nvidia : no error but black screen

Question

Possible Duplicate:
How can I get nVidia CUDA or OpenCL working on a laptop with nVidia discrete card/Intel Integrated Graphics?

So I went through a lot of questions and documentations (including of course NVidia's) and whatnot, hoping to find the solution to this problem but I couldn't.

The thing is, I want my ubuntu 12.04 precise (on a dell xps 15) to run its Xorg with the driver 304.54 that is shipped with cuda 5 (obviously because I want to use cuda 5, but would love to have the neat graphics when I don't).

The driver is installed properly (afaik)

any other driver has been removed with --purge. I have installed it with the "devdriver_5.0_linux_64_304.54.run" that comes inside "cuda_5.0.35_linux_64_ubuntu11.10-1.run"

I have uninstalled and reinstalled everything several times as indicated in several tutorials / answers to nvidia questions.

A few commands outputs :

$lsmod | grep nv
nvidia              11270980  0

$lspci | grep -i nvidia
01:00.0 VGA compatible controller: NVIDIA Corporation GF108 [GeForce GT 540M] (rev a1)

$nvidia-smi -L
GPU 0: GeForce GT 540M (UUID: GPU-0016ec97-5f26-f9cd-e618-c7109d7bf0d2)

$dkms status
nvidia, 304.54, 3.2.0-34-generic, x86_64: installed
vboxhost, 4.1.22, 3.2.0-29-generic, x86_64: installed
vboxhost, 4.1.22, 3.2.0-34-generic, x86_64: installed
vboxhost, 4.1.22, 3.2.0-34-virtual, x86_64: installed

dmesg | grep -i nvidia
[   18.366259] nvidia: module license 'NVIDIA' taints kernel.
[   20.558104] nvidia 0000:01:00.0: power state changed by ACPI to D0
[   20.558463] nvidia 0000:01:00.0: power state changed by ACPI to D0
[   20.558585] nvidia 0000:01:00.0: enabling device (0006 -> 0007)
[   20.558746] nvidia 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[   20.559427] nvidia 0000:01:00.0: setting latency timer to 64
[   20.574937] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  304.54  Sat Sep 29 00:05:49 PDT 2012

$Xorg -version
X.Org X Server 1.11.3
Release Date: 2011-12-16
X Protocol Version 11, Revision 0
Build Operating System: Linux 2.6.42-23-generic x86_64 Ubuntu
Current Operating System: Linux fordprefect 3.2.0-34-generic #53-Ubuntu SMP Thu Nov 15 10:48:16 UTC 2012 x86_64
Kernel command line: BOOT_IMAGE=/boot/vmlinuz-3.2.0-34-generic root=UUID=a35b41e6-f102-422b-9c57-c6ce26ac9ea0 ro quiet splash
Build Date: 29 August 2012  12:12:33AM
xorg-server 2:1.11.4-0ubuntu10.8 (For technical support please see http://www.ubuntu.com/support) 
Current version of pixman: 0.24.4

CRT-0 is the only display device found as connected (see from Xorg.0.log) :

[   320.555] (--) NVIDIA(0): Valid display device(s) on GeForce GT 540M at PCI:1:0:0
[   320.555] (--) NVIDIA(0):     CRT-0 (connected)
[   320.555] (--) NVIDIA(0):     DFP-0

Changing xorg.conf and observing Xorg.0.log :

• If I do not use the nvidia driver (using driver "fbdev"), the system runs some intel graphics chip. There are then no devices named /dev/nv*

• If I add the nvidia driver to xorg.conf, the Xorg fails saying it can't find the card on its own, and starts lightdm in low resolution

• If I add BusID, the Xorg.log complains that it can't read the EDID and says in starts with 640x480. /dev/nvidia0 and /dev/nvidiactl are created.

• If I add the EDID with Option "CustomEDID" "CRT-0:/path/to/edid.bin" (which I got from get-edid (in package read-edid) when running without nvidia driver) there is no error or warning, in the Xorg log. /dev/nvidia0 and /dev/nvidiactl are created.

• I have tried just for the sake of it to set the option "UseDisplayDevice" to DFP or DFP-0 (with and without "ConnectedMonitor"). But this does not help as expected - my screen is definitely CRT-0 as its name (AU Optronics something) appears in the log when I set Connectedmonitor.

The problem

In the three last cases I hear the sound of lightdm starting when it prompts for a password. There is just nothing on screen. I have tried to press the switch-screen button (Fn+F1 here) but that does not seem to do anything. So I can of ran out of things to try/look at here. If anybody can suggest something I haven't tried (and preferably not "try another driver version"), I would be most grateful.

---

Here are the Device blocks from xorg.conf, with nvidia :

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "Geforce GT 540M"
    BusID          "PCI:1:0:0"
    Option         "CustomEDID"        "CRT-0:/var/opt/edid.bin"
    Screen         0
EndSection

without nvidia :

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
    SubSection     "Display"
        Depth       24
    EndSubSection
EndSection

Everything in the rest of the xorg.conf was already there before I started looking at it, and is exactly identical in the two different configurations. Full xorg.conf file here : http://pastebay.net/1161671

Just to add a little and probably irrelevant detail : in the cases where Xorg thinks its working fine on the GPU, and I switch to a tty1-6 and back to tty7, the screen doesn't exactly go black, rather what is on-screen freezes and nothing happens. The screen gets back to life when i go to another tty again.