I wonder how ATI GPUs are throttled, and what the default throttling state is. I'm using an ATI Mobility Radeon HD 5830 card, which is not officially supported by the radeon or the radeonhd drivers, neither the official ATI binary driver. thus, I'm using the vesa driver to drive this card. the laptop I'm using this card in idles at around 56-59 C, with all CPU cores at their minimum (checked a number of ways, they really are). I wonder if the amount of heat could be in part as a result of the GPU always running at its maximum... could this be the case? -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
I wonder how ATI GPUs are throttled, and what the default throttling state is.
I'm using an ATI Mobility Radeon HD 5830 card, which is not officially supported by the radeon or the radeonhd drivers, neither the official ATI binary driver. thus, I'm using the vesa driver to drive this card.
the laptop I'm using this card in idles at around 56-59 C, with all CPU cores at their minimum (checked a number of ways, they really are). I wonder if the amount of heat could be in part as a result of the GPU always running at its maximum...
could this be the case?
Yup. In most cases fans are self-adjusting to heat. If you downclock GPU, it will produce less heat and so fan will slow down. To downclock you need working radeon (DDX) or KMS. Both can downclock GPU by AtomBIOS commands, but I'm not sure how stable support for R5xxx is. -- Rafał -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Rafal,
Yup. In most cases fans are self-adjusting to heat. If you downclock GPU, it will produce less heat and so fan will slow down.
To downclock you need working radeon (DDX) or KMS. Both can downclock GPU by AtomBIOS commands, but I'm not sure how stable support for R5xxx is.
as the radeon driver seems to no support the card at all - where can I read more about this magical KMS thing? how would I issue these AtomBIOS commands? Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
W dniu 18 lutego 2010 16:18 użytkownik Ákos Maróy
Rafal,
Yup. In most cases fans are self-adjusting to heat. If you downclock GPU, it will produce less heat and so fan will slow down.
To downclock you need working radeon (DDX) or KMS. Both can downclock GPU by AtomBIOS commands, but I'm not sure how stable support for R5xxx is.
as the radeon driver seems to no support the card at all - where can I read more about this magical KMS thing? how would I issue these AtomBIOS commands?
Support for evergreen (R5xxx) is almost the same in KMS and radeon (xf86-video-ati). I think radeon has some problem when going back to console, that's all difference. In both cases you have to use software from git. In case of radeon it's standard master branch of git, in case of KMS drm-radeom-testing branch will be needed from airled's repository. In case of radeon there is "ForceLowPowerMode" option for xorg.conf In case of drm-radeon-testinth there is "dynpm" parameter. -- Rafał -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
Rafal,
Yup. In most cases fans are self-adjusting to heat. If you downclock GPU, it will produce less heat and so fan will slow down.
To downclock you need working radeon (DDX) or KMS. Both can downclock GPU by AtomBIOS commands, but I'm not sure how stable support for R5xxx is.
as the radeon driver seems to no support the card at all - where can I read more about this magical KMS thing? how would I issue these AtomBIOS commands?
You need git master for evergreen (HD5xxx) UMS support. For KMS, you need the drm-radeon-testing branch of Dave's drm tree. http://cgit.freedesktop.org/xorg/driver/xf86-video-ati http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=shortlog;h=r... Atombios command tables are found in the vbios rom on the card. The table parser is available in the ddx and drm trees. Alex
Akos
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex,
You need git master for evergreen (HD5xxx) UMS support. For KMS, you need the drm-radeon-testing branch of Dave's drm tree. http://cgit.freedesktop.org/xorg/driver/xf86-video-ati http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=shortlog;h=r...
thanks for the pointers. the kernel compile fails with the following: make[1]: Entering directory `/root/src/drm-2.6' CHK include/linux/version.h CHK include/linux/utsrelease.h SYMLINK include/asm -> include/asm-x86 CALL scripts/checksyscalls.sh Building modules, stage 2. MODPOST 2044 modules ERROR: "__set_irq_handler" [drivers/mfd/pcf50633-core.ko] undefined! ERROR: "handle_level_irq" [drivers/mfd/pcf50633-core.ko] undefined! WARNING: modpost: Found 1 section mismatch(es). To see full details build your kernel with: 'make CONFIG_DEBUG_SECTION_MISMATCH=y' I cloned the xf86-video-ati repository, compiled & installed the X drivers. now indeed, I seem to have initial ATI 5000 support, from Xorg.0.log: ATI Mobility Radeon HD 5800 Series, ATI Radeon HD 5700 Series, ATI Radeon HD 5700 Series, ATI Radeon HD 5700 Series, ATI Mobility Radeon HD 5000 Series, ATI Mobility Radeon HD 5000 Series, ATI FirePro (FireGL) Graphics Adapter, ATI FirePro (FireGL) Graphics Adapter, ATI Radeon HD 5670, ATI Radeon HD 5570, ATI Radeon HD 5500 Series, REDWOOD, ATI Mobility Radeon HD 5000 Series, ATI Mobility Radeon HD 5000 Series, CEDAR, CEDAR, CEDAR, ATI FirePro (FireGL) Graphics Adapter, ATI FirePro (FireGL) Graphics Adapter, CEDAR, ATI Radeon HD 5450, though, Mobility Radeon HD 5830 is not listed, which is the card I have. it seems to register ATOM BIOS support: (II) RADEON(0): Primary V_BIOS segment is: 0xc000 (II) RADEON(0): ATOM BIOS detected (II) RADEON(0): ATOM BIOS Rom: SubsystemVendorID: 0x103c SubsystemID: 0x7009 IOBaseAddress: 0x4000 Filename: BR35952.001 BIOS Bootup Message: BROADWAY LP HP/Quanta Nikita with DDR3 64Mx16 (II) RADEON(0): Framebuffer space used by Firmware (kb): 20 (II) RADEON(0): Start of VRAM area used by Firmware: 0xfffec (II) RADEON(0): AtomBIOS requests 20kB of VRAM scratch space (II) RADEON(0): AtomBIOS VRAM scratch base: 0xfffec I also get an error later on: (WW) RADEON(0): Direct rendering disabled (EE) RADEON(0): Acceleration initialization failed (II) RADEON(0): Acceleration disabled
Atombios command tables are found in the vbios rom on the card. The table parser is available in the ddx and drm trees.
