Bug 102287 - Sax2 not enabling dynamic clocking of graphics card
Summary: Sax2 not enabling dynamic clocking of graphics card
Status: RESOLVED FIXED
Alias: None
Product: SUSE LINUX 10.0
Classification: openSUSE
Component: X.Org (show other bugs)
Version: Preview 3
Hardware: Other All
: P5 - None : Critical
Target Milestone: ---
Assignee: Stefan Behlert
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-08-05 00:35 UTC by Dirk Mueller
Modified: 2007-06-05 11:37 UTC (History)
6 users (show)

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
Patch required for enabling it by default on mobile chipsets (557 bytes, patch)
2005-08-21 21:40 UTC, Stefan Dirsch
Details | Diff
I forgot to adjust the manual page (1.00 KB, patch)
2005-08-22 09:30 UTC, Stefan Dirsch
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Dirk Mueller 2005-08-05 00:35:18 UTC
this line is missing in the generated xorg.conf:  
 
 Option "DynamicClocks" "on" 
 
section Section "Device" 
 
hwinfo --gfxcard:  
 level is implicitly set to a reasonable value.) 
schleppi:/etc/X11 # hwinfo --gfxcard 
22: PCI(AGP) 100.0: 0300 VGA compatible controller (VGA)         
  [Created at pci.271] 
  Unique ID: VCu0.Bl6sNg5xBX0 
  Parent ID: vSkL.1o+Z33xgwU4 
  SysFS ID: /devices/pci0000:00/0000:00:01.0/0000:01:00.0 
  SysFS BusID: 0000:01:00.0 
  Hardware Class: graphics card 
  Model: "IBM RV350 NT" 
  Vendor: pci 0x1002 "ATI Technologies Inc" 
  Device: pci 0x4e54 "RV350 NT" 
  SubVendor: pci 0x1014 "IBM" 
  SubDevice: pci 0x054f  
  Revision: 0x80 
  Memory Range: 0xe0000000-0xe7ffffff (rw,prefetchable) 
  I/O Ports: 0x3000-0x3fff (rw) 
  Memory Range: 0xc0100000-0xc010ffff (rw,non-prefetchable) 
  Memory Range: 0xc0120000-0xc013ffff (ro,prefetchable,disabled) 
  IRQ: 11 (20798 events) 
  I/O Ports: 0x3c0-0x3df (rw) 
  Driver Info #0: 
    XFree86 v4 Server Module: radeon 
  Config Status: cfg=new, avail=yes, need=no, active=unknown 
  Attached to: #11 (PCI bridge) 
 
Xorg.log says:  
(--) RADEON(0): Chipset: "ATI FireGL Mobility T2 (M10) NT (AGP)" (ChipID = 
0x4e54) 
 
Its a mobility chip == laptop. my laptop more easily overheats (and is louder) 
if graphics card is not properly scaled down while idle).
Comment 1 Marcus Schaefer 2005-08-09 13:16:09 UTC
reassigned to sndirsch 
Comment 2 Stefan Dirsch 2005-08-09 13:47:06 UTC
IIRC this is still experimental. Therefore I don't want to change the default,
which is currently "off" for this option.
Comment 3 Dirk Mueller 2005-08-09 13:56:43 UTC
Might have been interesting to change the default for the beta though to see 
if any real problems appear. It works just fine for me (means: it doesn't 
crash).  
Comment 4 Thomas Renninger 2005-08-21 17:55:23 UTC
Yes please do so, I have several positiv reports.
Some Thinkpad models need this option urgently to not overheat and shutdown with
a critical temperature alert.
If this will not be included in 10.0, please mention this somewhere where people
could find it easily. Affected models I know so far: T42p (#105981 - comment
#16, 53e #98178) and probably much more (also see:
http://mailman.linux-thinkpad.org/pipermail/linux-thinkpad/2005-August/thread.html
). 
I fear every Thinkpad could need this option that has a current embedded
controller firmware:
_______________
Symptom corrected by version 3.03 - 1RHT70WW
Reduced Fan noise in some models
_______________

Thinkpads are normally the laptops Novell can point to, as those are normally
working really well with Linux.

