Page 6 of 11

Posted: Mon Dec 17, 2007 7:34 pm
by Yhtomit
And is Apple responding to this?

Posted: Mon Dec 17, 2007 7:44 pm
by zeropoint
Not yet....other than an initial piece of bs from the applecare senior engineer (supposedly) who tried to tell me that the software I am using is obviously not coded for dual core processors. Fired a response back and now the issue has been "escalated". Obviously the more people that hit them up on this the better.....so....you know what to do people.....

Posted: Tue Dec 18, 2007 9:17 am
by Yhtomit
And does the problem also exist when using other cpu intensive programs like Logic, Cubase....?

Posted: Tue Dec 18, 2007 9:23 am
by Lo-Fi Massahkah
And does the problem also exist when using other cpu intensive programs like Logic, Cubase....?
Speed stepping per say is when the CPU is told to "slow down" to a lower frequency to conserve power. So it should be equal to all apps.

On Windows you actually have control over this and can set it to "full power", "conserve battery" and a few in between. Seems this can't be done in OSX natively. So you need an app like Coolbook to control it.

I'd like to know if OSX takes into account that my Mac Mini is not a lappy and doesn't really need conserving battery power.

-M

Posted: Tue Dec 18, 2007 9:25 am
by zeropoint
apparently yes.....it is system wide on my machine...certainly for all audio apps. Live 6 and 7, Reason 4, Chuck.......

Posted: Tue Dec 18, 2007 9:41 am
by Timur
I do wonder why Audio apps produce problems with Speedstep/CPUnow at all. These techniques can change the clock speed really fast and monitor the CPU load all the time. When they clock down the CPU it usually means that the CPU just is not busy enough to need a higher clock. But Live is one of those apps that tend to produce drop-outs and glitches nevertheless. My assumption is that they produce such short CPU spikes that Speedstep/CPUnow don't care to raise the clock speed. Hence the necessity to manually switch the CPU clock to maximum. This is too bad, because you're wasting alot of power just for those spikes.

In Windows there are several apps that allow to take over this process from the CPU driver and allow you to define your own settings like raising the clock speed sooner. But since these apps have to run at Realtime priority they can and will lead to dropouts themself.

Posted: Tue Dec 18, 2007 9:51 am
by bleepsnbreaks
Yep. for me its all the time, even when I run a stress test app. THe CPU never goes above 800 unless I set it hard in Coolbook.

Posted: Tue Dec 18, 2007 10:05 am
by Yhtomit
And does this mean your macbook pro is getting really hot? (and so noisy too?).. when cool book is running and communicating with the cpu

Posted: Tue Dec 18, 2007 10:45 am
by Timur
Coolbook does not load the CPU, it just clocks the CPU to its maximum clock speed. This will result in slightly more power consumption and thus a bit more heat, but not like if you were running the CPU at maximum load all the time. So don't worry. To give you some comparison. My desktop A64 X2 CPU runs at 1416 MHz and 1.136v when clocked down at an idle temperature of around 36/37°C. When clocked to maximum it runs at 2832 MHz and 1.392v and an idle temperature of around 37/39°C. Both measured with the fan running at the same speed.

Posted: Tue Dec 18, 2007 12:33 pm
by Yhtomit
And what does history tell you? Is apple good in fixing things like these? .... I hope tiger will get the patch too then................ I was considering the macbook pro + Live 6 + my Multiface II + TIGER!! .... not a good idea....................... :? :(

Posted: Tue Dec 18, 2007 12:52 pm
by zeropoint
Timur wrote:I do wonder why Audio apps produce problems with Speedstep/CPUnow at all. These techniques can change the clock speed really fast and monitor the CPU load all the time. When they clock down the CPU it usually means that the CPU just is not busy enough to need a higher clock. But Live is one of those apps that tend to produce drop-outs and glitches nevertheless. My assumption is that they produce such short CPU spikes that Speedstep/CPUnow don't care to raise the clock speed. Hence the necessity to manually switch the CPU clock to maximum. This is too bad, because you're wasting alot of power just for those spikes.

In Windows there are several apps that allow to take over this process from the CPU driver and allow you to define your own settings like raising the clock speed sooner. But since these apps have to run at Realtime priority they can and will lead to dropouts themself.
I don't think this is what was happening in my case....my pre Coolbook cpu stress test with Xbench showed a processor performing at one third of it's capability...41.2 as opposed 120.2.....the speedstepping just wasn't kicking in at all....

Posted: Tue Dec 18, 2007 3:36 pm
by R.J.Dubya
Mammalux wrote:
Timur wrote:I do wonder why Audio apps produce problems with Speedstep/CPUnow at all. These techniques can change the clock speed really fast and monitor the CPU load all the time. When they clock down the CPU it usually means that the CPU just is not busy enough to need a higher clock. But Live is one of those apps that tend to produce drop-outs and glitches nevertheless. My assumption is that they produce such short CPU spikes that Speedstep/CPUnow don't care to raise the clock speed. Hence the necessity to manually switch the CPU clock to maximum. This is too bad, because you're wasting alot of power just for those spikes.

In Windows there are several apps that allow to take over this process from the CPU driver and allow you to define your own settings like raising the clock speed sooner. But since these apps have to run at Realtime priority they can and will lead to dropouts themself.
I don't think this is what was happening in my case....my pre Coolbook cpu stress test with Xbench showed a processor performing at one third of it's capability...41.2 as opposed 120.2.....the speedstepping just wasn't kicking in at all....
Although it sounds like the santa rosa chips are the worst case, the oder mbp's are also inefficient with Live. I've got a 2.16 c2d, and with the performance test, I sometimes get crackles with the 8 tracks, but it always calms down after a while. Then when I duplicate tracks, it steadily crackles on the tenth (2nd one added), and doesn't get better. With coolbook, I can smoothly run 13! It only breaks up on the fourteenth.

So although the speed-stepping works better on my older mbp, there is still something very wrong about how live (and possbly other DAWs) communicate with the speed stepping function. - Logic is more efficient though, so it seems ableton is especially bad at this communication, or maybe it's very short spikes as someone else mentioned,

Posted: Tue Dec 18, 2007 5:05 pm
by Timur
It's not the task of Live or any other DAW to handle Speedstep. Speedstep has to be handled by the OS via a system/CPU driver.

Posted: Tue Dec 18, 2007 5:11 pm
by Lo-Fi Massahkah
Timur wrote:It's not the task of Live or any other DAW to handle Speedstep. Speedstep has to be handled by the OS via a system/CPU driver.
Thanks. I was getting nervous that I was confused here.

And it doesn't necessarilly mean that it jumps up/down. I know on windows, some of the power schemes set my processor to permanently work at for example 800 mhz, being a 2 ghz Pentium M. Saving battery, but not exactly giving the power that the CPU could deliver.

-M

Posted: Tue Dec 18, 2007 6:24 pm
by Timur
Power schemes in Windows allow control over CPU steppings, seems like there is no such thing on Mac OS (beside Coolbook). <shrug>