I'm not sure I know from this what to do next. the system idles at about the same temperature now as well... is there a way to tell the actual clock speed of the GPU? is there a way to change the actual clock speed? Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
Alex,
You need git master for evergreen (HD5xxx) UMS support. For KMS, you need the drm-radeon-testing branch of Dave's drm tree. http://cgit.freedesktop.org/xorg/driver/xf86-video-ati http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=shortlog;h=r...
thanks for the pointers.
the kernel compile fails with the following:
make[1]: Entering directory `/root/src/drm-2.6' CHK include/linux/version.h CHK include/linux/utsrelease.h SYMLINK include/asm -> include/asm-x86 CALL scripts/checksyscalls.sh Building modules, stage 2. MODPOST 2044 modules ERROR: "__set_irq_handler" [drivers/mfd/pcf50633-core.ko] undefined! ERROR: "handle_level_irq" [drivers/mfd/pcf50633-core.ko] undefined! WARNING: modpost: Found 1 section mismatch(es). To see full details build your kernel with: 'make CONFIG_DEBUG_SECTION_MISMATCH=y'
I cloned the xf86-video-ati repository, compiled & installed the X drivers. now indeed, I seem to have initial ATI 5000 support, from Xorg.0.log:
ATI Mobility Radeon HD 5800 Series, ATI Radeon HD 5700 Series, ATI Radeon HD 5700 Series, ATI Radeon HD 5700 Series, ATI Mobility Radeon HD 5000 Series, ATI Mobility Radeon HD 5000 Series, ATI FirePro (FireGL) Graphics Adapter, ATI FirePro (FireGL) Graphics Adapter, ATI Radeon HD 5670, ATI Radeon HD 5570, ATI Radeon HD 5500 Series, REDWOOD, ATI Mobility Radeon HD 5000 Series, ATI Mobility Radeon HD 5000 Series, CEDAR, CEDAR, CEDAR, ATI FirePro (FireGL) Graphics Adapter, ATI FirePro (FireGL) Graphics Adapter, CEDAR, ATI Radeon HD 5450,
though, Mobility Radeon HD 5830 is not listed, which is the card I have.
it seems to register ATOM BIOS support:
(II) RADEON(0): Primary V_BIOS segment is: 0xc000 (II) RADEON(0): ATOM BIOS detected (II) RADEON(0): ATOM BIOS Rom: SubsystemVendorID: 0x103c SubsystemID: 0x7009 IOBaseAddress: 0x4000 Filename: BR35952.001 BIOS Bootup Message: BROADWAY LP HP/Quanta Nikita with DDR3 64Mx16
(II) RADEON(0): Framebuffer space used by Firmware (kb): 20 (II) RADEON(0): Start of VRAM area used by Firmware: 0xfffec (II) RADEON(0): AtomBIOS requests 20kB of VRAM scratch space (II) RADEON(0): AtomBIOS VRAM scratch base: 0xfffec
I also get an error later on:
(WW) RADEON(0): Direct rendering disabled (EE) RADEON(0): Acceleration initialization failed (II) RADEON(0): Acceleration disabled
We do not support acceleration yet. Ignoring that, does X work fine under radeon control? Did it start correctly, with proper resolution, etc.?
Atombios command tables are found in the vbios rom on the card. The table parser is available in the ddx and drm trees.
I'm not sure I know from this what to do next. the system idles at about the same temperature now as well...
is there a way to tell the actual clock speed of the GPU? is there a way to change the actual clock speed?
As I said, xorg.conf and ForceLowPowerMode option in Driver section. -- Rafał -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Rafal,
We do not support acceleration yet. Ignoring that, does X work fine under radeon control? Did it start correctly, with proper resolution, etc.?
yes, it did. the only problem being that it won't shift back to text mode properly - but this is a know issue.
As I said, xorg.conf and ForceLowPowerMode option in Driver section.
