Disabling power management firmware decisions
Hi,
I would like to do some experiments with the x86 platform but I am facing one big issue.
When I use the rapl register to measure the energy consumption with turbostat and another tool written by me. I see no differences between my kernel changes while using the idlestat tool, I see big differences in the idle state durations with the traces.
I remember on x86 the firmware overrides all kernel decisions to do the power management. So I assume, it is possible my changes are not measurable because of that.
Is there a way to tell the firmware to stop overriding the kernel decisions in order to let the kernel handle idle/dvfs on its own ?
Thanks
-- Daniel
Signature · 5 lines
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs Follow Linaro: <http://www.facebook.com/pages/Linaro> Facebook | <http://twitter.com/#!/linaroorg> Twitter | <http://www.linaro.org/linaro-blog/> Blog
Quoted 11 lines · Daniel Lezcano
Hi,
I would like to do some experiments with the x86 platform but I am facing one big issue.
When I use the rapl register to measure the energy consumption with turbostat and another tool written by me. I see no differences between my kernel changes while using the idlestat tool, I see big differences in the idle state durations with the traces.
Quoted 6 lines
I remember on x86 the firmware overrides all kernel decisions to do the power management. So I assume, it is possible my changes are not measurable because of that.
Is there a way to tell the firmware to stop overriding the kernel decisions in order to let the kernel handle idle/dvfs on its own ?
The short answer to that is no, there isn't. It also may not be firmware, but hardware.
Rafael
Quoted 16 lines · Rafael J. Wysocki
Quoted 15 lines · Daniel Lezcano
Hi,
I would like to do some experiments with the x86 platform but I am facing one big issue.
When I use the rapl register to measure the energy consumption with turbostat and another tool written by me. I see no differences between my kernel changes while using the idlestat tool, I see big differences in the idle state durations with the traces.
I remember on x86 the firmware overrides all kernel decisions to do the power management. So I assume, it is possible my changes are not measurable because of that.
Quoted 5 lines
Quoted 2 lines
Is there a way to tell the firmware to stop overriding the kernel decisions in order to let the kernel handle idle/dvfs on its own ?
The short answer to that is no, there isn't. It also may not be firmware, but hardware.
mmrfh, that's not funny :(
Thanks for the answer.
-- Daniel
Signature · 5 lines
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs Follow Linaro: <http://www.facebook.com/pages/Linaro> Facebook | <http://twitter.com/#!/linaroorg> Twitter | <http://www.linaro.org/linaro-blog/> Blog