SpeedStep in MacOS = no whine & extended battery life

I installed all the updates - h/w and s/w, installed bootcamp and Win XP. Everything is working great ! But whine is still there.

So far no whine on my MBP when booted to my WinXP partition. THis could be in part to the fact that I have not stopped downloading and performing tests while in Windows. It seems that even under my Mac OS partition that as long as the CPU's are both busy the whine does not occur.

MacOS ABSOLUTELY allows the processor to clock up and down. You know not of which you speak. If it didn't, the machines would get anywhere NEAR 3 hours of battery life.

Smply because its not user controllable doesn't mean its not in use.

I just switched from an IBM Thinkpad and used the Rightmark CPU utility on that laptop:

http://cpu.rightmark.org/

It allows access to all the multiplier and voltage selections that the Intel mobile cpu's support - and much more granular control over cpu settings. It allows under-volting at various speeds to conserve battery life when needed. It can be set to load at start-up and run custom settings based on A/C or battery use. Unfortunately, it's WinXP only, but I found it to be extremely useful.

I haven't loaded XP yet, but I plan on giving it a try when I do.

unfortunately for rightmarks 'RMclock' the core duo cannot be undervolted. this is just the way Intel set it.

So the lowest you can go is .95V .

Okay, while I don't believe that OS X doesn't support speedstep, because it obviously gets decent battery life (barely decent) I do wonder why I get no whine whatsoever in Windows XP? But, in Mac OS X I usually have the whine, which has been better since the firmware and 10.4.6 (now it's usually only on with battery and no while on AC).

Seriously though, if it can be determined why there's not a whine in XP, but in OS X unless photobooth is loaded, then maybe it can be corrected?

Nick

Mac OS X does indeed support SpeedStep, and uses it extremely aggressively.

The CPU whining noise is most likely related to C state management ("nap" in Mac OS terms), which is related but orthogonal.

Typically, noise like the whine being discussed ad nauseam here is radiated from power supply components (capacitors and magnetics most of the time), and it is caused by sudden, repeated fluctuations in the current being drawn from the power supply.

SpeedStep transitions don't typically result in sudden changes in power supply current; when the VID bits change the power supply ramps up or down a little, and the load on the power supply changes proportionately.

Conversely, on a transition into or out of C4, the CPU is being switched almost completely off, or on; this represents an enormous change in load for the power supply. If the power supply is making noise, these large changes tend to result in a much louder noise than the smaller changes that you would see from a SpeedStep transition.

Note that the effectiveness of the Photobooth "hack" tends to support the argument that it is nap and not SpeedStep that results in the noise; realtime applications like Photobooth are likely to be inhibiting nap for latency reasons, but very unlikely to be inhibiting SpeedStep.

Mac OS X does indeed support SpeedStep, and uses it
extremely aggressively.

The CPU whining noise is most likely related to C
state management ("nap" in Mac OS terms), which is
related but orthogonal.

Typically, noise like the whine being discussed ad
nauseam here is radiated from power supply components
(capacitors and magnetics most of the time), and it
is caused by sudden, repeated fluctuations in the
current being drawn from the power supply.

SpeedStep transitions don't typically result in
sudden changes in power supply current; when the VID
bits change the power supply ramps up or down a
little, and the load on the power supply changes
proportionately.

Conversely, on a transition into or out of C4, the
CPU is being switched almost completely off, or on;
this represents an enormous change in load for the
power supply. If the power supply is making noise,
these large changes tend to result in a much louder
noise than the smaller changes that you would see
from a SpeedStep transition.

Note that the effectiveness of the Photobooth "hack"
tends to support the argument that it is nap and not
SpeedStep that results in the noise; realtime
applications like Photobooth are likely to be
inhibiting nap for latency reasons, but very unlikely
to be inhibiting SpeedStep.

I'm still trying to grok all this, so please humor me.
Based on the above, the speedstep state transitions are resulting in power fluctuations, leading to the whine? And the reason many don't find any whine in WinXP is because (speculation) it may not be doing appropriate SpeedStep transitions, resulting in additional heat (reported issue) and lower battery life in XP (but no whine)?

All that seems to make sense. The only thing that throws me off are the posts (potentially wrong) from folks claiming that they get BETTER battery life w/ XP (implying a more mature SpeedStep implementation in XP -- which is entirely possible), and they still have no whine (sounds impossible given everything above). Sorry for the long post, but from my understanding both can't be true, unless I'm completely misunderstanding the above. Can anyone provide insight?

Mac OS X doesnt have speed step. it only makes use of sleep states.

Can you justify this naked claim?

At the moment, I’m far more inclined to accept MSmith’s word than yours.

RRS

Actually there's another thread where if you check your system.log in terminal apparantly when OS X boots an error message is generated stating speedstep load has failed. This could be why there's the whine and the heat as well if cores aren't being stepped down properly.

http://discussions.apple.com/message.jspa?messageID=2227076#2227076

Nick