I added it, this is what I get in Xorg.0.log: (**) RADEON(0): Option "ForceLowPowerMode" "true" ... (II) RADEON(0): Dynamic Power Management Disabled (II) RADEON(0): Force Low Power Mode Enabled (II) RADEON(0): Power Mode Switch but the system temperature is the same. as a comparison, the log file without this option only contained this line from the above: (II) RADEON(0): Dynamic Power Management Disabled Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Rafal, I also tried to run rovclock to check on the GPU clock speeds, but the results are invalid: $ sudo rovclock -i Radeon overclock 0.6e by Hasw (hasw@hasw.net) Found ATI card on 01:00, device id: 0x68a1 I/O base address: 0x4000 Video BIOS shadow found @ 0xc0000 Invalid reference clock from BIOS: 0.0 MHz Memory size: 0 kB Memory channels: 1, CD,CH only: 0 tRcdRD: 3 tRcdWR: 5 tRP: 7 tRAS: 8 tRRD: 3 tR2W-CL: 2 tWR: 5 tW2R: 0 tW2Rsb: 0 tR2R: 1 tRFC: 13 tWL(0.5): 0 tCAS: 0 tCMD: 0 tSTR: 0 XTAL: 27.0 MHz, RefDiv: 192 Core: 27.0 MHz, Mem: 27.0 MHz is there any way to determine the real GPU clock speed? Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
I think the selected speed appears in the xorg log. -----Original Message----- From: Ákos Maróy [mailto:akos@maroy.hu] Sent: Thursday, February 18, 2010 3:39 PM To: Rafał Miłecki Cc: Alex Deucher; radeonhd@opensuse.org Subject: Re: [radeonhd] GPU throttling? Rafal, I also tried to run rovclock to check on the GPU clock speeds, but the results are invalid: $ sudo rovclock -i Radeon overclock 0.6e by Hasw (hasw@hasw.net) Found ATI card on 01:00, device id: 0x68a1 I/O base address: 0x4000 Video BIOS shadow found @ 0xc0000 Invalid reference clock from BIOS: 0.0 MHz Memory size: 0 kB Memory channels: 1, CD,CH only: 0 tRcdRD: 3 tRcdWR: 5 tRP: 7 tRAS: 8 tRRD: 3 tR2W-CL: 2 tWR: 5 tW2R: 0 tW2Rsb: 0 tR2R: 1 tRFC: 13 tWL(0.5): 0 tCAS: 0 tCMD: 0 tSTR: 0 XTAL: 27.0 MHz, RefDiv: 192 Core: 27.0 MHz, Mem: 27.0 MHz is there any way to determine the real GPU clock speed? Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Bridgman, John wrote:
I think the selected speed appears in the xorg log.
I see.. this is what I get: $ grep Clock /var/log/Xorg.0.log (II) RADEON(0): Default Engine Clock: 500000 (II) RADEON(0): Default Memory Clock: 800000 (II) RADEON(0): Maximum Pixel ClockPLL Frequency Output: 1200000 (II) RADEON(0): Minimum Pixel ClockPLL Frequency Output: 0 (II) RADEON(0): Maximum Pixel ClockPLL Frequency Input: 27000 (II) RADEON(0): Minimum Pixel ClockPLL Frequency Input: 6750 (II) RADEON(0): Maximum Pixel Clock: 400000 (II) RADEON(0): Reference Clock: 27000 XRes: 1920, YRes: 1080, DotClock: 138500 (II) RADEON(0): crtc(0) Clock: mode 138500, PLL 138500 Alex Deucher wrote:
The default clock is listed in the xorg log. ForceLowPowerMode reduces the clock to half that.
is there a way to verify? there seems to be no difference in terms of experience, i.e. the system temperature :( -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
Bridgman, John wrote:
I think the selected speed appears in the xorg log.
I see.. this is what I get:
$ grep Clock /var/log/Xorg.0.log (II) RADEON(0): Default Engine Clock: 500000
^^^^^^^^^^ engine clock is 500 Mhz.
(II) RADEON(0): Default Memory Clock: 800000 (II) RADEON(0): Maximum Pixel ClockPLL Frequency Output: 1200000 (II) RADEON(0): Minimum Pixel ClockPLL Frequency Output: 0 (II) RADEON(0): Maximum Pixel ClockPLL Frequency Input: 27000 (II) RADEON(0): Minimum Pixel ClockPLL Frequency Input: 6750 (II) RADEON(0): Maximum Pixel Clock: 400000 (II) RADEON(0): Reference Clock: 27000 XRes: 1920, YRes: 1080, DotClock: 138500 (II) RADEON(0): crtc(0) Clock: mode 138500, PLL 138500
Alex Deucher wrote:
The default clock is listed in the xorg log. ForceLowPowerMode reduces the clock to half that.
is there a way to verify? there seems to be no difference in terms of experience, i.e. the system temperature :(
It's doing it. How are you measuring temperature? GPU temperature support is not implemented yet in the open source driver. If it's a general case temperature, it may have more to do with the other stuff inside than the GPU. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex Deucher wrote:
2010/2/18 Ákos Maróy
: Bridgman, John wrote:
I think the selected speed appears in the xorg log. I see.. this is what I get:
$ grep Clock /var/log/Xorg.0.log (II) RADEON(0): Default Engine Clock: 500000
^^^^^^^^^^
engine clock is 500 Mhz.
but I get the same reading without the ForceLowPowerMode option.. indeed, the clock of this card is 500 MHz: http://www.amd.com/us/products/notebook/graphics/ati-mobility-hd-5800/Pages/... so you're saying that without any explicit indication, the ForeLowPowerMode option will make it run at half the speed..
It's doing it. How are you measuring temperature? GPU temperature support is not implemented yet in the open source driver. If it's a general case temperature, it may have more to do with the other stuff inside than the GPU.
there's a single ACPI temperature reading within the system, this is the only thing I can count on now. all else is subjective, like fan noise, subjective heat on the keyboard, etc. the machine is irrationally hot, and I'm looking for ways to make it cooler. the CPU seems to be handled already, verified that in a number of ways. my next possible choice was the GPU as an obvious source of heat... -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
Alex Deucher wrote:
2010/2/18 Ákos Maróy
: Bridgman, John wrote:
I think the selected speed appears in the xorg log. I see.. this is what I get:
$ grep Clock /var/log/Xorg.0.log (II) RADEON(0): Default Engine Clock: 500000
^^^^^^^^^^
engine clock is 500 Mhz.
but I get the same reading without the ForceLowPowerMode option..
indeed, the clock of this card is 500 MHz:
http://www.amd.com/us/products/notebook/graphics/ati-mobility-hd-5800/Pages/...
so you're saying that without any explicit indication, the ForeLowPowerMode option will make it run at half the speed..
