This is an FPU for ... what?

[mcdermd]

It certainly appears to be for an LC. The LC PDS and general size seem to hint at it. But the clock runs at 25 MHz. The only LC at 25 MHz would be the LC III and it has an FPU socket on the logic board. The packaging is marked as a 25 MHz FPU for a Classic II but the Classic II doesn't have an LC PDS slot.

Any ideas?

 

[uniserver]

well the LC-II runs at 16 mhz,

The crystal on there is at 25 mhz,

LC-II could benefit from this because there is no FPU socket, even the color classic has a fpu socket.

the Classic II could benefit from a FPU upgrade… but does not have a pds slot.

and the classic II runs at 16 mhz, Not 25 mhz.

very strange! 8-o

 

[Trash80toHP_Mini]

Not really, Quadra 605 or LC475? They shipped with the 64LC040 and a 25MHz 68882 on a card would be less $$$$$ than a 68040.

The ALT/SYS jumper might make it an OverClock/CoPro Card for the LC II, dunno, but it doesn't seem all that strange to me.

 

[uniserver]

this math co only works with 68030 CPU's

 

[TylerEss]

I seem to recall reading that you can't run a 68LC040 + 68882 combination.

I also seem to recall reading that when running a 68020 + 68881 or 68030 + 68882 that the clock speeds of the FPUs can differ from the ALU. (I also recall something about multiple 6888{1,2}'s per 680{2,3}0, with appropriate glue on the bus...)

So maybe it's a 25MHz FPU for the LC/LCII/LCIII?

 

[uniserver]

now that makes sense to me.

 

[Trash80toHP_Mini]

My bad. HRMMM?????

 

[Gorgonops]

Do you have a picture of the packaging? Two easy explanations:

A: The box is for a Classic II upgrade, but someone stuck an LC FPU board in there instead. (According to Apple the Classic II *does* have a slot that's intended for an FPU upgrade, and according to several sources some were made. link)

B: (Pure speculation) A vendor sold an FPU upgrade for the Classic II that involved packaging an LC FPU and an adapter board? (And the adapter is missing, possibly still stuck in the slot of the Classic II this was pulled from?)

There are passing references to FPU boards that ran asynchronously to the main CPU but a quick Google didn't turn up a specific product name. It could be that (it does appear to have a crystal onboard, and... do you suppose that "alt/sys" jumper might select whether the crystal or the motherboard clock is used?) or it could just be that when the board was populated 16Mhz FPUs weren't any cheaper than 25mhz ones.

 

[mcdermd]

Well ...

The "packaging" is simply the original label on the original static bag so I'm out of luck on any box copy.

We had thought of Gorgonops's explanation B while I was brainstorming with uniserver. That would be a pretty cool explanation but I think it's more than likely either a misprint or a swticheroo between the product and the packaging. I'll see about throwing it into an LC II soon and seeing what happens (thanks for the LCII, uniserver).

I also wrote a quick email to Lifetime asking if they knew but I'm not holding my breath on that one.

 

[beachycove]

I also recall something about multiple 6888{1,2}'s per 680{2,3}0, with appropriate glue on the bus...
So maybe it's a 25MHz FPU for the LC/LCII/LCIII?

I am pretty sure it is for one of these machines, or similar (e.g., LC 520), but I am now wondering what would happen if TWO FPUs were popped into an LC.

On the question of the Classic II, does anyone have one of the FPU cards, remember what they look like, or know where to find a photo of the FPU card? I have always assumed that the FPU card stood in the machine vertically, like a Macintosh Classic RAM card. It would be a turnup if this really were for a Classic II, since I also have one of these cards. Mine has a 16mhz 68882 in it.

 

[beachycove]

Not on this list, either -- not that that in and of itself means a lot: http://support.apple.com/kb/TA47024?viewlocale=en_US .

 

[bbraun]

FWIW, the 68882 Technical Summary documents how to connect 8 of them to a 680{2,3}0, and that they may run asynchronously clocked.

Both multiple 68882s and asynchronous clocking are covered in sections "Bus Interface Unit" and "Coprocessor Interface".

Summary as I understand it: each 68882 needs a different coprocessor ID and address decoding logic that connects to its /CS line. Then the CPU ID is encoded in the instruction so the 68030 knows which coprocessor to send the instruction to. So 1) just throwing another Macintosh FPU expansion on the bus will not work since they are all responding to the same coprocessor ID, and 2) even if you did get more than one setup properly, you'd need modified software to use it. I haven't seen any options in any of the dev tools I've seen to allow specifying the coprocessor ID of the generated instructions, so you'd probably be limited to writing new assembly to address your additional 68882's.

For asynchronous clocking, it seems to be a result of the 68882 never being a bus master, and all the bus timing is driving off of /AS and /DS.

Anyway, that's just my understanding as a naive software guy.

 

[TylerEss]

FWIW, the 68882 Technical Summary ...

Thanks for verifying that I'm not crazy!

 

[Gorgonops]

Re: multiple 68881/2s, there's actually an Apple knowledge base article floating around (I'm too lazy to look for it right now) that covers the implications of plugging an LC/LCII card with a coprocessor on it into the slot of an LCIII/III+ that has an onboard socket. The short answer is that, yes, both sockets will be mapped to the same location and will conflict with each other if there's also an FPU on the motherboard. The *one* reason, based on the knowledge base article, that you might ever want to us an LC coprocessor card in an LCIII is if you only have a 16Mhz 68881/2. The article says the slot is clocked at 16Mhz even in the faster 25/33mhz machines, so an FPU that won't work at full speed in the onboard socket *can* work in the slot, albeit more slowly.

(And of course, if you actually did that you'd essentially be doing exactly the opposite of using a faster-clocked FPU card with its own crystal in an LC/LCII.)

I'll second that it's amazing just how un-findable photos of a genuine Classic II FPU upgrade are.

 

[trag]

Anyway, that's just my understanding as a naive software guy.

Hah! You're dangerously close to being a hardware guy. You appear to have a better understanding of the protocols of the 68030 bus than anyone else posting here. Signal protocols. Pretty hardwary.

 

[Trash80toHP_Mini]

He's FirmWary, that's even scarier, almost more so than practitioners of (PSU design/Magic Smoke Containment) Black Magic! :approve:

 

[bbraun]

Heh thanks, I'll take the compliment but appearances can be deceiving. Specific knowledge from reading the datasheet shouldn't be mistaken as actual understanding.

 

[uniserver]

This would work good in the LC-I because the 68882 will work with the 68020 CPU,

Then you can install Linux on that bad boy

http://www.linux-m68k.org/faq/reqs.html

 

[Gorgonops]

Just for the record, no, read the fine print: (It's in the mac section of that FAQ you pasted the link to) you still can't run regular Linux on an original 68020 LC after adding an FPU because the LC also lacks a 68851 PMMU and there's no straightforward way to add one. (IE, no socket, unlike the original 68020 Mac II.) It's mildly interesting how the FAQ says there's a "hardware hack"; if there's a way to add a PMMU *other* than upgrading the CPU to a 68030 or better it seems to be rather difficult to find a reference.