How to troubleshoot intel 3D performance?

Alex Bennee kernel-hacker at
Tue Feb 17 01:46:33 PST 2009

2009/2/17 Colin Guthrie <gmane at>:
> 'Twas brillig, and Vasily Khoruzhick at 17/02/09 09:05 did gyre and gimble:
>> On Tuesday 17 February 2009 10:12:56 you wrote:
>>> 2009/2/16 Vasily Khoruzhick <anarsoul at>:
>>>> Hi, performance issue is known thing on pre-965 chipsets with new intel
>>>> stuff (gem, uxa).
>>>> I've filed bugs about them:
>>>> AFAIK there's no fix at the moment, but as workaround you can use 2.6.27
>>>> kernel and EXA accelleration.
>>> Looking at the code it looks like a failure of the kernel side. Is
>>> there a kernel tree that tracks the Linus one with the latest pending
>>> fixes I can try?
>> I suspect this one:
>> git://
> I believe the "drm-next" branch is usually the bleeding edge stuff fro
> the next stable kernel (but I'm not 100% sure on this).

I'll give that a spin.

An additional data point, Dave Airle's tree (commit
eaaffe54249d4444bb4de3631b9d968456bb0458) gave a slightly different
error on start up:

(II) [drm] DRM open master succeeded.
(II) intel(0): [drm] Using the DRM lock SAREA also for drawables.
(II) intel(0): [drm] framebuffer mapped by ddx driver
(II) intel(0): [drm] added 1 reserved context for kernel
(II) intel(0): X context handle = 0x1
(II) intel(0): [drm] installed DRM signal handler
(**) intel(0): Framebuffer compression disabled
(**) intel(0): Tiling enabled
(==) intel(0): VideoRam: 262144 KB
(II) intel(0): Attempting memory allocation with tiled buffers.
(EE) intel(0): Failed to set tiling on front buffer: Invalid argument
(EE) intel(0): Failed to set tiling on back buffer: Invalid argument
(EE) intel(0): Failed to set tiling on depth buffer: Invalid argument
(II) intel(0): Tiled allocation successful.
(II) intel(0): [drm] Registers = 0xdfd00000
(II) intel(0): [drm] Initialized kernel agp heap manager, 33554432
(II) intel(0): [dri] visual configs initialized
(II) intel(0): Page Flipping disabled
(II) intel(0): vgaHWGetIOBase: hwp->IOBase is 0x03d0, hwp->PIOOffset is 0x0000
(II) EXA(0): Offscreen pixmap area of 25804800 bytes
(II) EXA(0): Driver registered support for the following operations:
(II)         Solid
(II)         Copy
(II)         Composite (RENDER acceleration)
(==) intel(0): Backing store disabled
(==) intel(0): Silken mouse enabled
(II) intel(0): Initializing HW Cursor
(II) intel(0): [DRI] installation complete
(II) intel(0): Current clock rate multiplier: 8
(II) intel(0): xf86BindGARTMemory: bind key 0 at 0x0c759000 (pgoffset 51033)
(II) intel(0): xf86BindGARTMemory: bind key 1 at 0x0dff5000 (pgoffset 57333)
(II) intel(0): Fixed memory allocation layout:
(II) intel(0): 0x00000000-0x00009fff: HW cursors (40 kB,
0x00000000b7800000 physical
(II) intel(0): 0x0000a000-0x0000afff: overlay registers (4 kB,
0x00000000b780a000 physical
(II) intel(0): 0x007bf000:            end of stolen memory
(II) intel(0): 0x007bf000-0x0c758fff: DRI memory manager (196200 kB)
(II) intel(0): 0x0c759000-0x0dff4fff: exa offscreen (25200 kB)
(II) intel(0): 0x0dff5000-0x0fff4fff: classic textures (32768 kB)
(II) intel(0): 0x10000000:            end of aperture
(II) intel(0): BO memory allocation layout:
(II) intel(0): 0x007bf000:            start of memory manager
(II) intel(0): 0x007df000-0x0101efff: depth buffer (8448 kB)
(II) intel(0): 0x017df000-0x0201efff: back buffer (8448 kB)
(II) intel(0): 0x02800000-0x0351ffff: front buffer (13440 kB)
(II) intel(0): 0x027df000-0x027e6fff: logical 3D context (32 kB)
(II) intel(0): 0x0c759000:            end of memory manager

But that may well be because it went through a different call path
still being EXA. I shall give the latest "official" intel tree a go
and report back.

> Col
> --
> Colin Guthrie
> gmane(at)
> Day Job:
>   Tribalogic Limited []
> Open Source:
>   Mandriva Linux Contributor []
>   PulseAudio Hacker []
>   Trac Hacker []
> _______________________________________________
> xorg mailing list
> xorg at

Alex, homepage:

More information about the xorg mailing list