It would be good idea to implement reading current engine clock in radeon. There is AtomBIOS command for that. -- Rafał -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Rafał Miłecki
2010/2/18 Ákos Maróy
: Alex Deucher wrote:
2010/2/18 Ákos Maróy
: Bridgman, John wrote:
I think the selected speed appears in the xorg log. I see.. this is what I get:
$ grep Clock /var/log/Xorg.0.log (II) RADEON(0): Default Engine Clock: 500000
^^^^^^^^^^
engine clock is 500 Mhz.
but I get the same reading without the ForceLowPowerMode option..
indeed, the clock of this card is 500 MHz:
http://www.amd.com/us/products/notebook/graphics/ati-mobility-hd-5800/Pages/...
so you're saying that without any explicit indication, the ForeLowPowerMode option will make it run at half the speed..
It would be good idea to implement reading current engine clock in radeon. There is AtomBIOS command for that.
There's no interface for reading it. The current ddx code only sets the default clock and default clock divided by 2 so you always know what the clock is. In the case of the power management options we don't print the clock to avoid spamming the log with tons of clock changes. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Alex Deucher
2010/2/18 Rafał Miłecki
: It would be good idea to implement reading current engine clock in radeon. There is AtomBIOS command for that.
There's no interface for reading it. The current ddx code only sets the default clock and default clock divided by 2 so you always know what the clock is. In the case of the power management options we don't print the clock to avoid spamming the log with tons of clock changes.
I wanted to make sure engine was really changed. Maybe it needs some additional magic for evergreen or sth? Just for debugging purposes. -- Rafał -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Thu, Feb 18, 2010 at 5:47 PM, Rafał Miłecki
2010/2/18 Alex Deucher
: 2010/2/18 Rafał Miłecki
: It would be good idea to implement reading current engine clock in radeon. There is AtomBIOS command for that.
There's no interface for reading it. The current ddx code only sets the default clock and default clock divided by 2 so you always know what the clock is. In the case of the power management options we don't print the clock to avoid spamming the log with tons of clock changes.
I wanted to make sure engine was really changed. Maybe it needs some additional magic for evergreen or sth? Just for debugging purposes.
Well, it's just reading back the same registers that the set function programmed, so it should be the same. Evergreen is no different. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Thu, Feb 18, 2010 at 4:05 PM, Ákos Maróy
Alex Deucher wrote:
2010/2/18 Ákos Maróy
: Bridgman, John wrote:
I think the selected speed appears in the xorg log. I see.. this is what I get:
$ grep Clock /var/log/Xorg.0.log (II) RADEON(0): Default Engine Clock: 500000
^^^^^^^^^^
engine clock is 500 Mhz.
but I get the same reading without the ForceLowPowerMode option..
indeed, the clock of this card is 500 MHz:
http://www.amd.com/us/products/notebook/graphics/ati-mobility-hd-5800/Pages/...
so you're saying that without any explicit indication, the ForeLowPowerMode option will make it run at half the speed..
yes. that messges just prints the default clock. ForceLowPowerMode sets the clock to half of the default value.
It's doing it. How are you measuring temperature? GPU temperature support is not implemented yet in the open source driver. If it's a general case temperature, it may have more to do with the other stuff inside than the GPU.
there's a single ACPI temperature reading within the system, this is the only thing I can count on now. all else is subjective, like fan noise, subjective heat on the keyboard, etc.
the machine is irrationally hot, and I'm looking for ways to make it cooler. the CPU seems to be handled already, verified that in a number of ways. my next possible choice was the GPU as an obvious source of heat...
Really depends where the thermal chip is and how well it's calibrated. If it's right under the north bridge or something, it well always run hot. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex Deucher wrote:
yes. that messges just prints the default clock. ForceLowPowerMode sets the clock to half of the default value.
but how does one know if this setting succeeds?
Really depends where the thermal chip is and how well it's calibrated. If it's right under the north bridge or something, it well always run hot.
yes, but the machine is quite hot to the touch as well, and the fan is always quite loud... it's also strange that there seems to be no difference in temperature between the forced low power mode and normal mode. on my old laptop with also an ATI card, the temperature difference was significant, depending on GPU load.
There's no interface for reading it. The current ddx code only sets the default clock and default clock divided by 2 so you always know what the clock is. In the case of the power management options we don't print the clock to avoid spamming the log with tons of clock changes.
well, a /proc like interface to read it on demand would prevent spamming, but provide the information as needed. or some signal to dump the current state into the log file. or some dedicated program that just shows this information on stdout. Rafał Miłecki wrote:
I wanted to make sure engine was really changed. Maybe it needs some additional magic for evergreen or sth? Just for debugging purposes.
yes, something like this would be nice.. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Thu, Feb 18, 2010 at 5:59 PM, Ákos Maróy
Alex Deucher wrote:
yes. that messges just prints the default clock. ForceLowPowerMode sets the clock to half of the default value.
but how does one know if this setting succeeds?
The GPU didn't lock up. You can call the get engine clock table if you want to verify.
Really depends where the thermal chip is and how well it's calibrated. If it's right under the north bridge or something, it well always run hot.
yes, but the machine is quite hot to the touch as well, and the fan is always quite loud... it's also strange that there seems to be no difference in temperature between the forced low power mode and normal mode. on my old laptop with also an ATI card, the temperature difference was significant, depending on GPU load.
There's no interface for reading it. The current ddx code only sets the default clock and default clock divided by 2 so you always know what the clock is. In the case of the power management options we don't print the clock to avoid spamming the log with tons of clock changes.
well, a /proc like interface to read it on demand would prevent spamming, but provide the information as needed. or some signal to dump the current state into the log file. or some dedicated program that just shows this information on stdout.
