I sent this to Alex a few days ago but felt I should post it on this list as well:
I've been testing radeon KMS PM with 2.6.34-rc* for a few days now and I wanted to send you my testcase. Even though PM is enabled, the defaults of my card are somewhat insane so no real powermanagement takes place - i.e. the card doesn't get clocked down. is there any way to force the card to use a different setting where less power is being consumed? Thanks!
Klaus
Below you'll find the output of dmesg | grep drm which should provide you with the needed information:
klaus@klaus-imac:/etc/acpi/events$ dmesg | grep drm [ 1.011163] [drm] Initialized drm 1.1.0 20060810 [ 1.069646] [drm] radeon kernel modesetting enabled. [ 1.071157] [drm] initializing kernel modesetting (RV630 0x1002:0x9583). [ 1.071247] [drm] register mmio base: 0xD0520000 [ 1.071249] [drm] register mmio size: 65536 [ 1.071454] [drm] Clocks initialized ! [ 1.071456] [drm] Internal thermal controller without fan control [ 1.071458] [drm] 4 Power State(s) [ 1.071460] [drm] State 0 Default (default) [ 1.071461] [drm] 16 PCIE Lanes [ 1.071462] [drm] 3 Clock Mode(s) [ 1.071463] [drm] 0 engine/memory: 680000/750000 [ 1.071465] [drm] 1 engine/memory: 680000/750000 [ 1.071466] [drm] 2 engine/memory: 680000/750000 [ 1.071468] [drm] State 1 Performance [ 1.071469] [drm] 16 PCIE Lanes [ 1.071470] [drm] 3 Clock Mode(s) [ 1.071471] [drm] 0 engine/memory: 110000/750000 [ 1.071473] [drm] 1 engine/memory: 680000/750000 [ 1.071474] [drm] 2 engine/memory: 680000/750000 [ 1.071475] [drm] State 2 Default [ 1.071476] [drm] 16 PCIE Lanes [ 1.071477] [drm] 3 Clock Mode(s) [ 1.071479] [drm] 0 engine/memory: 680000/750000 [ 1.071480] [drm] 1 engine/memory: 680000/750000 [ 1.071482] [drm] 2 engine/memory: 680000/750000 [ 1.071483] [drm] State 3 Performance [ 1.071484] [drm] 16 PCIE Lanes [ 1.071485] [drm] 3 Clock Mode(s) [ 1.071486] [drm] 0 engine/memory: 300000/750000 [ 1.071488] [drm] 1 engine/memory: 680000/750000 [ 1.071489] [drm] 2 engine/memory: 680000/750000 [ 1.071493] [drm] radeon: dynamic power management enabled [ 1.071494] [drm] radeon: power management initialized [ 1.073069] [drm] Detected VRAM RAM=256M, BAR=256M [ 1.073072] [drm] RAM width 128bits DDR [ 1.073144] [drm] radeon: 256M of VRAM memory ready [ 1.073145] [drm] radeon: 512M of GTT memory ready. [ 1.073206] [drm] radeon: using MSI. [ 1.073236] [drm] radeon: irq initialized. [ 1.073236] [drm] GART: num cpu pages 131072, num gpu pages 131072 [ 1.073874] [drm] Loading RV630 Microcode [ 1.115484] [drm] ring test succeeded in 1 usecs [ 1.115555] [drm] radeon: ib pool ready. [ 1.115614] [drm] ib test succeeded in 0 usecs [ 1.115616] [drm] Enabling audio support [ 1.115637] [drm] Default TV standard: NTSC [ 1.115711] [drm] Default TV standard: NTSC [ 1.115713] [drm] Radeon Display Connectors [ 1.115714] [drm] Connector 0: [ 1.115715] [drm] DVI-I [ 1.115716] [drm] HPD1 [ 1.115718] [drm] DDC: 0x7e40 0x7e40 0x7e44 0x7e44 0x7e48 0x7e48 0x7e4c 0x7e4c [ 1.115719] [drm] Encoders: [ 1.115720] [drm] CRT1: INTERNAL_KLDSCP_DAC1 [ 1.115722] [drm] DFP1: INTERNAL_KLDSCP_TMDS1 [ 1.115723] [drm] TV1: INTERNAL_KLDSCP_DAC2 [ 1.115724] [drm] Connector 1: [ 1.115725] [drm] LVDS [ 1.115727] [drm] DDC: 0x7e50 0x7e50 0x7e54 0x7e54 0x7e58 0x7e58 0x7e5c 0x7e5c [ 1.115728] [drm] Encoders: [ 1.115729] [drm] LCD1: INTERNAL_LVTM1 [ 1.169669] [drm] Requested: e: 68000 m: 75000 p: 16 [ 