HOWTO: Jaunty Intel Graphics Performance Guide

[DavetheDude]

Thanks for the help, I followed the safe procedure and compiz runs notably faster.

A couple of strange things though... on restart it didn't seem to work and I was taken to a screen suggesting that I allow it to attempt to 'autofix' the problem. I proceeded and the greeter started as normal and have been fine since. I would have catalogued this process more carefully, only I didn't detect any problems until now, some days later. Today I noticed that I no longer have access to virtual terminals via the ctrl-alt-function. Instead, I get some flickering and then the ubuntu greeter. This also seems to end my original gnome session (ctrl-alt-f7). Trying this a few times produced the message 'the greeter application appears to be crashing, attempting to use different one'.

So, as something of an epic n00b, I humbly request assistance from any who may be able to provide it.

Cheers.

[Andrew May]

Hi,
I am running Jaunty on a desktop machine.
I have been following the evolution of these instructions and have taken the plunge as a newbie and applied the changes for optimal effect.
They have done nothing except give me 1280x1024 resolution which was not available before.
I cannot activate visual effects to normal setting and when I run PPRacer I have about 2 (yes two!) fps, which is what I had before the changes. I have restarted several times, and it tells me it is already applied:
Supplying corrected MTRR ranges to /proc/mtrr
doing nothing, MTRR range already set up
reg00: base=0x000000000 ( 0MB), size= 1024MB, count=1: write-back
reg01: base=0x03f800000 ( 1016MB), size= 8MB, count=1: uncachable
reg02: base=0x0f0000000 ( 3840MB), size= 128MB, count=1: write-combining

I have a dual core P4 3GHz with intel integrated graphics:

00:02.0 VGA compatible controller: Intel Corporation 82865G Integrated Graphics Controller (rev 02)
Subsystem: ASUSTeK Computer Inc. Device 80a5
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
Latency: 0
Interrupt: pin A routed to IRQ 10
Region 0: Memory at f0000000 (32-bit, prefetchable) [size=128M]
Region 1: Memory at fbe80000 (32-bit, non-prefetchable) [size=512K]
Region 2: I/O ports at d800 [size=8]
Capabilities: <access denied>
Kernel modules: intelfb

What am I doing wrong? any ideas?
The rest of the computer functions fine...just no games or effects.
thanks,
Andrew

[gambothell]

Thanks for such great support. I used safe/optimal method on a Dell Inspiron 1420 with an Intel GM965. My wife now can watch Hulu ...


Your the Man

[Vadi]

Can someone tell how to stop the LCD randomly flashing?
That's all I want.
I repeat: I don't care about 3D, or EXA, or UXA, or no accel.
Just tell me how to put an Intel X3100+965GMA powered laptop display to a sane state.

Email support@intel.com asking them why did they make such buggy drivers and didn't finish new ones in time? Ubuntu isn't the one that makes the Intel drivers and breaks them.

[lucis]

Email support@intel.com asking them why did they make such buggy drivers and didn't finish new ones in time? Ubuntu isn't the one that makes the Intel drivers and breaks them.

I would personally be more interested in hearing Ubuntu's side of the story. The new infrastructure in the kernel and in xorg are experimental, so why did the Ubuntu devs include them in the first place? I think it's pretty appalling that this thread has grown to 47 pages and I have yet to hear any official word from Ubuntu on the matter.

When Jaunty breaks on my laptop, that's fine. I'm a geek and I know how to fix it. When it breaks on the machine of my non-geek friend who wants a viable alternative to virus-ridden Windows, that's a totally different issue. Where does she go when she can't do basic tasks without huge headaches? What a debacle.

[psyke83]

I would personally be more interested in hearing Ubuntu's side of the story. The new infrastructure in the kernel and in xorg are experimental, so why did the Ubuntu devs include them in the first place? I think it's pretty appalling that this thread has grown to 47 pages and I have yet to hear any official word from Ubuntu on the matter.