You need a kernel driver to use /proc. the ddx is in userspace.
Rafał Miłecki wrote:
I wanted to make sure engine was really changed. Maybe it needs some additional magic for evergreen or sth? Just for debugging purposes.
yes, something like this would be nice..
kms already has this. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Thu, Feb 18, 2010 at 6:28 PM, Ákos Maróy
Alex Deucher wrote:
The GPU didn't lock up. You can call the get engine clock table if you want to verify.
how would I do that?
You'd need to patch the code. You can grab the get_engine_clock code from kms in radeon_atombios.c. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex,
You'd need to patch the code. You can grab the get_engine_clock code from kms in radeon_atombios.c.
I'm a bit confused, I'm not finding a gen_engine_clock call in radeon_atombios.c in the xf86-video-ati source tree, nor in the drm-2.6 kernel source code base you sent a link about. I'll look into it later, and try to come up with something... Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/19 Ákos Maróy
Alex,
You'd need to patch the code. You can grab the get_engine_clock code from kms in radeon_atombios.c.
I'm a bit confused, I'm not finding a gen_engine_clock call in radeon_atombios.c in the xf86-video-ati source tree, nor in the drm-2.6 kernel source code base you sent a link about.
I'll look into it later, and try to come up with something...
There is not ready command for that, you have to write it. AtomBIOS table is defined for that (GetEngineClock and it's parameters). -- Rafał -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Rafał Miłecki wrote:
There is not ready command for that, you have to write it. AtomBIOS table is defined for that (GetEngineClock and it's parameters).
indeed. I've implemented two functions to get the engine and the memory clocks, see below. the output after the low power mode setting is: (II) RADEON(0): Force Low Power Mode Enabled (II) RADEON(0): Power Mode Switch (II) RADEON(0): Engine Clock Set To 249980 (II) RADEON(0): Memory Clock Set To 799870 thus, the engine clock is indeed set to 250MHz, the memory clock is 800MHz. it seems the engine clock is half of the default one, as Alex also said, while the memory clock is unchanged at 800MHz: (II) RADEON(0): Default Engine Clock: 500000 (II) RADEON(0): Default Memory Clock: 800000 I wonder if the clock can be set lower, or if it makes sense to change the memory clock? here's the diff of what I did - the 10x multiplier I put there after I found this sample for radeonhd: http://lists.opensuse.org/radeonhd/2009-04/msg00163.html actually the possibility in this patch to set the desired clock looks quite good. I wonder what are the acceptable clock speeds... (BTW, I can't use radeonhd, as it says it cannot find a display device) diff --git a/src/radeon_atombios.c b/src/radeon_atombios.c index 0fa53ff..ab719aa 100644 --- a/src/radeon_atombios.c +++ b/src/radeon_atombios.c @@ -578,6 +578,42 @@ atombios_static_pwrmgt_setup(ScrnInfoPtr pScrn, Bool enable } +uint32_t +atombios_get_memory_clock(ScrnInfoPtr pScrn) +{ + RADEONInfoPtr info = RADEONPTR(pScrn); + GET_MEMORY_CLOCK_PARAMETERS mem_p; + AtomBiosArgRec data; + + RADEONWaitForIdleMMIO(pScrn); + + data.exec.index = GetIndexIntoMasterTable(COMMAND, GetMemoryClock); + data.exec.pspace = &mem_p; + + if (RHDAtomBiosFunc(info->atomBIOS->scrnIndex, info->atomBIOS, ATOMBIOS_EXE + return mem_p.ulReturnMemoryClock * 10ul;; + } + return 0; +} + +uint32_t +atombios_get_engine_clock(ScrnInfoPtr pScrn) +{ + RADEONInfoPtr info = RADEONPTR(pScrn); + GET_ENGINE_CLOCK_PARAMETERS eng_p; + AtomBiosArgRec data; + + RADEONWaitForIdleMMIO(pScrn); + + data.exec.index = GetIndexIntoMasterTable(COMMAND, GetEngineClock); + data.exec.pspace = &eng_p; + + if (RHDAtomBiosFunc(info->atomBIOS->scrnIndex, info->atomBIOS, ATOMBIOS_EXE + return eng_p.ulReturnEngineClock * 10ul; + } + return 0; +} + int atombios_set_engine_clock(ScrnInfoPtr pScrn, uint32_t engclock) { diff --git a/src/radeon_atombios.h b/src/radeon_atombios.h index 1f21c46..ebcd6ed 100644 --- a/src/radeon_atombios.h +++ b/src/radeon_atombios.h @@ -122,6 +122,12 @@ atombios_clk_gating_setup(ScrnInfoPtr pScrn, Bool enable); extern int atombios_static_pwrmgt_setup(ScrnInfoPtr pScrn, Bool enable); +extern uint32_t +atombios_get_memory_clock(ScrnInfoPtr pScrn); + +extern uint32_t +atombios_get_engine_clock(ScrnInfoPtr pScrn); + extern int atombios_set_engine_clock(ScrnInfoPtr pScrn, uint32_t engclock); diff --git a/src/radeon_pm.c b/src/radeon_pm.c index d5152c8..3b45418 100644 --- a/src/radeon_pm.c +++ b/src/radeon_pm.c @@ -754,6 +754,7 @@ static void RADEONSetStaticPowerMode(ScrnInfoPtr pScrn, RADE { RADEONInfoPtr info = RADEONPTR(pScrn); int i; + int clock; for (i = 0; i < info->pm.num_modes; i++) { if (info->pm.mode[i].type == type) @@ -779,6 +780,12 @@ static void RADEONSetStaticPowerMode(ScrnInfoPtr pScrn, RAD info->pm.current_mode = i; xf86DrvMsg(pScrn->scrnIndex, X_INFO, "Power Mode Switch\n"); + + clock = atombios_get_engine_clock(pScrn); + xf86DrvMsg(pScrn->scrnIndex, X_INFO, "Engine Clock Set To %d\n", clock); + + clock = atombios_get_memory_clock(pScrn); + xf86DrvMsg(pScrn->scrnIndex, X_INFO, "Memory Clock Set To %d\n", clock); } -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/19 Ákos Maróy
Rafał Miłecki wrote:
There is not ready command for that, you have to write it. AtomBIOS table is defined for that (GetEngineClock and it's parameters).
indeed.
