Printable View
1. Is still / has ever INTEL_BATCH=1 worked ?
2. Why is the gma3150 clocked at 200Mhz while the gma950 was clocked 400Mhz ?
3. Can the gma3150 be overclocked (safely) ?
4. If I limit the memory in the kernel line in the boot menu will the gma use then this memory preferably or ignore it?
Thanks ahead and greetings!
1) Gallium3D for the first next gen hardware which will need a complete driver overhaul
2) Where is G45 va-api? With latest code and swapbufferswait off (which is not suitable for video) it does barely match software decoding on my 1,4GHz Core 2 Duo ULV. With SNA va-api is even slower (which is very strange anyway)
3) GM45 support in upstream kernels: https://bugs.freedesktop.org/show_bug.cgi?id=27169
4) Unfair or not, please dismiss the Poulsbo/Medfield team and open the specs, mobile is the future and Intel is just not ready for the future with that crap of a driver
Thanks
You are the only one left in the classic Mesa land. From the beginning you were worried that there are no production ready Gallium3D drivers. There are now.
Do you consider switching to Gallium3D, and clear up the stack after classic removal?
AFAIK this could speed up gallium path considerately. Thank you.
For a long time I hoped to get tearing free video playback using XV on my old box using i945 graphics when compositing is enabled. At some point, this finally happened.
Then I upgraded to a Sandy Bridge box and have tearing video again, currently using Debian Sid, Gnome 3.2 and recent kernels. I tried XV and VAAPI playback in different players (mplayer, vlc, MythTV) and all were affected. This really makes me sad.
Any hints what I'm doing wrong or, if it's not my fault, when tearing free playback will be possible on Sandy Bridge?
NVidia provides really nice deinterlacing in VDPAU. Intel VAAPI on Sandy Bridge is very limited regarding this. Will there be better deinterlacing in VAAPI for Sandy Bridge?
Both nVidia binary driver and the X server are having problems with hybrid graphics. The nvidia driver does not yet switch the mux, it just does nothing if the NV GPU is not connected to the screen, which usually results in lack of 3D or a black screen. With Nouveau it's at least able to fall back to Intel's 3D acceleration if the mux is not switched to NV.Quote:
Originally Posted by ciplogic
First out, let me say thanks for the work you do. Intel Linux drivers far surpass the
Windows drivers in quality where it matters (OpenGL).
1) My experience matches Kano's, in that if you happen to use even one different version
than mentioned in the quarterly package, it's unstable. This doesn't happen on other FOSS
drivers (ati, nouveau).
2) Since you focus on the classic driver, g3d MLAA doesn't concern you really. But the
user-space version[1] is known to fail on Intel. Perhaps you could take a look?
The minimum hw was GF6/R500, so at least Sandy should be able to run it hw-wise (the
requirement comes from shader length, branch support, and texlod support).
There's no bug open, because I don't own any recent Intel hw.
3) Even though as a dev I understand why you did it, as a customer I'm still pissed about
how you dropped the support for i8xx hw. I own several i8xx Intel chipsets.
They went from acceptable 3d and video to slow and buggy 2d. I don't think even XV works
anymore.
[1] https://github.com/clbr/MLAA-test-app
First thank you for great work done for Linux and opensource drivers. Only reason i bought SNB cpu/mobo for new XBMC PC.
Question: Option that HDMI is always ON and is used even "other side" of HDMI connector is not ON.
Problem I have: Powering htpc before TV/amplifier will result in black screen, only solution is to turn everything off, then first power amplifier, TV then htpc.
Answer i found on IRC or Forums on this question was: its way Intel driver works.. Is this correct?
HW: Zotac H61-ITX WiFi + Intel Celeron Single-Core G440
Hope I there is solution for this. As its quite annoying, as i dont need all time TV/amplifier on, but as htpc doesn't consume much power its at my place always on, doing some compile or serving other clients..
I would like to know if intel has some plan to work in joint with AMD and NVIDIA to solve this bad hybrid graphic switching situation when using AMD/NVIDIA proprietary drivers. A lot of computers are equipped with both intel IGP and discrete GC. AMD actually supports the switch from intel and fglrx with an Xorg restart with the proprietary fglrx binary driver (ok this is an Xorg limitation and there is work to add support for hotplugging). But let's face it: radeon and nouveau are not good for everything, i'm sad to say it but i use fglrx from a long time maily for the superior powersave, but from time to time i also play some game like xonotic. When i buy a new computer i have to choose: intel IGP with its low power and low graphics capability (i really hope intel IGP will be more powerfull in the future), or an AMD/NVIDIA GC? The problem is currently a lot of laptops are hybrid, so i have to spend hours checking if the laptop actually works on linux (and with works, i mean all hw included works, if it has an NVIDIA optimus and it will not work, i will never buy it of course, even if using only the intel IGP works as expected). It would be really nice to not worry about this kind of hardware support when buying a new computer and be sure if you buy whatever combination of intel/amd/nvidia it will work. And with proprietary drivers from other vendors, since intel is the only one providing a windows-(almost?)-on-par driver for linux (thank you very much!).
It must be a myth that fglrx+intel switching really works. I can not confirm this.