Adding a project manager for final decission and increase severity to get his
attention.
Comment 5 Thomas Renninger 2005-08-21 18:25:16 UTC
A post from the thinkpad list:
Its been enabled for months, so yes.
...
$ grep Dynamic /var/log/Xorg.0.log
(**) RADEON(0): Option "DynamicClocks" "true"
(II) RADEON(0): Dynamic Clock Scaling Enabled 
Comment 6 Stefan Dirsch 2005-08-21 21:40:43 UTC
Created attachment 46796 [details]
Patch required for enabling it by default on mobile chipsets
Comment 7 Stefan Dirsch 2005-08-21 21:41:49 UTC
Egbert, Matthias. Any comments about enabling it by default? 
Comment 8 Stefan Dirsch 2005-08-22 09:30:30 UTC
Created attachment 46845 [details]
I forgot to adjust the manual page
Comment 9 Egbert Eich 2005-08-22 10:23:22 UTC
Our 'beta' is generally not for testing experimental options but to make sure
all pieces we ship and have modified work together well (and sometimes they
don't despite of the beta phase). I would much prefer to make such things
available to our testers much earlier in the release cycle. This way such
experimental things could be given a much wider exposure and the open source
community would actually benefit from it.
I can say very little about the side effects this change may have from looking
at the code. I would suggest to enable it for selected chipsets however not
enable it by default.
Comment 10 Stefan Dirsch 2005-08-22 10:50:13 UTC
Sure we can do this, although the coverage would be *very* small. So it would be
the task of the "Mobile Devices" team to collect the data for this. I would need
the Device ID for mobile ATI chips on which this option works then.
Comment 11 Stefan Dirsch 2005-08-22 12:37:06 UTC
I've added this option now for 0x4e54. I assign this now to Thomas to collect
the data. Please assign it back to me until the end of the week, so I can update
the x11DB in time (monday) for Beta4.
Comment 12 Thomas Renninger 2005-08-22 13:20:20 UTC
Where do I find this string?
Is it graphics card or graphics driver dependant, probably both?
Thanks.
Comment 13 Stefan Dirsch 2005-08-22 13:30:13 UTC
> Where do I find this string?
lspci -n | grep 0300 :-)
Comment 14 Matthias Hopf 2005-08-22 14:08:08 UTC
I agree that it is quite late for adding an option that can make a huge
difference stability-wise. However, I would opt for including the option (off)
in /etc/X11/xorg.conf, so an interested user will stumble over it.
Comment 15 Stefan Dirsch 2005-08-22 14:10:48 UTC
Marcus, can we add this to the radeon driver profile?

  Option "DynamicClocks" "off" 
Comment 16 Egbert Eich 2005-08-22 14:43:17 UTC
Better would be:
# Option "DynamicClocks" "on"
Then the user only would have to remove the '#'.
Comment 17 Stefan Dirsch 2005-08-22 15:00:06 UTC
Don't know, whether this is possible. Marcus?
Comment 18 Thomas Renninger 2005-08-25 13:28:13 UTC
Do as you think it's best.
I am on holidays for three days..., can't collect graphics card info so shortly.
Comment 19 Stefan Dirsch 2005-08-25 13:41:49 UTC
I do not either ... Stefan, it's up to you ...
Comment 20 Egbert Eich 2005-08-25 14:38:32 UTC
This affects only a number of chipsets and probably only on certain systems.
We can add this (in a commented out form) for all mobility chips (or the newer
ones starting with a certain generation (I don't know where to draw the line) or
we can add it only for chipsets that we know need tis option. One is named above 
(0x4e54) but I don't think it's Stefan's job to match  laptop models that are
reported having this problem to chipsets.
Comment 21 Stefan Dirsch 2005-08-25 21:58:17 UTC
It shouldn't hurt to add this option in a commented out form for all chips,  
which use the radeon driver. Unfortunately I still don't know, whether this is  
possible. Marcus? (see comment 16/17) 
  
Nobody wants to maintain a list for Radeon mobility chips to add this option.  
I didn't ask for a list of radeon chips, on which this option does *not* work.  
I asked for a list of radeon chips, on which this option has been tested and  
works. No single chip has been tested so far besides the one of Dirk. 
Therefore I assume this is not that critical as it supposed to be after 
reading comment #4 ... 
Comment 22 Stefan Dirsch 2005-09-01 10:54:11 UTC
I don't expect any more test results. dmuell will get happy since the option
will be used for his gfx chip. Others might run into a problem.