X Server 1.9 Window Closing After RandR 1.4 Pull
Phoronix: X Server 1.9 Window Closing After RandR 1.4 Pull
There's good news for the Ubuntu camp and others releasing in the September-October time-frame: development work on X.Org Server 1.9 is still going as planned for an August release and its merge window is about to be closed. In the past it's been tough for the X.Org project to release server updates in a timely manner that's on schedule, but continuing from their X.Org Server 1.8 success, 1.9 is shaping up nicely too...
I always see this abbreviation used but never explained - what is a "CRTC"?
A CRTC, or Cathode Ray Tube Controller, generates the video timings and reads video data from a RAM attached to the CRTC, to output it via an external character generator ROM, (for text modes) or directly, (for high resolution graphics modes) to the video output shift register. Because the actual capabilities of the video generator depend to a large degree on the external logic, video generator based on a CRTC chip can have a wide range of capabilities. From very simple (text mode only) systems to very high resolution systems supporting a wide range of colours. Sprites however are normally not supported by these systems.
In other words:
A crtc is basically a display controller (crt controller). What it does is that it is linked to the physical ports, like DVI...
It is what takes a certain sement of the frambuffer and sends it to your screen.
It can do all sorts of fancy stuff. For example if you have a dual screen setup, a crtc copies the segment of the frambuffer for your left screen and send that to the monitor 1, and another crtc takes the right image and sends it to screen 2 connected to it...
Are you sure about that?
This is the first time I've heard this component named so I could be wrong, but it seems as though this sits between, as you said, the phy interface and cathode ray. The splitting of the screen you are speaking of seems as though would have to occur at the driver level since the card is responsible for splitting the image and sending the appropriate signals to each monitor. The crtc would never "know" it didn't have the whole image.
Originally Posted by V!NCENT
Right. A CRTC just knows how to turn an area of memory into a stream of digital data with the appropriate sync and blanking signals.
Note that most of our GPUs before Evergreen have 2 CRTCs (some of the really old ones only had 1), while the Evergreen parts have either 4 or 6 depending on model.
Out of curiosity: why do we need that many?
Originally Posted by bridgman
Most users only use a couple of displays at a time, but a typical graphics card has 3 or 4 outputs and a docked laptop might have 5 or 6. In principle you could get by with 2 CRTCs and a complex switching arrangement between CRTCs, encoders and outputs, but you still end up with sitations where you can't run output X and output Y independently because they share "something" -- so dedicating a CRTC to each potential output provides more flexibility for the user and lets you eliminate a lot of switching logic. The switching logic to map CRTCs onto encoders & outputs was getting wider, faster and more complex every year, to the point that having more CRTCs was a better solution than the switching logic required to let 2 CRTCs cover all the possible output scenarios.
It also turned out to be mighty fine for multiscreen gaming, but that wasn't the only reason
Does this possibly mean that there will be support for using passive displayport converters for 2+ monitor setups and "legacy" (as in purchased last year) DVI panels? Please say yes, since I'm the unhappy owner of 4 Saumsung 2233's, 2 NVIDIA 9800GT's and a Composite-less desktop.
Originally Posted by bridgman