I've implemented two functions to get the engine and the memory clocks, see below. the output after the low power mode setting is:
(II) RADEON(0): Force Low Power Mode Enabled (II) RADEON(0): Power Mode Switch (II) RADEON(0): Engine Clock Set To 249980 (II) RADEON(0): Memory Clock Set To 799870
thus, the engine clock is indeed set to 250MHz, the memory clock is 800MHz. it seems the engine clock is half of the default one, as Alex also said, while the memory clock is unchanged at 800MHz:
(II) RADEON(0): Default Engine Clock: 500000 (II) RADEON(0): Default Memory Clock: 800000
I wonder if the clock can be set lower, or if it makes sense to change the memory clock?
It does make sense to change it, but the memory clock is trickier to deal with than the engine clock because you need make sure you don't lower it too much for your current bandwidth requirements (3D engine, displays, etc.) or you'll get underflow and display or rendering problems.
here's the diff of what I did - the 10x multiplier I put there after I found this sample for radeonhd: http://lists.opensuse.org/radeonhd/2009-04/msg00163.html actually the possibility in this patch to set the desired clock looks quite good. I wonder what are the acceptable clock speeds...
Atom tables use 10 khz units. There are power tables in the bios that have sets of known good clock combinations. The ddx doesn't currently use them, however, kms does. If you really want to play with power management KMS is your best bet. The ddx options are more of a stop gap until the kms code is more mature.
(BTW, I can't use radeonhd, as it says it cannot find a display device)
rhd doesn't have support for evergreen cards at the moment. Alex
diff --git a/src/radeon_atombios.c b/src/radeon_atombios.c index 0fa53ff..ab719aa 100644 --- a/src/radeon_atombios.c +++ b/src/radeon_atombios.c @@ -578,6 +578,42 @@ atombios_static_pwrmgt_setup(ScrnInfoPtr pScrn, Bool enable
}
+uint32_t +atombios_get_memory_clock(ScrnInfoPtr pScrn) +{ + RADEONInfoPtr info = RADEONPTR(pScrn); + GET_MEMORY_CLOCK_PARAMETERS mem_p; + AtomBiosArgRec data; + + RADEONWaitForIdleMMIO(pScrn); + + data.exec.index = GetIndexIntoMasterTable(COMMAND, GetMemoryClock); + data.exec.pspace = &mem_p; + + if (RHDAtomBiosFunc(info->atomBIOS->scrnIndex, info->atomBIOS, ATOMBIOS_EXE + return mem_p.ulReturnMemoryClock * 10ul;; + } + return 0; +} + +uint32_t +atombios_get_engine_clock(ScrnInfoPtr pScrn) +{ + RADEONInfoPtr info = RADEONPTR(pScrn); + GET_ENGINE_CLOCK_PARAMETERS eng_p; + AtomBiosArgRec data; + + RADEONWaitForIdleMMIO(pScrn); + + data.exec.index = GetIndexIntoMasterTable(COMMAND, GetEngineClock); + data.exec.pspace = &eng_p; + + if (RHDAtomBiosFunc(info->atomBIOS->scrnIndex, info->atomBIOS, ATOMBIOS_EXE + return eng_p.ulReturnEngineClock * 10ul; + } + return 0; +} + int atombios_set_engine_clock(ScrnInfoPtr pScrn, uint32_t engclock) { diff --git a/src/radeon_atombios.h b/src/radeon_atombios.h index 1f21c46..ebcd6ed 100644 --- a/src/radeon_atombios.h +++ b/src/radeon_atombios.h @@ -122,6 +122,12 @@ atombios_clk_gating_setup(ScrnInfoPtr pScrn, Bool enable); extern int atombios_static_pwrmgt_setup(ScrnInfoPtr pScrn, Bool enable);
+extern uint32_t +atombios_get_memory_clock(ScrnInfoPtr pScrn); + +extern uint32_t +atombios_get_engine_clock(ScrnInfoPtr pScrn); + extern int atombios_set_engine_clock(ScrnInfoPtr pScrn, uint32_t engclock);
diff --git a/src/radeon_pm.c b/src/radeon_pm.c index d5152c8..3b45418 100644 --- a/src/radeon_pm.c +++ b/src/radeon_pm.c @@ -754,6 +754,7 @@ static void RADEONSetStaticPowerMode(ScrnInfoPtr pScrn, RADE { RADEONInfoPtr info = RADEONPTR(pScrn); int i; + int clock;
for (i = 0; i < info->pm.num_modes; i++) { if (info->pm.mode[i].type == type) @@ -779,6 +780,12 @@ static void RADEONSetStaticPowerMode(ScrnInfoPtr pScrn, RAD info->pm.current_mode = i;
xf86DrvMsg(pScrn->scrnIndex, X_INFO, "Power Mode Switch\n"); + + clock = atombios_get_engine_clock(pScrn); + xf86DrvMsg(pScrn->scrnIndex, X_INFO, "Engine Clock Set To %d\n", clock); + + clock = atombios_get_memory_clock(pScrn); + xf86DrvMsg(pScrn->scrnIndex, X_INFO, "Memory Clock Set To %d\n", clock); }
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex,
It does make sense to change it, but the memory clock is trickier to deal with than the engine clock because you need make sure you don't lower it too much for your current bandwidth requirements (3D engine, displays, etc.) or you'll get underflow and display or rendering problems.
