Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Audio Decoding KILLS MacPro (2009) green factor

Until today Apple engineers has no answer to my simple question:

Why does my Mac Pro uses up 50W of power just to play an audio file in iTunes while it only uses up 30W of power to run 700 RAW image conversion to JPEG in Photoshop Lightroom and stitching a 60MP image in PTGui?

I still remembered one joker at Apple Singapore said: Some 3rd party apps are not optimised for Snow Leopard yet.

WTF???? :eek:
 
AZREOSpecialist said:
With iTunes playing music in the background the score is 17,330, without iTunes the score is 18,069. This represents a 3.1% performance hit.

I have a 3.33 GHz W3580 Xeon which I installed myself.
Click to expand...

I make that -4.0898776% performance dip.

Is this a clue - your installing a 3.33GHz CPU?
What was involved in that, what changed from the standard system from Apple?
 
FWIW
Cinebench R10 on 09MP 2x2.66GHz Xeon in 64bit mode, GTX285, 10.6.2
Code: 
                             nothing iTunes   %diff
Rendering (Single   CPU):     3777     2772  -26.6%
Rendering (Multiple CPU):    23724    22486   -5.2%
Shading (OpenGL Standard):    6113     5109  -16.4%
Multiprocessor Speedup:   8.11
 
smacman said:
A few weeks ago I tried to get the spotlight on this issue but nothing came of it. The fact that every second page or so somebody posts claiming their machine is not affect or that it's a non-issue probably does not help our cause...
Click to expand...

Doesn't Arn use a 2009 MP himself? I could have sworn he said in the thread when they first came out that he was upgrading to one.
 
I really don't know how to relate to the difference in performance after my Cinebench test the other day. The numbers show that when iTunes is playing, there's a drop in performance as illlustrated below:

Rendering (Single CPU): -26%
Rendering (Multiple CPU): -2,2%
Multiprocessor Speedup: 5.21
Shading (OpenGL Standard): -20%

What does this mean? Shouldn't one expect a performance drop when there's stuff going on in the background?..

What concerns me more though are the temperatures. After 15 minutes of iTunes:

CPU A Temperature Diode: 75C.
CPU Cores: 82-84C.

Is there a possibility that the different temperature monitor programs out there aren't showing correct values?

Don't get me wrong, I strongly suspect that something isn't right with the Nehalems when it comes to audio decoding, I'm just trying to foresee what Apple Care will tell me when I give them my results.

Thanks.
 
fat binary said:
I really don't know how to relate to the difference in performance after my Cinebench test the other day. The numbers show that when iTunes is playing, there's a drop in performance as illlustrated below:

Rendering (Single CPU): -26%
Rendering (Multiple CPU): -2,2%
Multiprocessor Speedup: 5.21
Shading (OpenGL Standard): -20%

What does this mean? Shouldn't one expect a performance drop when there's stuff going on in the background?..

What concerns me more though are the temperatures. After 15 minutes of iTunes:

CPU A Temperature Diode: 75C.
CPU Cores: 82-84C.

Is there a possibility that the different temperature monitor programs out there aren't showing correct values?

Don't get me wrong, I strongly suspect that something isn't right with the Nehalems when it comes to audio decoding, I'm just trying to foresee what Apple Care will tell me when I give them my results.

Thanks.
Click to expand...

Your results are in line with what most other's have reported (myself included). A CPU Temperature Diode reading of 75C indicates that your CPU is operating at or above Intel's maximum specified temperature of 67.9C, and the performance loss is clearly visible. Keep in mind that when doing a Cinebench test on ANY other Mac (ie. 2008 Mac Pro, Macbook Pro, 27" i7 iMac, etc.), there is less than 0.5% difference in results with and without audio playback. Something is definitely wrong with these machines, and I encourage you to persist with Applecare. It will be a frustrating experience, but unless everyone does their part, we may be stuck with these lemons...
 
handheldgames said:
I've talked to my top software/hardware engineers... the only thing we can come up with is that it's the software based APU or somethings broken somewhere.

Intel Nehalem (microarchitecture) is touted at providing 30% lower power usage for the same performance.....
Click to expand...


Such a long thread, I didn't read all messages so maybe i'll be overposted here. We have a 2.6Ghz 2006 quad core MP with 8Gb RAM and 1.5Tb HD RAID. We use this lady to record live audio for 64 SDII tracks (48Khz, 24bit) with Logic, using a MOTU PCI-424 and 3 2408's optical access from a PM5D mixer. Never had ANY problem for recordings for up to 3 non stop hours. We've noticed a heavy speed fan action, strong HD noise and up to 58 ºC temp on processors at start, but when system stabilizes (after 12 minutes or so) fans get to regular speeds, HD's are not noisy and temps go down to 50 ºC. Initial hump takes this baby to the limits!

We've noticed that sometimes we had to record audio with a not so clean HD array and it took a LOT of MP's resources, but when using a fresh formatted and initialized array things go smoother.

I don't think that a regular MP3 out of iTunes takes so much resources, never had to look over though, Strange behavior for such an easy process...

Our machine is working as expected, maybe the new MP's have an issue here? We will never change our MB as long as it lasts.
 
maldoblaz said:
Such a long thread, I didn't read all messages so maybe i'll be overposted here. We have a 2.6Ghz 2006 quad core MP with 8Gb RAM and 1.5Tb HD RAID. We use this lady to record live audio for 64 SDII tracks (48Khz, 24bit) with Logic, using a MOTU PCI-424 and 3 2408's optical access from a PM5D mixer. Never had ANY problem for recordings for up to 3 non stop hours. We've noticed a heavy speed fan action, strong HD noise and up to 58 ºC temp on processors at start, but when system stabilizes (after 12 minutes or so) fans get to regular speeds, HD's are not noisy and temps go down to 50 ºC. Initial hump takes this baby to the limits!