1.169671] [drm] Setting: e: 68000 m: 75000 p: 16 [ 1.520024] [drm] fb mappable at 0xC0141000 [ 1.520025] [drm] vram apper at 0xC0000000 [ 1.520027] [drm] size 9216000 [ 1.520028] [drm] fb depth is 24 [ 1.520029] [drm] pitch is 7680 [ 1.520061] fb0: radeondrmfb frame buffer device [ 1.520067] [drm] Initialized radeon 2.3.0 20080528 for 0000:01:00.0 on minor 0 [ 1.524098] [drm] Requested: e: 68000 m: 75000 p: 16 [ 1.524101] [drm] Setting: e: 68000 m: 75000 p: 16 [ 2.130063] [drm] Requested: e: 68000 m: 75000 p: 16 [ 2.130066] [drm] Setting: e: 68000 m: 75000 p: 16 [ 2.141345] [drm] not in vbl for pm change 00020002 00000000 at entry [ 497.545444] [drm] Requested: e: 68000 m: 75000 p: 16 [ 497.545459] [drm] Setting: e: 68000 m: 75000 p: 16 [ 497.545514] [drm] not in vbl for pm change 00020002 00000000 at entry [ 497.545684] [drm] not in vbl for pm change 00020002 00000000 at exit [ 498.650121] [drm] Requested: e: 68000 m: 75000 p: 16 [ 498.650131] [drm] Setting: e: 68000 m: 75000 p: 16 [ 501.560135] [drm] Requested: e: 68000 m: 75000 p: 16 [ 501.560145] [drm] Setting: e: 68000 m: 75000 p: 16 [ 504.560137] [drm] Requested: e: 68000 m: 75000 p: 16 [ 504.560146] [drm] Setting: e: 68000 m: 75000 p: 16
On Tue, Apr 27, 2010 at 5:35 AM, Klaus Doblmann B.A. klaus.doblmann@gmail.com wrote:
I sent this to Alex a few days ago but felt I should post it on this list as well:
You might want to check the address you used as I never got the email.
I've been testing radeon KMS PM with 2.6.34-rc* for a few days now and I wanted to send you my testcase. Even though PM is enabled, the defaults of my card are somewhat insane so no real powermanagement takes place - i.e. the card doesn't get clocked down. is there any way to force the card to use a different setting where less power is being consumed?
The current code doesn't handle a lot of cases properly. Please try my latest patch set against Dave's drm-next tree: http://people.freedesktop.org/~agd5f/pm3/ It allows you to enable dynamic pm or force a static power mode via sysfs.
Alex
Thanks!
Klaus
Below you'll find the output of dmesg | grep drm which should provide you with the needed information:
klaus@klaus-imac:/etc/acpi/events$ dmesg | grep drm [ 1.011163] [drm] Initialized drm 1.1.0 20060810 [ 1.069646] [drm] radeon kernel modesetting enabled. [ 1.071157] [drm] initializing kernel modesetting (RV630 0x1002:0x9583). [ 1.071247] [drm] register mmio base: 0xD0520000 [ 1.071249] [drm] register mmio size: 65536 [ 1.071454] [drm] Clocks initialized ! [ 1.071456] [drm] Internal thermal controller without fan control [ 1.071458] [drm] 4 Power State(s) [ 1.071460] [drm] State 0 Default (default) [ 1.071461] [drm] 16 PCIE Lanes [ 1.071462] [drm] 3 Clock Mode(s) [ 1.071463] [drm] 0 engine/memory: 680000/750000 [ 1.071465] [drm] 1 engine/memory: 680000/750000 [ 1.071466] [drm] 2 engine/memory: 680000/750000 [ 1.071468] [drm] State 1 Performance [ 1.071469] [drm] 16 PCIE Lanes [ 1.071470] [drm] 3 Clock Mode(s) [ 1.071471] [drm] 0 engine/memory: 110000/750000 [ 1.071473] [drm] 1 engine/memory: 680000/750000 [ 1.071474] [drm] 2 engine/memory: 680000/750000 [ 1.071475] [drm] State 2 Default [ 1.071476] [drm] 16 PCIE Lanes [ 1.071477] [drm] 3 Clock Mode(s) [ 1.071479] [drm] 0 engine/memory: 680000/750000 [ 1.071480] [drm] 1 engine/memory: 680000/750000 [ 1.071482] [drm] 2 engine/memory: 680000/750000 [ 1.071483] [drm] State 3 Performance [ 1.071484] [drm] 16 PCIE Lanes [ 1.071485] [drm] 3 Clock Mode(s) [ 1.071486] [drm] 0 engine/memory: 300000/750000 [ 1.071488] [drm] 1 engine/memory: 680000/750000 [ 1.071489] [drm] 2 engine/memory: 680000/750000 [ 1.071493] [drm] radeon: dynamic power management enabled [ 1.071494] [drm] radeon: power management initialized [ 1.073069] [drm] Detected VRAM RAM=256M, BAR=256M [ 1.073072] [drm] RAM width 128bits DDR [ 1.073144] [drm] radeon: 256M of VRAM memory ready [ 1.073145] [drm] radeon: 512M of GTT memory ready. [ 1.073206] [drm] radeon: using MSI. [ 1.073236] [drm] radeon: irq initialized. [ 1.073236] [drm] GART: num cpu pages 131072, num gpu pages 131072 [ 1.073874] [drm] Loading RV630 Microcode [ 1.115484] [drm] ring test succeeded in 1 usecs [ 1.115555] [drm] radeon: ib pool ready. [ 1.115614] [drm] ib test succeeded in 0 usecs [ 1.115616] [drm] Enabling audio support [ 1.115637] [drm] Default TV standard: NTSC [ 1.115711] [drm] Default TV standard: NTSC [ 1.115713] [drm] Radeon Display Connectors [ 1.115714] [drm] Connector 0: [ 1.115715] [drm] DVI-I [ 1.115716] [drm] HPD1 [ 1.115718] [drm] DDC: 0x7e40 0x7e40 0x7e44 0x7e44 0x7e48 0x7e48 0x7e4c 0x7e4c [ 1.115719] [drm] Encoders: [ 1.115720] [drm] CRT1: INTERNAL_KLDSCP_DAC1 [ 1.115722] [drm] DFP1: INTERNAL_KLDSCP_TMDS1 [ 1.115723] [drm] TV1: INTERNAL_KLDSCP_DAC2 [ 1.115724] [drm] Connector 1: [ 1.115725] [drm] LVDS [ 1.115727] [drm] DDC: 0x7e50 0x7e50 0x7e54 0x7e54 0x7e58 0x7e58 0x7e5c 0x7e5c [ 1.115728] [drm] Encoders: [ 1.115729] [drm] LCD1: INTERNAL_LVTM1 [ 1.169669] [drm] Requested: e: 68000 m: 75000 p: 16 [ 1.169671] [drm] Setting: e: 68000 m: 75000 p: 16 [ 1.520024] [drm] fb mappable at 0xC0141000 [ 1.520025] [drm] vram apper at 0xC0000000 [ 1.520027] [drm] size 9216000 [ 1.520028] [drm] fb depth is 24 [ 1.520029] [drm] pitch is 7680 [ 1.520061] fb0: radeondrmfb frame buffer device [ 1.520067] [drm] Initialized radeon 2.3.0 20080528 for 0000:01:00.0 on minor 0 [ 1.524098] [drm] Requested: e: 68000 m: 75000 p: 16 [ 1.524101] [drm] Setting: e: 68000 m: 75000 p: 16 [ 2.130063] [drm] Requested: e: 68000 m: 75000 p: 16 [ 2.130066] [drm] Setting: e: 68000 m: 75000 p: 16 [ 2.141345] [drm] not in vbl for pm change 00020002 00000000 at entry [ 497.545444] [drm] Requested: e: 68000 m: 75000 p: 16 [ 497.545459] [drm] Setting: e: 68000 m: 75000 p: 16 [ 497.545514] [drm] not in vbl for pm change 00020002 00000000 at entry [ 497.545684] [drm] not in vbl for pm change 00020002 00000000 at exit [ 498.650121] [drm] Requested: e: 68000 m: 75000 p: 16 [ 498.650131] [drm] Setting: e: 68000 m: 75000 p: 16 [ 501.560135] [drm] Requested: e: 68000 m: 75000 p: 16 [ 501.560145] [drm] Setting: e: 68000 m: 75000 p: 16 [ 504.560137] [drm] Requested: e: 68000 m: 75000 p: 16 [ 504.560146] [drm] Setting: e: 68000 m: 75000 p: 16
-- Klaus Doblmann B.A. - http://straightrazorguy.net - FSF member #7570 PGP-Key: http://www.doblmann.de/pgp_key.asc
A: Because it messes up the order in which people normally read text. See http://en.wikipedia.org/wiki/Posting_style Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
On Tue, 27 Apr 2010 13:42:05 -0400 Alex Deucher alexdeucher@gmail.com wrote:
On Tue, Apr 27, 2010 at 5:35 AM, Klaus Doblmann B.A. klaus.doblmann@gmail.com wrote:
I sent this to Alex a few days ago but felt I should post it on this list as well:
You might want to check the address you used as I never got the email.
