Alright will do next time I boot into Linnie, I've got backups so I can fiddle about all I want.

OK so I spent some time trying to activate twinview and Googling and all that.
I encountered some lovely Nvidia documentation that told me to do things that were impossible.
And no 'Clone' or 'Mirror' or whatever option, but there is one for 'Same Position'...
Which apparently is the same thing.. just with what I find to be a very strange descriptor.
But, as long as it works.
It might seem like a strange thing to obsess about, but my computer is hooked up to my TV which is in front of my bed.
Duplicating the screen allows me to have full access to my computer for browsing and (big screen) gaming from said bed.
It's too bad xrandr apparently doesn't work or I would've scripted a switcharoo script for the settings.
But thank you very much for your help.

Now I just need to find out what the hell kind of layout my keyboard has.
It's a lot like the UK layout, except that the top left key is next to bottom right ? key and the top left is for ~ and #.
Goshdarn Steelseries and their weird layouts.

What xrandr arguments were you trying? ...because I seem to remember changing those sort of things via xrandr a few months ago on my triple-head TwinView setup.

Yw, glad you found the solution :)

Have you tried browsing through available layouts by using your DE's Keyboard or Keyboard Layout program?

I'm using i3 so I don't think there is a tool like that, I suppose I could install KDE or something to check it.

Eh cvt and gft to find the resolution modeline and then xrandr --newmode and xrandr --addmode.
Newmode worked, but trying to add the mode to my monitor failed with a bunch of stuff I don't understand ;)
In order to really 'clone' the displays I need to bring my monitor down to 1920x1080 (same as the TV), but adding the resolution didn't work. It is listed in the Nvidia-Settings though, as 1920x1080 (scaled).

For lighter alternative you could install mate-control-center.

Ahh. Modelines. That's definitely trickier.

I've only used xrandr for doing stuff after they're already set up properly and, the only time I can remember ever having to manually put together a modeline, I set it in Xorg.conf on my ARM-based HTPC because our HDTV returns garbage EDID information over HDMI.

Have you looked into ways to generate a modeline from the currently active mode? (That'd let you set the resolution up in nvidia-settings, then dump the active settings for reuse later in a more scriptable way.)

That aside, I CAN actually tell you how to change your display settings in a scriptable fashion using TwinView and nvidia-settings. Here's an example of how my current configuration looks if set dynamically:

nvidia-settings --assign CurrentMetaMode="DVI-I-1: nvidia-auto-select +1280+0 { ForceFullCompositionPipeline = On }, HDMI-0: nvidia-auto-select +2560+0 { ForceFullCompositionPipeline = On }, DP-0: nvidia-auto-select +0+0 { ForceFullCompositionPipeline = On }"

It takes the same syntax as the MetaModes Xorg.conf line, which is well-documented. However, I'll give a quick overview.

The syntax is as follows:
1. It's a semicolon-separated list of metamodes but you can use the magic value "nvidia-auto-select" for the resolution to have one metamode expand into a whole list of autodetected ones as I've done.
2. Each metamode is a comma-separated list of monitors.
3. Each monitor consists of a monitor name (from xrandr), a colon, and a space-separated set of parameters.
4. The first parameter (the only required one) is a WxH resolution like 1280x1024 or NULL to indicate the monitor should turn off in that metamode.
5.You can specify the position of the monitor within the larger desktop by using +X+Y syntax like +1280+0 as I've done.
6. You can specify a panning domain (desktop larger than the monitor where the viewport follows the mouse) by adding an @WxH like @1600x1200.
7. You can add a set of further attributes inside curly braces as I've done. (ForceFullCompositionPipeline=On fixes tearing issues)

In your case, what you want to do is specify a metamode where both monitors have the same resolution (1920x1080) and offset (+0+0) and you'll get your cloning behaviour. (As far as I can tell, TwinView works by rendering to one giant virtual screen, then feeding rectangles from it to the displays which may overlap to any degree you want.)

By using TwinView and MetaModes, you avoid mucking around with modelines and just tell the nVidia driver "I want these resolutions at these positions. Figure it out yourself."

Also, nifty trick: If you omit the nvidia-auto-select token, you can lock your desktop to a specific set of resolutions. As soon as I can find time to write an LD_PRELOAD hook to fix games that insist on windowed mode using a valid fullscreen resolution, I plan to lock my desktop at 3840x1024 (three 1280x1024 monitors) so games can't scramble my window positions by changing resolution when fullscreening (both Openbox and Kwin like to drop the ball there).

Anyone know how to change the .conf file for dosinstalled games on PoL?

I've installed Screamer 2 using the install script in PoL, and it runs great, but the game is not using 100% of the screen about a 640x480 block in the middle. I've tried editing the playonlinux_dos.cfg file but it seems get ignored, and the one in the PlayOnLinux_DOSBox_tmp dir gets overwritten each time you launch the game, and reading the Wiki doesn't help :(

I'm not familiar with Screamer 2, but why do you need Wine for a DosBox game?

since GOG doesn't have a linux install :(. And just ran it that way :D


EDIT: Got it running on ykhwong's Dosbox branch, but still unable to get 3dfx working :(

I just unpack them with innoextract, adjust the .conf file for things like OpenGL output instead of DirectX, and run them using a Linux-native DOSBox build.

Do you have all 3 monitors connected to one GPU or did you find a way to set up TwinView with multiple GPUs?

One GPU. I bought an EVGA GeForce GTX750 and one of those passive DisplayPort-to-DVI adapters.

(If your GPU has enough CRTCs to drive three displays, which means Maxwell and beyond, then the only thing preventing those passive adapters from working is if the board manufacturer was an idiot and only connected the traces needed for genuine DisplayPort data. That's the difference between regular DisplayPort and the requisite DisplayPort++ on Maxwell cards.)

Sadly, the mobo I'm currently stuck with for budgetary reasons only has one PCI-E slot and my SATA ports are off the end of it, so I'm constrained to a single half-length card unless I want to pull some of my hard drives. (It'd be nice to at least have the option of hooking up the fourth display that my GPU apparently should support.)

As for TwinView across multiple cards, nVidia calls it Mosaic (BaseMosaic is the Xorg config term for doing it without an SLI bridge) but, while it was originally useful on Linux, driver version 302.17 crippled it down to only 3 displays to "match the Windows version" (If you want Base Mosaic with more than 3 displays on Windows, you have to buy a Quadro card), which makes it not much more useful than TwinView.

I was afraid of that answer. But at least I could hope. :)

My card is too old, it's not supported by Mosaic. So it looks like I'm definitely limited to Xinerama until I buy a new graphics card. But nevertheless thanks for your detailed answer!