PDA

View Full Version : Any way to view/visualize Timer Coalescing?


adcx64
Jun 14, 2013, 07:48 PM
Hello. I'm really interested in this new feature, upon installing to my MBA, I am curious if there is any way to see this in action? Activity Monitor now shows App Nap, but what about when Timer Coalescing kicks in?

w0lf
Jun 14, 2013, 07:51 PM
It's always on when on battery power.

Also timer coalescing has already existed in Windows since 7. It's not really gonna blow your mind.

adcx64
Jun 14, 2013, 08:02 PM
It's always on when on batter power.

Also timer coalescing has already existed in Windows since 7. It's not really gonna blow your mind.

interesting. i just thought it was really neat to be honest.

andrezuppe
Jun 14, 2013, 08:11 PM
It's always on when on battery power.


Can we force it to be always active also on ac power (maybe to keep macs cooler)?

Bye,

Andrea

rainydays
Jun 15, 2013, 08:11 AM
Can we force it to be always active also on ac power (maybe to keep macs cooler)?

Bye,

Andrea

Wondering this too. Most of the time I'm willing to sacrifice a little performance for a cooler running CPU.

adcx64
Jun 15, 2013, 11:53 AM
Correct me if I'm wrong, but isn't the point of the feature keeping the cpu cool while gaining performance as well?

chown33
Jun 15, 2013, 02:01 PM
Correct me if I'm wrong, but isn't the point of the feature keeping the cpu cool while gaining performance as well?

No, the point is reducing power consumption. CPU temperature is a consequence of power consumption, not its cause.

adcx64
Jun 15, 2013, 02:03 PM
No, the point is reducing power consumption. CPU temperature is a consequence of power consumption, not its cause.

So the freed cpu downtime does not equate to better performance?

mrapplegate
Jun 15, 2013, 02:09 PM
By chance did you look over Apple's PDF about it?
http://images.apple.com/osx/preview/docs/OSX_Mavericks_Core_Technology_Overview.pdf

Some interesting graphs, but you won't be able to visualize it on the actual system.

w0lf
Jun 15, 2013, 02:15 PM
Correct me if I'm wrong, but isn't the point of the feature keeping the cpu cool while gaining performance as well?

No the point of the feature is to allow your cpu longer idle periods and thus using less energy when on battery power.

chown33
Jun 15, 2013, 04:53 PM
So the freed cpu downtime does not equate to better performance?

It depends on exactly what you mean by "better performance".

Timer coalescing should result in MORE CPU downtime, not less. That is, the CPU is placed in a less power-consuming lower-speed state for a larger percentage of time. From a power-consumption viewpoint, this is "better performance", in that it uses less power than if timers weren't coalesced. It's also "better performance" from a temperature viewpoint, because less power consumed means less heat to dissipate.

There is no CPU downtime that's "freed" by timer coalescing. I'm not even sure what you mean by that. "Down" is a state the CPU is in, not a resource that can be used or freed. The CPU is either up or down, or at some intermediate power level. If anything, CPU downtime is increased, not decreased. In any case, increased downtime is a consequence of timer coalescing, not a cause. That is, by performing timer coalescing, the amount of downtime is increased (hopefully). It's not that downtime is somehow "freed" which then results in timers being coalesced.

Conversely, if "better performance" means "less latency" or "higher speed", then increasing CPU downtime may negatively affect both of those. That is, it may increase the latency for timers that are coalesced, or it may increase the granularity of timers, such as from 10 ms granularity to 100 ms. Things like human-perceived responsiveness, network speed, etc. may be affected by increases in latency.

So it depends entirely on what you mean by "better performance".