Hi Alex, weird, I double-checked and I sent it to the address you replied from...
I've been testing radeon KMS PM with 2.6.34-rc* for a few days now and I wanted to send you my testcase. Even though PM is enabled, the defaults of my card are somewhat insane so no real powermanagement takes place - i.e. the card doesn't get clocked down. is there any way to force the card to use a different setting where less power is being consumed?
The current code doesn't handle a lot of cases properly. Please try my latest patch set against Dave's drm-next tree: http://people.freedesktop.org/~agd5f/pm3/ It allows you to enable dynamic pm or force a static power mode via sysfs.
Alex
Will do so at the end of the week when (hopefully) 2.6.34-rc6 is out and I have mor etime on my hands. Could you a bit more explicit on how I could set a static power mode via sysfs? I've never worked with sysfs before so some hints would be greatly appreciated.
Thanks, Klaus
PS: On another note, is the KMS PM code supposed to work with r100 chips (Mobility 7500)? I've tried it yesterday and the machine (IBM T40) hardlocks when loading GDM (Ubuntu 10.04, 2.6.34-rc5 without any additional patches).
Thanks!
Klaus
[...]
On Tue, Apr 27, 2010 at 1:53 PM, Klaus Doblmann B.A. klaus.doblmann@gmail.com wrote:
On Tue, 27 Apr 2010 13:42:05 -0400 Alex Deucher alexdeucher@gmail.com wrote:
On Tue, Apr 27, 2010 at 5:35 AM, Klaus Doblmann B.A. klaus.doblmann@gmail.com wrote:
I sent this to Alex a few days ago but felt I should post it on this list as well:
You might want to check the address you used as I never got the email.
Hi Alex, weird, I double-checked and I sent it to the address you replied from...
I've been testing radeon KMS PM with 2.6.34-rc* for a few days now and I wanted to send you my testcase. Even though PM is enabled, the defaults of my card are somewhat insane so no real powermanagement takes place - i.e. the card doesn't get clocked down. is there any way to force the card to use a different setting where less power is being consumed?
The current code doesn't handle a lot of cases properly. Please try my latest patch set against Dave's drm-next tree: http://people.freedesktop.org/~agd5f/pm3/ It allows you to enable dynamic pm or force a static power mode via sysfs.
Alex
Will do so at the end of the week when (hopefully) 2.6.34-rc6 is out and I have mor etime on my hands. Could you a bit more explicit on how I could set a static power mode via sysfs? I've never worked with sysfs before so some hints would be greatly appreciated.
enable/disable dynpm: echo 1 > /sys/class/drm/card0/device/dynpm
force a static power state: echo 1.0 > /sys/class/drm/card0/device/power_state
Alex
Thanks, Klaus
PS: On another note, is the KMS PM code supposed to work with r100 chips (Mobility 7500)? I've tried it yesterday and the machine (IBM T40) hardlocks when loading GDM (Ubuntu 10.04, 2.6.34-rc5 without any additional patches).
In theory, although I haven't had time to test older chips much yet.
Alex
Thanks!
Klaus
[...]
-- Klaus Doblmann B.A. - http://straightrazorguy.net - FSF member #7570 PGP-Key: http://www.doblmann.de/pgp_key.asc
A: Because it messes up the order in which people normally read text. See http://en.wikipedia.org/wiki/Posting_style Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail?
On Tue, 27 Apr 2010 14:06:41 -0400 Alex Deucher alexdeucher@gmail.com wrote:
[...]
enable/disable dynpm: echo 1 > /sys/class/drm/card0/device/dynpm
force a static power state: echo 1.0 > /sys/class/drm/card0/device/power_state
Alex
Oh, now that you've mentioned it, I've noticed I've been using sysfs for years in my customized fan-script for my machine without knowing^^
Thanks for clearing that up!
PS: On another note, is the KMS PM code supposed to work with r100 chips (Mobility 7500)? I've tried it yesterday and the machine (IBM T40) hardlocks when loading GDM (Ubuntu 10.04, 2.6.34-rc5 without any additional patches).
