Re: cpu frequencies
Re: cpu frequencies
- Subject: Re: cpu frequencies
- From: Michael Smith <email@hidden>
- Date: Sat, 21 Jun 2008 12:03:00 -0700
On Jun 21, 2008, at 7:06 AM, Jean-luc Chasseriau wrote:
Thanks all for these information!
I would like to quickly give some feedbacks from your help.
1) sysctl doesn't seem to provide the _current_ cpu frequency since
*cpu* values don't change while time is passing by.
That is correct. These values are maxima.
3.c) why retrieving this value?
The main goal is quite stupide but the background (how to) is still
interesting to know. I would like to make a simple widget which is
displaying cpus frequencies on the menu bar. (I saw that on linux +
gnome, very useless I agree, but fun! :))
Not only useless, but inaccurate. You might as well generate random
numbers and display them. That is one of the reasons the information
is not made available; see below.
Michael, could you give me an entry point to help me where can I
start to look to gather this?
No.
The Darwin kernel uses core P-states very aggressively on many
platforms. The very act of running code to examine the current P-
state and draw something based on that value will change it. It is
not uncommon to see over 100 P-state transitions per second on a given
core.
Likewise, C-state transitions are very common (tens to hundreds per
second, depending on workload). The details of the code that manages
these states (not the policy) can be found in the published Darwin
sources, but all this will tell you is that the system is P/C-state
agile on a core by core basis, and that these states are intimately
tied to the work state of individual cores. There is no user-relevant
concept of "current P/C state" for a given core.
= Mike
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden