Thanks for your polite answer, I'll be looking forward to an official announcement then ;-)
In the meantime, I have good news on my part: I've tried with the stock HD Graphics driver for Windows, on an "embedded" motherboard with QM57 (i.e. the "mobile" chipset), namely the Advantech AIMB-270, which even has an LVDS port on a header, but I need to use the DB15 VGA port as primary. I've saved my settings as a custom "user profile" in the Intel Graphics systray applet, pulled he VGA cable, and rebooted. Waited a while for Windows XP to boot, and then plugged the VGA cable back. And voila: the picture was still there :-) The driver now did not default to the LVDS port! It seems that this solves my key problem known in the past :-)
I've noticed that the standard HD Graphics driver now allows you to define a custom resolution and give it a name ("add a resolution profile" to the default list). There are two modes of the config dialog:
1) set the basic width vs. height in pixels, the desired horizontal refresh rate, select your timing style (select GTF and CVT, there's no DMT nor CVT with reduced blanking) and there's even a word about "underscan percentage" (by default at 0%) - is this possibly meant to compensate the ugly overscanning done by modern flatscreen TV's to their HDMI input? Or is it just another name for "reduced blanking"?
2) set all the gory details, down H/V sync and porches and pixel clock. You need to know the GTF maths and have some examples of DMT/CVT/CVT-R modes, but if you've been there and done that (e.g. in the IEGD CED, or in Xwindows), this is your ultimate knob on resolutions.
This is good :-) In my "embedded" practice, there have been occasions where I needed to tweak timings a bit, to make a particular monitor feel better, or to dodge some analog bandwidth limit on inferior TMDS cabling.
The one thing that I'm missing (that IEGD could provide) is an option to get rid of DDC/EDID. The IEGD can be told to force output to a particular display, regardless of whether or not DDC indicates a monitor attached to that port. There's no such option in the stock driver yet. The stock driver only allows you to select among inputs where some display is "detected" (possibly by DDC on the digital outputs). Curiously the "analog CRT RGB DB15 VGA" seems to be always available - that's a good choice.
I've noticed that the QM57 only allows me to select two display outputs at a time, either in clone mode or in "extended desktop" mode (which translates into independent "display" entities in the Windows XP screen properties). My motherboard has 2x DVI, analog CRT and LVDS - and I always have a choice of two, out of all the displays "detected attached". I believe this "maximum of two displays active at the same time", out of the total number of 4-5, corresponds to the FDI link between the CPU and the PCH being a dual-link pipe, which gets switched by some crossbar in the PCH to the physical PCH output ports (about 5 of them I believe, with slight differences between different chipset flavours). This arrangement seems to ring a bell, after I've fumbled through the datasheets of 5 and 6 series Intel chipsets a while ago. I believe the older chipsets (3 and 4 series) would allow you to have more than 2 display outputs active at a time, but I don't recall ever using that option :-)
One last note: it seems to me that the QM57 doesn't accept and ADD2 card (using the PCI-e x16 graphical port's alternative function as SDVO). At least not the old card that's known to work in i915 through i3 and i4 series GMCH's. Might be a chipset feature (missing), might be a motherboard-level design issue (deficiency), might be an incompatibility with my old ADD2 board (a new generation of pinout expected by the modern mobo or chipset).
1 of 1 people found this helpful
In regards to your concerns, please be aware that the Intel(R) HD Graphics is indeed limited to 2 video outputs at a time. Triple monitor configurations are not supported. Regarding ADD2+ cards, please bear in mind that support for these devices is no longer provided. We apologize for the inconvenience.
Okay, thanks for confirming those points :-)