In theory, although I haven't had time to test older chips much yet.
Alex
I'll test the new patches ith .34-rc6 on the weekend too and will report back. Whenever you need a testcase for ancient r100 hardware feel free to mail me (I'll be reading the list too)
Klaus
On Tue, 27 Apr 2010 14:06:41 -0400 Alex Deucher alexdeucher@gmail.com wrote:
I've been testing radeon KMS PM with 2.6.34-rc* for a few days now and I wanted to send you my testcase. Even though PM is enabled, the defaults of my card are somewhat insane so no real powermanagement takes place - i.e. the card doesn't get clocked down. is there any way to force the card to use a different setting where less power is being consumed?
The current code doesn't handle a lot of cases properly. Please try my latest patch set against Dave's drm-next tree: http://people.freedesktop.org/~agd5f/pm3/ It allows you to enable dynamic pm or force a static power mode via sysfs.
Alex
Will do so at the end of the week when (hopefully) 2.6.34-rc6 is out and I have mor etime on my hands. Could you a bit more explicit on how I could set a static power mode via sysfs? I've never worked with sysfs before so some hints would be greatly appreciated.
enable/disable dynpm: echo 1 > /sys/class/drm/card0/device/dynpm
force a static power state: echo 1.0 > /sys/class/drm/card0/device/power_state
Alex
Hi Alex,
I just built drm-radeon-testing as I saw the patches have already been merged there. It built fine, I updated the firmware files as the Xor-Wiki suggests, but when I boot into the resulting kernel with dynpem on my screen goes all wonky, a portion on the right side is missing and I've got vertical lines running down the screen. Somehow I managed to look at dmesg and saw that the speed was being set correctly (it even set the lowest one available).
When I tried "sudo echo 0 > /sys/class/drm/card0/device/dynpm" to see whether switching off dynpm would fix the "drunken" screen the system told me I was not permitted to access the file...
I am attaching the syslog from the faulty boot in case it's of any help...
Klaus
On Sun, May 2, 2010 at 3:13 PM, Klaus Doblmann B.A. klaus.doblmann@gmail.com wrote:
On Tue, 27 Apr 2010 14:06:41 -0400 Alex Deucher alexdeucher@gmail.com wrote:
I've been testing radeon KMS PM with 2.6.34-rc* for a few days now and I wanted to send you my testcase. Even though PM is enabled, the defaults of my card are somewhat insane so no real powermanagement takes place - i.e. the card doesn't get clocked down. is there any way to force the card to use a different setting where less power is being consumed?
The current code doesn't handle a lot of cases properly. Please try my latest patch set against Dave's drm-next tree: http://people.freedesktop.org/~agd5f/pm3/ It allows you to enable dynamic pm or force a static power mode via sysfs.
Alex
Will do so at the end of the week when (hopefully) 2.6.34-rc6 is out and I have mor etime on my hands. Could you a bit more explicit on how I could set a static power mode via sysfs? I've never worked with sysfs before so some hints would be greatly appreciated.
enable/disable dynpm: echo 1 > /sys/class/drm/card0/device/dynpm
force a static power state: echo 1.0 > /sys/class/drm/card0/device/power_state
Alex
Hi Alex,
I just built drm-radeon-testing as I saw the patches have already been merged there. It built fine, I updated the firmware files as the Xor-Wiki suggests, but when I boot into the resulting kernel with dynpem on my screen goes all wonky, a portion on the right side is missing and I've got vertical lines running down the screen. Somehow I managed to look at dmesg and saw that the speed was being set correctly (it even set the lowest one available).
Thing should work better with the patches here: http://people.freedesktop.org/~agd5f/pm-drt/ on top of d-r-t.
When I tried "sudo echo 0 > /sys/class/drm/card0/device/dynpm" to see whether switching off dynpm would fix the "drunken" screen the system told me I was not permitted to access the file...
You need to be root to access the file. sudo can't handle redirects, so you need something like: sudo bash -c "echo 0 > /sys/class/drm/card0/device/dynpm"
I am attaching the syslog from the faulty boot in case it's of any help...
Klaus
-- Klaus Doblmann B.A. - http://straightrazorguy.net - FSF member #7570 PGP-Key: http://www.doblmann.de/pgp_key.asc http://twitter.com/klausdoblmann
A: Because it messes up the order in which people normally read text. See http://en.wikipedia.org/wiki/Posting_style Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail?
dri-devel@lists.freedesktop.org
-
Alex Deucher -
Klaus Doblmann B.A.