Mac mini: How to arrange a dual display setup?

[register]

Hello, I would like to connect two displays simultaneously to the same Mac mini, using a DVI to DVI + VGA Y-cable. The cable splits the DVI port into a DVI port (digital) and a VGA port (analogue). Actually, I want do this with an early Intel machine, but the same should work with the G4 incarnation of the Mac mini. I found a VGA monitor running perfectly in this setup, as long as the DVI connector is not used. As soon as a display is connected to the DVI output connector (digital on), the VGA monitor gets no signal any more (analogue off). Is there a procedure to invoke both displays at the same time? If yes, please be so kind and share your knowledge.

 

[cosmichobo]

https://everymac.com/systems/apple/mac_mini/specs/mac_mini_cs_1.5.html

According to this link on Everymac the mini is only capable of powering 1 display...

(Of course - I expect if there's a work around, someone here will know it?)

 

[Byrd]

Unfortunately you’re never going to get dual display out out from a single DVI port - I’ve never heard of it being done, the chipset will never support such a thing. You might be able to find an OS X compatible USB video card to get dual (albeit slow) displays on a G4 or early Intel mini.

 

[trag]

The early 2009 Intel Mac Minis were the first models to support dual displays.

Early 2009 Mini

Minis on Everymac

Now, in theory, the VGA pins of a DVI port are separate from the digital pins of a DVI port and one could send different outputs to each set of pins...

So, first question would be whether the graphics chip in the earlier Mac Minis support more than one display -- most did by that era.   Then the second question is how does the output from the graphics chip get to the video out port(s).   If Apple just wired up one set of outputs from the graphics chip to whatever PHYs are needed to turn the output into DVI and VGA, then it's not possible to get different  outputs.   If the two PHYs that serve DVI and VGA have separate outputs on the graphics chip, then it might be possible to hack multi monitor support into the system through software/firmware.  If you can convince the graphics chip that the two outputs have distinct frame buffers.

 

[Gorgonops]

I didn't realize that cables like this were even a "thing", but they do indeed exist... but also apparently only work with certain graphics cards. The Mac Mini developer note doesn't say anything about the DVI port supporting any sort of dual supply splitters. It seems like in principle the Radeon 9200 chip in the Mini *could* support such a configuration, as it looks like it's capable of driving a grand total of three displays simultaneously (two analog and a single digital link... that latter limitation is why the Mac Mini can't use a 30" monitor) using "native" PHYs on the chip, but if Apple omitted the firmware support for it it's probably going to be... nontrivial, to enable it.

Some versions of the iBook G4 and the eMac use similar Radeon 9200 GPUs. However, the author of the "Screen Spanning Doctor" software specifically says the following:
 

Mac mini



These machines can NOT use the patch:


Unfortunately the patch does not work with ANY Mac mini configuration.
It is not possible to use both the DVI and the VGA part of the display connector simultaneously.

Breaking this down it *may* well be that it's still actually a software limitation (my guess is that the DVI port still just runs the digital and analog lines to separate PHYs on the Radeon chip instead of doing something weird like using an external DAC on the digital lines), but that doesn't mean you can easily get around it. My vague understanding of how the Screen Spanning Doctor works is it patches the information associated with an already present Open Firmware node for the external VGA port so the driver will allow that second port to be associated with a separate framebuffer instead of only mirroring of the primary display. My guess is that the Mac Mini only has a single "port object" on its device tree; that object supports both analog and digital PHYs, but it's still only one *port*. If that's the case then the only way to undo it would be to patch the ROM driver in a far more extensive manner than SSDoctor does.

 

[trag]

Screen Spanning Doctor

This software is particularly useful on certain Macs that have two video out ports, but in which Apple only enabled screen mirroring, and not screen spanning, such as the 800MHz G4 17" iMac.  

Your idea is very interesting, in that I don't think any of us had ever considered the possibility before, but as Gorgonops writes, it would a non-trivial hack.

