Cpu Freq Merge with Scheduler

Cpu Freq Merge with Scheduler

[nick]

Greetings Fellow Developers,
I am curious about the work going into making the kernel scheduler more CPU power efficient. I have done some googling on this and
am curious about what how is going into the ideas/patches for this work.
Nick

Cpu Freq Merge with Scheduler

[Robert P. J. Day]

Quoting nick <xerofoify@gmail.com>:

> Greetings Fellow Developers,

> I am curious about the work going into making the kernel scheduler  
> more CPU power efficient. I have done some googling on this and
> am curious about what how is going into the ideas/patches for this work.

> Nick

   OFFS ... why are you asking about a clearly advanced kernel feature
on the kernel *newbies* list? And by the way, Nick, this is another
of your less endearing features -- your habit of posting one or two
line posts, effectively saying, "I am interested in topic <X>, could
everyone drop what they're doing and explain it to me in astonishing
detail to save me the trouble of doing any work?"

   You've done this before -- the networking stack, BTRFS, and probably
more. Rather than do enough research to ask *detailed* and *specific*
questions, you instead simply request that people here explain something
to you, and it's getting kind of tiring.

   Do your own research. Stop asking everyone else to do it for you.

rday

Cpu Freq Merge with Scheduler

[nick]

Robert,
After doing some research I found on the lkml logs and through google that the latest discussion about this on the kernel
lists was in 2013. I am curious as to where we are in terms of merging cpuidle and cpufreq into the scheduler, I known we have
a basic patch series written in 2013 but seems that's about it. We also don't seem to have any test software to test this either.
through seems there was some discussion at the plumber's conference in October of this year in Germany. Seems for my understanding
we have hit a few roadblocks.
1. We have no benchmark tools, due to power management being hard to benchmark( lots of user cases)
2. Power Management is complex and doing a best fix all solution is impossible or near impossible
3. Merging cpuidle and cpufreq into the scheduler is a long and complex issue, we need to also benchmark and test for regressions as we 
are prone to error when merging these subsystems in the scheduler
4. This takes me back to my first point. we don't have have benchmark tools or a suite, therefore we shouldn't merge due to not being
able to test for bugs or regressions we cause during the merge of these two subsystems into the scheduler.

My question stands as follows have we got any farther on the above list of issues or not?
Regards Nick 

On 2014-12-18 01:23 PM, Robert P. J. Day wrote:
> Quoting nick <xerofoify@gmail.com>:
> 
>> Greetings Fellow Developers,
> 
>> I am curious about the work going into making the kernel scheduler  
>> more CPU power efficient. I have done some googling on this and
>> am curious about what how is going into the ideas/patches for this work.
> 
>> Nick
> 
>    OFFS ... why are you asking about a clearly advanced kernel feature
> on the kernel *newbies* list? And by the way, Nick, this is another
> of your less endearing features -- your habit of posting one or two
> line posts, effectively saying, "I am interested in topic <X>, could
> everyone drop what they're doing and explain it to me in astonishing
> detail to save me the trouble of doing any work?"
> 
>    You've done this before -- the networking stack, BTRFS, and probably
> more. Rather than do enough research to ask *detailed* and *specific*
> questions, you instead simply request that people here explain something
> to you, and it's getting kind of tiring.
> 
>    Do your own research. Stop asking everyone else to do it for you.
> 
> rday
> 
> 
> _______________________________________________
> Kernelnewbies mailing list
> Kernelnewbies at kernelnewbies.org
> http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
>