When Jaunty breaks on my laptop, that's fine. I'm a geek and I know how to fix it. When it breaks on the machine of my non-geek friend who wants a viable alternative to virus-ridden Windows, that's a totally different issue. Where does she go when she can't do basic tasks without huge headaches? What a debacle.

I suggest you do some research before making accusations. The Ubuntu developers did not ship Jaunty with the new acceleration framework for the reason that it was less stable.

The default Jaunty driver uses EXA by default (whereas the same upstream version uses UXA), and using UXA caused system freezes for many users. Whilst EXA suffered from some performance regressions, its stability was roughly the same as Intrepid.

In other words, what you think should have happened, did happen.

[Arup]

Why blame Ubuntu, they are giving us a choice of extremely stable and quick 8.04.2 LTS with support till 2011 and then you have cutting edge Jaunty which for sake of progress has to incorporate the latest stuff and only then can Ubuntu move on. Point being that Ubuntu of all the distros is far less prone to sacrifice stability for the sake of latest and speed, they always act with responsibility compared to many others who add all the latest kernel, xorg, etc. just to make their release look attractive at the sake of compatibility and stability.

In case you have issues with Jaunty, reverting back to 8.04 LTS is your best bet, its still supported and it runs fast as well.

Speaking about Intel drivers, they have myriads of issues in Windows world as well and I have used them on Windows so I speak from experience. Primarily till today Intel's latest 5029 still can't recognize the EDID of some leading monitors, LCD and CRT so you have to to a registry hack to get them working. Kinda annoying to see when your shiny new LCD monitor can't be set to its native resolution and thereby negating the who purpose of getting a new monitor. Then there are issues which have been typical with Intel from its early days of i740. Many games refuse to run and these don't just include high end stuff like the ones from Crysis etc. Even basic games, also Intel always has been known to skip complex frames to get their drivers to benchmark well, this still persists till today. I could go on and on but I would end it by repeating psyke83's advice of doing some research before plunging in. As a matter of fact, the amount of advice and hard work he and other give in this forum can't be touched by any Windows forum. When I ran into the resolution issue, there was no answer from Windows newsgroup, Intel reverted me back to a older modified driver, I needed to do some serious digging to find out the hack needed for the native resolution to function properly.

[VMC]

...
I have a dual core P4 3GHz with intel integrated graphics:

00:02.0 VGA compatible controller: Intel Corporation 82865G Integrated Graphics Controller (rev 02)

What am I doing wrong? any ideas?
The rest of the computer functions fine...just no games or effects.
thanks,
Andrew

Andrew, Your not doing anything wrong. I also have an Intel 82865G intregreated chip. I couldn't get my video working correctly following these instructions.

I found the Revert_to_2.4 method worked best for my chip.

[HotShotDJ]

psyke83: Only the Bleeding-Edge instructions worked for me, and it worked brilliantly. Thank you for your HOWTO.

I have only one concern. On boot-up with the RC5 Kernel, I see the message:

mounting securityfs on /sys/kernel/security   [FAILED]

Do you know what the problem is and/or what the solution may be? I haven't seen any problems on my system, but any failure (real or imagined) in my system's security makes me a bit nervous.

EDIT: Ooops... just rechecked message. Edited to correct.

[markbl]

Only the Bleeding-Edge instructions worked for me, and it worked brilliantly. Thank you for your HOWTO.

Don't count your chickens too early. It always works well just after a reboot. I've been updating my laptop in sync with the RC kernels etc and find significant stability problems still occur. I'm currently on RC5 with the bleeding edge drivers and find that after a solid day of using my laptop (with a second external screen) that the system can get very slow and frequent screen/font corruptions start occurring. A reboot will clear this but it is a pain.

It is also a massive inconvenience that uxa/dri2 (which is the only mode intel driver will soon support) can not drive a second screen side by side with chipsets before 965 (unless total screen res is < 2048 which is not feasible in this day and age). I can only configure my second screen logically below my laptop screen and in spite of using this for a while it still feels really awkward. There is no solution in near sight.

Refer https://bugs.freedesktop.org/show_bug.cgi?id=21190 and https://bugs.freedesktop.org/show_bug.cgi?id=10479.