There is a forum associated with the Screen Spanning Doctor.  I don't know if it's still active (looks like things may have been idle since 2005) but that would be a good place to try to get the attention of folks specializing in this kind of task.

There are people trying to get OS9 to work on PPCs that only support OSX.  I doubt that this video idea is more complicated than that.  So it's probably feasible.    But it will require work from someone with skills and time.  AFAIK it is not an already solved problem.

If you do pursue this further, please keep us updated on the results, positive or negative.   I want to hear about it.

 

[Cory5412]

I would expect this to be a hard limitation. The iBooks/iMacs/eMacs that can use screen span hacks have the distinction of being able to mirror those displays first. I suspect the difference is whatever hardware is needed to do this does exist in those hackable Macs, and does not exist in the Mac mini.

Is there any documentation on this specific type of thing working with other DVI-I ports? The only examples I know of, personally, of two display outputs coming out of one port are the DMS59 connector, which was explicitly designed to be two DVI, two VGA, or one DVI and one VGA for multiple displays, and DisplayPort with multi-stream-transport, which can either power very huge displays or multiple displays in tandem.

 

[register]

Actually, it is possible to boot a Mac mini G4 into Mac OS 9. Might this permit use of appropriate Radeon driver software to run the Mac mini's Radeon 9200 chip in a dual display mode? If someone has this configuration (Mac mini with Mac OS 9 and ATI OS 9 Mac Software Update January 2005) up and running, could you be so kind and check this detail, please.

 

[Cory5412]

I'd be interested in seeing if anyone has a mini hanging around booting OS 9 and whether or not it's stable and has all the driver needed to run wel.

But, if I had to guess, the reason you can't use both the analog and digital portions of the Mac mini's DVI port is because they're wired as such, physically. If I had to guess, the VGA port simply displays a mirror of what's on the DVI port.

The other machines on which that extension work have their internal display on a port and a VGA or DVI port that's physically separate, but Apple nerfed the original drivers, basically for product lineup differentiation reasons.

 

[Gorgonops]

But, if I had to guess, the reason you can't use both the analog and digital portions of the Mac mini's DVI port is because they're wired as such, physically. If I had to guess, the VGA port simply displays a mirror of what's on the DVI port.

One problem that comes up in really trying to dive into this is it's gotten to the point that it's hard to quickly google any in-depth information about the inner workings of the Radeon 9200. The stuff I've found leads me to believe that *unless* ATI made different versions of the chip that have various options relating to number of virtual framebuffers and how they map to the PHYs hard-set (IE, by blowing option PROM fuses or offering different physical die skus or... something) so that the analog and digital parts of that port are permanently mirrored to each other it's *theoretically possible* that with low-level reprogramming you could get the Mac Mini to do separate heads on the analog and digital halves of the port. But I think to do it you're looking at having to hack the chip at the raw register-banging level. My educated guess is that the normal MacOS drivers (9 and X) rely entirely on the Open Firmware device tree interface for mode switching and framebuffer definition and the driver in the Mini's ROM simply *does not* allow for the possibility.

Realistically speaking, I don't think it's going to happen unless someone with deep arcane knowledge of Radeon chipset programming steps up to the plate and essentially writes a new BIOS for the machine. And, again, while I'm *somewhat* skeptical that the machine is absolutely unable to do it at the hardware level I wouldn't actually bet money on it. It is *possible* that ATI did sell versions of that chip that only support a single virtual framebuffer.

 

[Cory5412]

The other computers that use the Radeon 9200 and have dual display output all seem to have a second port physically connected to an internal display of some sort.  I would bet, again without having looked, so this could be wrong, that that's fused off or not connected to anything in the mini.

PCI and AGP Radeon 9200s I can find online have two separate ports, often one VGA and one DVI. It's not unreasonable to think that there's a video port that's simply not connected.

 

[Gorgonops]

PCI and AGP Radeon 9200s I can find online have two separate ports, often one VGA and one DVI. It's not unreasonable to think that there's a video port that's simply not connected.

The documentation I found on the 9200 basically boils down to this: the chip has 3 physical PHYs; two are analog, one is digital, and supports two simultaneous framebuffers or "heads", each of which can be mapped to one or more PHYs. If the chip is used in a laptop the internal LCD occupies the digital port so only analog PHYs are available for external ports; the most common PCI retail card configuration had the digital part of the DVI-I port attached to the digital PHY and shared an analog PHY with a TV out connector. (The second analog PHY was mapped to a VGA port.) The card could do a separate "head" on any two ports or, here's the important part, could actually run *three* monitors at once consisting of one digital and one analog monitor and the TV port mirroring *either* monitor. Reference for display limitations on the retail card, page 6.

The fact that it *can* mirror either monitor on the TV port is the reason for believing that it might be *technically possible* to put two different heads on the DVI-I port in a Mini given sufficiently clever reprogramming of the chip registers. The TV port is using the *same* PHY/DAC as the analog portion of the DVI-I port, that's why it's an unsupported configuration to have a CRT monitor connected via adapter to the DVI port and the TV port active at the same time. Unless the chip in the Mini is a "brain damaged" version that doesn't allow more than one PHY to be active (or has disabled the ability to activate more than one head) it seems like it should be "possible", by some loose definition of possible.

 

[trag]

From macos9lives:

The Trampoline parcels structure can be used to add properties to the device tree for certain devices. This is already done to insert 'NDRV's for some crucial devices before Mac OS 9 boots. If you don't want to replace the rom on the card, you could mess around with ELN's newworld-rom repo and make a Mac OS ROM file that inserts the correct 'NDRV'.

It sounds to me like they may already have the tools needed to do the low level hack needed.  Whether it is low enough level, I don't know.

 

[Gorgonops]

Whether it is low enough level, I don't know.

Yeah, I have no idea myself. The difference between the Mac Mini and the other machines that, say, Screen spanning doctor, were able to make work is that the device tree for those already has two "ports" defined by the firmware under the parent device, it merely deletes a flag that signals the driver to disable using anything but mirror mode. IE, it sounds to me like the code path for having two PHYs enabled simultaneously is already there, it's just a driver-level thing that prevents attaching two framebuffers. If the Mini's firmware is hardcoded to treat the DVI-I port as a "single" port that can only have *either* the digital or analog PHY active at once the driver itself might not have the code to override it; it all depends on whether the driver "bit-bangs" the chip for that function or calls the firmware.

(The reason I have for uncertainty here is for a *long* time on the x86 side of the house protected-mode operating system video drivers often would still use the processor's vm86 mode to make real-mode calls to the VGA bios for mode setting. It's a horribly awkward way to do it but was considered "safer" than making very poorly documented calls against the raw hardware. Open Firmware is *designed* to avoid awkwardness like that, so it would make sense for the Macintosh Radeon drivers to use it. So if the firmware in the Mac Mini genuinely thinks it has "one port, one port only" you could be SOL unless you replace the whole BIOS.)

 

[register]

Dear fellows, I absolutely love you. I have enjoyed reading your illuminative treatise on possible implementations and uses of the 9200 chip in the Mini. Thank you!

 

[Cory5412]

The dev note for the G4 Mac mini is available online. However, it doesn't shine a lot of information onto this, it basically just says there's one video output and it can do svideo, VGA, and DVI.

My guess is it's wired up like the mini-DVI port on any other machines of it's era and there's an unoccupied video output port not connected to anything.

Or: on a desktop Radeon card featuring both VGA and DVI, did those VGA/DVI splitters like the one mentioned in the first post ever work, or did Apple never implement it, even on its PowerMac cards?

That would be the most telling thing to me and it would be extremely like Apple, I think, for them to just leave an unoccupied video header, so you'd need to upgrade to a Power Mac to get dual displays.