I see.. I already encountered flickering when chaning the engine clock to 1/4th of the original value.
Atom tables use 10 khz units. There are power tables in the bios that have sets of known good clock combinations. The ddx doesn't currently use them, however, kms does. If you really want to play with power management KMS is your best bet. The ddx options are more of a stop gap until the kms code is more mature.
this KMS you're mentioning, this would be the drm-2.6 kernel tree you sent a git URL about, and which failed for me when compiling. did I see correct that this is a kernel branch made sometime during 2.6.2x? Unfortunately for other hardware I need a 2.6.32+ kernel tree...
rhd doesn't have support for evergreen cards at the moment.
thanks for the info. is there an ETA on evergreen support in any of the radeon drivers - radeon, radeonhd and ATI binary official? is this like weeks, months, years? Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
this KMS you're mentioning, this would be the drm-2.6 kernel tree you sent a git URL about, and which failed for me when compiling. did I see correct that this is a kernel branch made sometime during 2.6.2x? Unfortunately for other hardware I need a 2.6.32+ kernel tree...
The drm-2.6 git tree is based on 2.6.29-rc2, according to the Makefile in the top level of the kernel source tree. There is no Evergreen support there. For Evergreen support, you need the branch of drm-2.6 called "drm-radeon-testing"; this is based on kernel 2.6.32, which is the current stable version of the Linux kernel. (Not for long, though, since 2.6.33 will be released quite soon.)
rhd doesn't have support for evergreen cards at the moment.
thanks for the info.
is there an ETA on evergreen support in any of the radeon drivers - radeon, radeonhd and ATI binary official? is this like weeks, months, years?
The earliest beginnings of Evergreen support have appeared in the "radeon" driver (xf86-video-ati). I believe that the proprietary Catalyst/fglrx has support, but I haven't actually looked into it. (I'm only interested in the open source support.) The "radeonhd" driver (xf86-video-radeonhd) is barely being touched these days, and maintainers for many distributions are describing it as dead and recommending their users not to even touch it. The developers of "radeonhd" have not announced that it is dead, though, and every once in a while some commits are still made to the "radeonhd" git tree, so I don't know what to think about it. HTH, Dave W. PS: I'm not a developer; nothing that I say should be taken as authoritative. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Fri, Feb 19, 2010 at 1:29 PM, Ákos Maróy
Alex,
It does make sense to change it, but the memory clock is trickier to deal with than the engine clock because you need make sure you don't lower it too much for your current bandwidth requirements (3D engine, displays, etc.) or you'll get underflow and display or rendering problems.
I see.. I already encountered flickering when chaning the engine clock to 1/4th of the original value.
The ForceLowPowerMode option also lowers the number of PCIE lanes, so you may be seeing the result of reduced bus bandwidth.
Atom tables use 10 khz units. There are power tables in the bios that have sets of known good clock combinations. The ddx doesn't currently use them, however, kms does. If you really want to play with power management KMS is your best bet. The ddx options are more of a stop gap until the kms code is more mature.
this KMS you're mentioning, this would be the drm-2.6 kernel tree you sent a git URL about, and which failed for me when compiling. did I see correct that this is a kernel branch made sometime during 2.6.2x? Unfortunately for other hardware I need a 2.6.32+ kernel tree...
You need to checkout the drm-radeon-testing branch.
rhd doesn't have support for evergreen cards at the moment.
thanks for the info.
is there an ETA on evergreen support in any of the radeon drivers - radeon, radeonhd and ATI binary official? is this like weeks, months, years?
It's already supported in radeon and KMS (modesetting only), and fully supported with accel in the binary driver. Work on open acceleration code and documentation has already started in house. It should be available in the next few months. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex,
You need to checkout the drm-radeon-testing branch.
and where is the drm-radeon-testing branch exactly?
It's already supported in radeon and KMS (modesetting only), and fully supported with accel in the binary driver. Work on open acceleration
I see. when I look at the 'download drivers' section on the right hand side at http://www.amd.com/uk/Pages/AMDHomePage.aspx , it only lists Mobility Radeon HD up to the 4000 Series - while I have a 5830 card... is this supported in some beta / testing binary driver?
code and documentation has already started in house. It should be available in the next few months.
can't wait to see support for it.. Akos -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Fri, Feb 19, 2010 at 2:14 PM, Ákos Maróy
Alex,
You need to checkout the drm-radeon-testing branch.
and where is the drm-radeon-testing branch exactly?
In the drm tree link I sent you previously: http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=shortlog;h=r...
It's already supported in radeon and KMS (modesetting only), and fully supported with accel in the binary driver. Work on open acceleration
I see. when I look at the 'download drivers' section on the right hand side at http://www.amd.com/uk/Pages/AMDHomePage.aspx , it only lists Mobility Radeon HD up to the 4000 Series - while I have a 5830 card... is this supported in some beta / testing binary driver?
All the current models are listed here: http://www.amd.com/us/Pages/AMDHomePage.aspx Alex
code and documentation has already started in house. It should be available in the next few months.
can't wait to see support for it..
