Difference between revisions of "Talk:How to make use of Graphics Chips Power Management features"

From ThinkWiki
Jump to: navigation, search
(aticonfig and Xorg.0.log don't match)
(aticonfig and Xorg.0.log don't match)
 
Line 62: Line 62:
  
 
--[[User:Dave abrahams|Dave abrahams]] 21:21, 30 August 2006 (CEST)
 
--[[User:Dave abrahams|Dave abrahams]] 21:21, 30 August 2006 (CEST)
 +
 +
:Same here on my Z61m wit x1400
 +
:--[[User:Zhenech|Zhenech]] 00:50, 6 September 2006 (CEST)
  
 
== DynamicClocks setting on suspend/resume? ==
 
== DynamicClocks setting on suspend/resume? ==
  
 
I am using the opensource ATI driver. I noticed in the article that it says that the DynamicClocks setting is lost on hibernation to disk. I am wondering if it is also lost on suspend. How can I check to see the current state of DynamicClocks?
 
I am using the opensource ATI driver. I noticed in the article that it says that the DynamicClocks setting is lost on hibernation to disk. I am wondering if it is also lost on suspend. How can I check to see the current state of DynamicClocks?

Latest revision as of 23:50, 5 September 2006

Experimentally, it seems that rovclock determines the maximum frequency, and "DynamicClocks" tells the chip to a lower frequency when possible. They are thus complementary. --Thinker 18:59, 27 Oct 2005 (CEST)

show current power state with fglrx?

Switching power states using aticonfig seems to work fine. Seems, because I can't really see in which state the ATI chip is currently in. Or can I?

--spiney 14:40, 20 Nov 2005 (CET)


You can (destructively) check whether it's in a specific state by trying to switch to that state. If it's alredy there, it will give an error. If not, it will switch and (on my machine) cause a brief screen blink. Indeed, brilliant engineering.

--Thinker 15:32, 20 Nov 2005 (CET)


aticonfig and Xorg.0.log don't match

I was testing this PowerPlay business and I saw that aticonfig --lsp outputs:

  core/mem      [flags]
---------------
1: 105/122 MHz  [low voltage]
2: 209/182 MHz  [low voltage]
3: 297/230 MHz  [default state]

But Xorg.0.log reports that the states are:

(II) fglrx(0): POWERplay version 3.  4 power states available:
(II) fglrx(0):   1. 297/230MHz @ 60Hz [enable load balancing]
(II) fglrx(0):   2. 105/122MHz @ 60Hz [low voltage, enable sleep]
(II) fglrx(0):   3. 250/230MHz @ 60Hz [thermal diode mode]
(II) fglrx(0):   4. 209/182MHz @ 60Hz [low voltage]

Am I the only one that think this is kind of odd? I'm using a R52 with a X300 card. Omarkj



I can confirm I have the same values on the same hardware. Also, it seems rovclock can't read the correct speeds. It reads values well over 400MHz. --Micampe 14:11, 2 Dec 2005 (CET)


On my T60p it's even weirder:

     core/mem      [flags]
 -----------------
   1: 128/135 MHz  [low voltage]
 *    209/135 MHz  [low voltage]
   2: 324/135 MHz
   3: 398/324 MHz  [default state]

This is the result after

 aticonfig --set-powerstate=1

Note that the "real" state 1 is unreachable and the state I'm actually in is unnamed.

 (II) fglrx(0): POWERplay version 3.  5 power states available:
 (II) fglrx(0):   1. 398/324MHz @ 60Hz [enable load balancing]
 (II) fglrx(0):   2. 128/135MHz @ 60Hz [low voltage, enable sleep]
 (II) fglrx(0):   3. 209/135MHz @ 60Hz [low voltage, enable sleep]
 (II) fglrx(0):   4. 324/135MHz @ 60Hz [enable sleep]
 (II) fglrx(0):   5. 344/324MHz @ 60Hz [enable sleep, thermal diode mode]

--Dave abrahams 21:21, 30 August 2006 (CEST)

Same here on my Z61m wit x1400
--Zhenech 00:50, 6 September 2006 (CEST)

DynamicClocks setting on suspend/resume?

I am using the opensource ATI driver. I noticed in the article that it says that the DynamicClocks setting is lost on hibernation to disk. I am wondering if it is also lost on suspend. How can I check to see the current state of DynamicClocks?