[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: The nightmare of Intel Integrated GPUs under Linux in general and Debian in particular



Miguel A. Vallejo composed on 2020-02-11 20:07 (UTC+0100):

> What are your recommendations / experiences?

Nothing like yours or all those bug reports. I've been using Intel, Nvidia and
AMD/ATI since over 15 years ago, but more with Intel than the others in recent
years, and across several distros other than just Debian. This PC is a (MSI)
Haswell, with HD Graphics 4400 (but only openSUSE). I have two with HD Graphics
630 and Buster on Kaby Lake, i3-7100t and G4600. On neither of these latter two
are any cmdline options re graphics necessary, nor is anything in xorg.conf*. My
trouble with Intel to the exclusion of the other two brands disappeared what now
seems like a decade or more. All mine supported by the modesetting DDX are using
it. The two Kaby Lake GPUs are supporting 3 displays simultaneously on stock
Buster 4.19 kernels. I spend very little time running Chrom*. Mostly I use latest
SeaMonkey, Firefox ESR, and New (Pale) Moon. I don't have any Debian installations
running KDE, only TDE (current, next, & devel) and IceWM on those. And, I do no
virtualization except for DOS on an OS/2 derivative.

I do get no working Xorg from some installations, but those are mostly worked
around by not disabling compositing globally in /etc/X11/xorg.conf.d/.

OTOH, support forums I frequent seem to have far more complaints about NVidia than
the other two combined, out of proportion to each's installed base, very roughly
split between tainting driver installation, and maintenance (laptop and desktop),
and multi-GPU problems (Optimus laptops mostly).

If you want a solution with absence of reverse-engineering or kernel tainting, I
suggest keeping after the devs and making sure to answer any requests you might
get there promptly in https://gitlab.freedesktop.org/drm/intel/issues/673, and
offer additional testing and logs on the intel-gfx@lists.freedesktop.org mailing
list referencing that bug report. This has obviously been going on much too long
for (against) you. :-(

Until this thread and that bug report, I was completely unaware of the extent of
recent Intel GPU complaints that that bug represents.

Have you ever run several hours (4-10, or overnight) of memtest86 (not
memtest86+)? Intel GPUs all run on shared RAM. All mine are running paired RAM
sticks in dual channel, and with equally balanced CAS ratings, e.g. 15-15-15-36 or
17-17-17-39 vs. 15-17-17-35 or 16-15-15-37.
-- 
Evolution as taught in public schools is religion, not science.

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/


Reply to: