm somehow being unfair.

Since when is it unfair to expect upgrade hardware to at least match the feature set of twenty year old hardware?

[Iggy]

I imagine implementing SATA is more than just adding the right socket to the board. IDE is a much simpler design and consequently a quick win.

Im with you on this though , SATA does make sense since thats the protocol everything uses now.

As for SSD , will the Vampire make the Amiga go fast enough to keep up with one ,and are the sorts of file sizes most people will transfer large enough for it to be of benefit? Also you need to think about support for them on AmigaOS . AFAIK theres no support for TRIM .  Im thinking hybrid drive would probably be a better (and cheaper).
 
I wouldnt be suprised if SATA comes in a future iteration - but already this is better than anything we could have ever imagined !

Thanks

Nick

Well, we still have SATA to IDE converters, and you're right that the transfer speeds will be limited with either type of device.
SSDs are fairly pointless for this application, and my experience with them is that when they fail, they do so with zero warning signs (making routine backups  essential).

[Iggy]

I think what these guys want is for developers to update their old apps to suit this new environment. Recompiling old software and modifying it so that it doesn't require an FPU is surely a good way forward.

No, that's just ridiculous. Requiring recompilation of legacy code where the source might not be available...might as well be running on ColdFire at that point.
Look these excuses are no more than that.
And if Gunnar want to implement a superset of fpu commands for new or recompiled software, I'm all for it.

But this isn't a 'new environment', its new hardware designed for old code.
If the OS required an fpu, would the development team still be dragging their heels on it or suggesting recompilation?

Is there some problem the developers aren't letting us onto?
Like maybe they've had difficulty creating a functional floating point component?
I know they claim to have something, but that so far is just a claim.

And compatibility is important, if you're talking legacy applications.
Otherwise, why wouldn't you move to NG?
Its certainly faster.

We see full support for register level video and sound compatibility.
Is there something wrong with Motorola's designs for the cpu that the rest of us aren't aware of?
Because the 'we have a better way' argument isn't proving out.

[Iggy]

No need to recompile old apps for no FPU, just use Femu for now or wait for FPU on the V4.

That seems like a likely solution, at least for the time being.
And with the new fpga, there is room for the additional components.
I'd still like the mmu, as I want to get back to using BSD.

But,  it actually think the new specs look pretty cool.
It's going to be a lot faster.

And, as much as everyone must be tiring of my fixation on these features, at least they can be added to the Vampire.
Tabor is permanently crippled.

[Iggy]

My experience with regular hard drives is the same.  One day everything is going swimmingly and the next day it fails to start.

I've been pretty lucky in being able to recover data from failing mechanical drives, and the SSDs I've had that have had fail have simply stopped working, totally, at once, with zero warning.

But we should all back up our data.

On my PCs these days, I back up to USB connected hard drives that I remove after backup. USB3.0 is a bit slower than SATA connected directly, but its still light years faster than some of the legacy hardware I have.

And I can't argue against the speed of SSDs. I even have an OWC unit in my PowerMac G4. After all, the size limitations for MorphOS drives rather suits SSD capacities.

Right.  SSDs fail too, but unless you need huge capacities solid state is the way to go nowadays.

I use large drives as secondary data storage now, with SSDs for my boot/primary drive.

Combining the two gives me fast operation AND cheap mass storage.

Then again, I still use optical drives, so maybe you don't want to follow my lead...;-).

I even have a PCI SATA controller stashed for use in an future X5000, so I can support two optical drives along with an SSD boot drive and a large mechanical hard drive for storage.