Anyone know what this NuBUS card might be?

[cheesestraws]

Yeah, it's not worth anyone doing anything serious—I didn't expect the kind of digging that @jeremywork and you did!—I was mostly asking for curiosity and whether that could be used to confirm the hardware differences thing. But I think that's adequately confirmed now, so...

 

[Melkhior]

I wonder how the AMD (?) RAMDAC in the newer revision is implementing "thousands". The BT473 has a 15-bits bypass mode, but it sets the LSB to 0 - Apple normally set them to a repeat of the MSB (i.e. they expand 'abcde' to 'abcdeabc' and not to 'abcde000'). maybe that's why they didn't enable 16-bits? The BT473 also seems to be the only one in the '91 datasheet with 15-bits support (I don't remember anyone other than Apple actually using 15-bits, most workstation vendors moved from 8-bits to 24/32-bits, and the PC world used 16-bits by adding an extra 6th bits to the green (?) channel).

 

[mg.man]

The BT484/485 are unlikely candidates

So... I was pretty sure, so, checked and found it :

Could this be the answer?


Same Apple P/N as my GC, but note the "Bt" P/N is still there...

 

[Phipli]

So... I was pretty sure, so, checked and found it :
View attachment 52969
Could this be the answer?
View attachment 52970

Same Apple P/N as my GC, but note the "Bt" P/N is still there...

Cool, note the colour of the Apple info is different. Must have relabelled some parts for Apple that already had stock labelling.

That's a 4•8, but with the VRAM SIMMs it is functionally identical to the 8•24. The PCBs are the same, just populated differently.

 

[mg.man]

Must have relabelled some parts for Apple that already had stock labelling.

Yeah, that was my thinking.

 

[Melkhior]

Same Apple P/N as my GC, but note the "Bt" P/N is still there...

Unfortunately, it doesn't seem to be a public Brooktree part number that I can track. Expected, as they were 3-digits in those days, and RAMDACs were all BT4xx AFAICT.

Also, looking up AMD's "Personal Computer Products" book from '89, they don't have a matching PLCC-68 RAMDAC either (from @MrFahrenheit pictures it's an AMD chip on some boards). Just PC-oriented stuff and clones of Brooktree or INMOS devices.

I suspect Apple had its own packaging. @MrFahrenheit: does it look like the pin-outs are the same than the one reported by @jeremywork ? It would explain things if they just repackaged third-party silicon to simplify PCB design...

 

[Phipli]

from @MrFahrenheit pictures it's an AMD chip on some boards

The same PCBs seem to use the Bt and AMD parts. The AMD part seems to be pin compatible, but additionally support 16bit colour. Otherwise electrically identical (and perhaps functionally too). The updated ROM for the AMD part is backwards compatible with the Bt part.

 

[Melkhior]

The same PCBs seem to use the Bt and AMD parts. The AMD part seems to be pin compatible, but additionally support 16bit colour.

I can't find a lot of information on AMD's RAMDAC - the '89 book I mentioned earlier doesn't even have a truecolor product. They definitely did 'clones' of Brooktree products, rather famously back in the day, so a pin/SW-compatible replacement for a BT product is somewhat logical.

As I've yet to see a "standard" RAMDAC doing the Apple's version of 'thousands'/15-bits, I'm starting to think they were all custom-made.

Did many third-party boards (NuBus, PDS, whatever) offer 'thousands' ?If so, it might be interesting to look at the RAMDAC of those boards offering it to see what they used - though they may have also done the 'simple' 15-bits thing of zero-padding the LSB.

 

[Phipli]

thousands'/15-bits

Can I ask how come you say 15bit? Isn't it 16bit? I always understood it was 65,536 colours? 2^16?

 

[Phipli]

Did many third-party boards (NuBus, PDS, whatever) offer 'thousands'

In truth, it was missing from many 3rd party Nubus boards, but most/all later ones did it (PCI era).

 

[Melkhior]

Can I ask how come you say 15bit? Isn't it 16bit? I always understood it was 65,536 colours? 2^16?

It is for everybody else, but Apple defined 'thousands' as 32768 colours (2^15), with exactly 5 bits per channel. Everybody else uses the 16th bit as an extra 6th bit for the green (?) channel, but for Apple it is unused. See "Imaging With QuickDraw", p1-20, "Direct Colors". The translation scheme between depth is in section "Color QuickDraw’s Translation of RGB Colors to Pixel Values" p4-13.

It is normally stored as a 16-bit value in memory. Same as in post-80s systems, 24-bits (8 per channels) is commonly stored as a 32-bits value, with the extra 8 bits either unused or used for e.g. alpha channel. In the eighties, it would have been more common to use 'true' 24-bits storage, usually using 3 banks of 8-bits memory. They would be either concatenated in a single 24-bits value for 'chunky' representation, or used as three 8-bits plane for 'planar' representation.

 

[MrFahrenheit]

@Melkhior You called? What do you need to know from my card(s)?