We've noticed that sometimes we had to record audio with a not so clean HD array and it took a LOT of MP's resources, but when using a fresh formatted and initialized array things go smoother.

I don't think that a regular MP3 out of iTunes takes so much resources, never had to look over though, Strange behavior for such an easy process...

Our machine is working as expected, maybe the new MP's have an issue here? We will never change our MB as long as it lasts.
Click to expand...

This issue is specific to (Early 2009) Mac Pros.
 
I think a new thread is in order because people are getting the wrong ideas.

One with the facts, screenshots and everything...
 
Transporteur said:
Hardware monitor. It reads out all core diode temperatures and not only the heatsink sensors like iStat does.
Click to expand...

Hardware monitor does not read the currents etc unless you buy it...

Besides temps are a symptom not the cause of our problems, so I'm forging the performance path, only!
 
Concorde Rules said:
For your personal information: iStatMenus reads the same as HW monitor
Click to expand...


Well, seems as if I've got a very special computer again.

HW Monitor CPU A Temperature Diode: 66°
HW Monitor CPU B Temperature Diode: 53°
HW Monitor CPU A Heatsink: 55°
HW Monitor CPU B Heatsink: 43°

iStatMenus CPU A: 55°
iStatMenus CPU B: 43°

So yes, iStatMenus reads out the same sensors as HW monitor, but apparently the wrong ones. The Heatsink temperature is completely out of interest, as it is considerably lower than the actual core temperatures and often doesn't exceed Intels maximum operating temperatures for the core temperature.
What we need is the core temp!
In addition to that, HW monitor reads out the temperature of every single core, not only of the processor tray in general.
That said, my core Temps for CPU A are currently at 74°+! These are the temperatures that are of interest as they exceed the limits set by Intel dramatically if you are stressing your computer.
 
Transporteur said:
Well, seems as if I've got a very special computer again.

HW Monitor CPU A Temperature Diode: 66°
HW Monitor CPU B Temperature Diode: 53°
HW Monitor CPU A Heatsink: 55°
HW Monitor CPU B Heatsink: 43°

iStatMenus CPU A: 55°
iStatMenus CPU B: 43°

So yes, iStatMenus reads out the same sensors as HW monitor, but apparently the wrong ones. The Heatsink temperature is completely out of interest, as it is considerably lower than the actual core temperatures and often doesn't exceed Intels maximum operating temperatures for the core temperature.
What we need is the core temp!
In addition to that, HW monitor reads out the temperature of every single core, not only of the processor tray in general.
That said, my core Temps for CPU A are currently at 74°+! These are the temperatures that are of interest as they exceed the limits set by Intel dramatically if you are stressing your computer.
Click to expand...

That is normal. iStat has always only reported the Heatsink temp. For this reason, I have always taken peoples reported temps with a grain of salt until they clarify which temps they are referring to. The ones that matter, are the diode temps, which only Temperature Monitor reports.
 
smacman said:
That is normal. iStat has always only reported the Heatsink temp.
Click to expand...

I'm well aware of that, that's why I always read the temps with HW monitor.
Heatsink temps are completely out of interest, currently mine are 23° below the actual core temperature.
 
Transporteur said:
I'm well aware of that, that's why I always read the temps with HW monitor.
Heatsink temps are completely out of interest, currently mine are 23° below the actual core temperature.
Click to expand...

Ahhh.. I'm sorry. I took
Well, seems as if I've got a very special computer again.
Click to expand...
to mean you thought your readings were unique.
 
Nah, that was irony. Next time I put a smilie after it. ;)

However, my suggestion is to read the temperatures with HW monitor and the power usage with iStatMenus. That would provide the most accurate measurements, at least for people that don't have an external power meter.
 
I think all CR is compiling right now are simple Cinebench results with and without audio. I see his point as Apple seems to love to deem any Temp and Wattage results as "within limits". When it comes to a performance degradation, what is "within limits"?
 
smacman said:
I think all CR is compiling right now are simple Cinebench results with and without audio. I see his point as Apple seems to love to deem any Temp and Wattage results as "within limits". When it comes to a performance degradation, what is "within limits"?
Click to expand...

I believe it, but it seems bizarre - not just to be selectively blind like that, but a company wide uniform mindset. Seems to infer a clamp down and dictate from on high. So just what are they covering up and frightened of.

Reminds me of police advice here. Asked if they would use video footage of crime, they told me NO, unless it was incontrovertible proof of a crime. So the thug lunging forward with a knife, the sounds of stabbing, followed by the scene of the villain with bloody knife and victim bleeding - all ignored and they refuse even to look at it.

One wonders why those apparently charged with dealing with such issues, that are of obvious concern to us, work so hard to avoid the evidence.
 
Are the 'Early 2009 MP's' affected aswell? I bought mine in July '09. The post over at Ars mentions the 'Late 2009' models or the Oct '09 models.

I usually have 5 HDD's and a MOTU interface connected at all times. I haven't noticed any drops, but like the post mentions, the fans do not kick in. I usually run Ableton Live/Logic Pro and haven't noticed any performance drops.

If this issue does exist on mine, I should be worried :( I only use iStat for monitoring, and haven't noticed any bumps. I don't use iTunes much, but use Spotify and VLC for movies all the time.


P.S.- What version of iStat allows viewing of Power consumption?
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back