Akos
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Alex Deucher wrote:
All the current models are listed here: http://www.amd.com/us/Pages/AMDHomePage.aspx
same thing, mobility radeon is only listed up to series 4000, while I have a 5830, which would be series 5000 I guess. also, the downloaded 10.2 driver fails both when installing, and when trying to create an ubuntu package for it. as this is the radeonhd list, I guess this is not the best place to post these issues here. is there a mailing list / forum for issues with the binary ATI driver? -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Bug tracker for the proprietary driver is at http://ati.cchtml.com Best forum is probably the AMD/ATI Linux area at Phoronix : http://www.phoronix.com/forums/forumdisplay.php?f=19 There were problems with the package-builder scripts for Ubuntu in 10.1 but I haven't heard of issues with the scripts in 10.2. There is a reported issue with compositing on 10.2 which didn't show up in our testing, so not sure what the system/version/platform specificity of that one is yet. -----Original Message----- From: Ákos Maróy [mailto:akos@maroy.hu] Sent: Friday, February 19, 2010 2:39 PM To: Alex Deucher Cc: Rafa? Mi?ecki; Bridgman, John; radeonhd@opensuse.org Subject: Re: [radeonhd] GPU throttling? Alex Deucher wrote:
All the current models are listed here: http://www.amd.com/us/Pages/AMDHomePage.aspx
same thing, mobility radeon is only listed up to series 4000, while I have a 5830, which would be series 5000 I guess. also, the downloaded 10.2 driver fails both when installing, and when trying to create an ubuntu package for it. as this is the radeonhd list, I guess this is not the best place to post these issues here. is there a mailing list / forum for issues with the binary ATI driver? -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
And no, this is not the best place to discuss Catalyst Linux issues ;) -----Original Message----- From: Bridgman, John [mailto:John.Bridgman@amd.com] Sent: Friday, February 19, 2010 2:43 PM To: Ákos Maróy; Alex Deucher Cc: Rafa? Mi?ecki; radeonhd@opensuse.org Subject: RE: [radeonhd] GPU throttling? Bug tracker for the proprietary driver is at http://ati.cchtml.com Best forum is probably the AMD/ATI Linux area at Phoronix : http://www.phoronix.com/forums/forumdisplay.php?f=19 There were problems with the package-builder scripts for Ubuntu in 10.1 but I haven't heard of issues with the scripts in 10.2. There is a reported issue with compositing on 10.2 which didn't show up in our testing, so not sure what the system/version/platform specificity of that one is yet. -----Original Message----- From: Ákos Maróy [mailto:akos@maroy.hu] Sent: Friday, February 19, 2010 2:39 PM To: Alex Deucher Cc: Rafa? Mi?ecki; Bridgman, John; radeonhd@opensuse.org Subject: Re: [radeonhd] GPU throttling? Alex Deucher wrote:
All the current models are listed here: http://www.amd.com/us/Pages/AMDHomePage.aspx
same thing, mobility radeon is only listed up to series 4000, while I have a 5830, which would be series 5000 I guess. also, the downloaded 10.2 driver fails both when installing, and when trying to create an ubuntu package for it. as this is the radeonhd list, I guess this is not the best place to post these issues here. is there a mailing list / forum for issues with the binary ATI driver? -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
Bridgman, John wrote:
Bug tracker for the proprietary driver is at http://ati.cchtml.com
Best forum is probably the AMD/ATI Linux area at Phoronix :
thanks...
And no, this is not the best place to discuss Catalyst Linux issues ;)
thought so :) -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
Rafal,
I also tried to run rovclock to check on the GPU clock speeds, but the results are invalid:
Don't use rovclock, it's only applicable to r1xx-r4xx radeons (so it won't work on your card), and even then it doesn't take into account the post divider properly. Alex
$ sudo rovclock -i Radeon overclock 0.6e by Hasw (hasw@hasw.net)
Found ATI card on 01:00, device id: 0x68a1 I/O base address: 0x4000 Video BIOS shadow found @ 0xc0000 Invalid reference clock from BIOS: 0.0 MHz Memory size: 0 kB Memory channels: 1, CD,CH only: 0 tRcdRD: 3 tRcdWR: 5 tRP: 7 tRAS: 8 tRRD: 3 tR2W-CL: 2 tWR: 5 tW2R: 0 tW2Rsb: 0 tR2R: 1 tRFC: 13 tWL(0.5): 0 tCAS: 0 tCMD: 0 tSTR: 0 XTAL: 27.0 MHz, RefDiv: 192
Core: 27.0 MHz, Mem: 27.0 MHz
is there any way to determine the real GPU clock speed?
Akos
-- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
2010/2/18 Ákos Maróy
Rafal,
I also tried to run rovclock to check on the GPU clock speeds, but the results are invalid:
$ sudo rovclock -i Radeon overclock 0.6e by Hasw (hasw@hasw.net)
Found ATI card on 01:00, device id: 0x68a1 I/O base address: 0x4000 Video BIOS shadow found @ 0xc0000 Invalid reference clock from BIOS: 0.0 MHz Memory size: 0 kB Memory channels: 1, CD,CH only: 0 tRcdRD: 3 tRcdWR: 5 tRP: 7 tRAS: 8 tRRD: 3 tR2W-CL: 2 tWR: 5 tW2R: 0 tW2Rsb: 0 tR2R: 1 tRFC: 13 tWL(0.5): 0 tCAS: 0 tCMD: 0 tSTR: 0 XTAL: 27.0 MHz, RefDiv: 192
Core: 27.0 MHz, Mem: 27.0 MHz
is there any way to determine the real GPU clock speed?
The default clock is listed in the xorg log. ForceLowPowerMode reduces the clock to half that. Alex -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
participants (5)
-
Alex Deucher
-
Bridgman, John
-
Dave Witbrodt
-
Rafał Miłecki
-
Ákos Maróy