It could be that OS X is uncovering some hardware bug because it's
using the hardware in a way that other OS's are not. I wouldn't like
to guess how likely that is, though. Perhaps not very likely at all.
Got a tip for us?
Let us know
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
- Thread starter handheldgames
- Start date
- Sort by reaction score
It could be that OS X is uncovering some hardware bug because it's
using the hardware in a way that other OS's are not. I wouldn't like
to guess how likely that is, though. Perhaps not very likely at all.
http://discussions.apple.com/thread.jspa?threadID=2228019&tstart=0&start=435
I hope this really is going to happen. Sick and tired of it.
I hope this really is going to happen. Sick and tired of it.
I think it's OSX only since this doesn't happen in Windows. I would think that if it were a hardware bug (which firmware would fix) it would happen in any OS.
Hey, Folks;
First-time poster here. I reg'd and am posting mainly because this thread was referred to by the ongoing discussion about this issue at Apple Discussion forums.
I am going to post my numbers below (which I posted at Apple Disc) and what apple care told me yesterday.
_______________________
2009 Mac Pro Audio experiment posted at Apple Discussions
Set up: fans running at Apple's own parameters; no fan controller being used during the tests. Also, the machine was on, awake and idling for about 20 minutes before beginning any of these. Tests were done on a practically brand-new installation of Leopard 10.5.8. EDIT: My Mac is a 2009 8-core 2.26, 6GB RAM, GeForce 120. Also, it is a refurb unit I bought from apple's refurb site.
My tests:
The temps I am giving are for CPUA Diode and CPUB Diode, as people indicated that was a more important measure than the CPU heatsink. Where you see #/#, the first one is for CPUA, the second is for CPUB. Northbridge seemed not to get higher than 60. All temperatures in Celsius.
QuickTime: Opened a DV video clip with sound which I recorded. Stereo sound, NTSC Starting Temps: 37/31. Hit Play. At 25 seconds 46/38; 1:15 48/40; 2:30 50/42; 4min 52/43; 5 mins 52/44 (Northbridge up to 60) Stopping clip didn't precipitate temps falling but quitting Quicktime did!
iTunes:
Starting Temps: 38/30: play music...at 2:19: 50/42 4min: 52/43; 12 mins: 55/45. temps fall immediately upon quitting iTunes
Blender: test.blend file You can download this file yourself. This is the file blender folks use to bench their machines against everyone else's. It is a cross-platform test which uses Radiosity. As you probably know, Radiosity really guzzles the juice, so it's a good CPU test.
I increased the size of the render by a factor of 3; the stock file renders in just 23 seconds on my machine. I wanted a minutes-long render session for purposes of gathering temps data.
Blender set to 8 threads; render size set to 2400 x 1800pixels.
Starting Temps: 39/31. Initiated Render. 6 cores kept maxed, others spiking here and there. 790% CPU usage in Top. at 1 min: 46/39. 3 mins: 49/38
As render nears end and more "Parts" completed, fewer cores needed to continue rendering, temps begin dropping. Render finished at 3 mins 45 seconds
I believe that this covers it, as far as I have applications to test. I know that some people did multiple demanding tasks in conjunction with audio, but it's clear that playing audio causes the temp spikes. At no time did I hear the fans rev up. Not very scientific, I admit, but my license for Hardware Monitor hasn't arrived, so...sound is my only gauge of fan speed at this time. Most telling is that no test other than the Blender render caused anything near even 1 core to max or even to spike fully. Blender by orders of magnitude was the most demanding test I ran.
________________________________________
I also observed a 65-watt spike in Power drawn while playing music from iTunes.
Called Apple Care yesterday, and related the experiments above, that I thought there was a bug, etc, and was told that Apple knows about this issue and is working on a fix...that it would be a "firmware-like" fix, not tied to the OS, as it was determined to be a hardware issue, not an OS issue (their words).
I was also told that I would be called on phone when the fix was ready, and that the time frame would be "two weeks max." Also, their words.
This is all very positive, of course, but I feel that we should continue our efforts until after the fix is known to be working, as several have commented that Apple's ears need to be opened a bit and the whole relationship between users and Apple needs to be more responsive and respectful.
My little ears nearly burst into flames as I read account after account of Mac Pro users given the brush off on this issue...either getting the Deny All Knowledge, or "Within specifications" brush-off.
That this issue has been known about by users far and wide for the best part of 1 year, and that the problem can be easily duplicated by any 2009 Mac Pro User, and that Apple did nothing for almost 1 year is beyond the pale.
In the meantime, how much extra juice have we all burned through? Extra electric bills? Potential damage to our expensive rigs?
When we call Apple with issues, especially ones as grave as this one, they need to sit up straight and pay attention, and not reach for the "Placate and get rid of caller" cue card.
That is full-on malarkey.
Thanks is due to several posters who have sweated it out for 6 months or more, and to the sites like Ars and others who deemed this to be news worthy.
cheers!
phzr
First-time poster here. I reg'd and am posting mainly because this thread was referred to by the ongoing discussion about this issue at Apple Discussion forums.
I am going to post my numbers below (which I posted at Apple Disc) and what apple care told me yesterday.
_______________________
2009 Mac Pro Audio experiment posted at Apple Discussions
Set up: fans running at Apple's own parameters; no fan controller being used during the tests. Also, the machine was on, awake and idling for about 20 minutes before beginning any of these. Tests were done on a practically brand-new installation of Leopard 10.5.8. EDIT: My Mac is a 2009 8-core 2.26, 6GB RAM, GeForce 120. Also, it is a refurb unit I bought from apple's refurb site.
My tests:
The temps I am giving are for CPUA Diode and CPUB Diode, as people indicated that was a more important measure than the CPU heatsink. Where you see #/#, the first one is for CPUA, the second is for CPUB. Northbridge seemed not to get higher than 60. All temperatures in Celsius.
QuickTime: Opened a DV video clip with sound which I recorded. Stereo sound, NTSC Starting Temps: 37/31. Hit Play. At 25 seconds 46/38; 1:15 48/40; 2:30 50/42; 4min 52/43; 5 mins 52/44 (Northbridge up to 60) Stopping clip didn't precipitate temps falling but quitting Quicktime did!
iTunes:
Starting Temps: 38/30: play music...at 2:19: 50/42 4min: 52/43; 12 mins: 55/45. temps fall immediately upon quitting iTunes
Blender: test.blend file You can download this file yourself. This is the file blender folks use to bench their machines against everyone else's. It is a cross-platform test which uses Radiosity. As you probably know, Radiosity really guzzles the juice, so it's a good CPU test.
I increased the size of the render by a factor of 3; the stock file renders in just 23 seconds on my machine. I wanted a minutes-long render session for purposes of gathering temps data.
Blender set to 8 threads; render size set to 2400 x 1800pixels.
Starting Temps: 39/31. Initiated Render. 6 cores kept maxed, others spiking here and there. 790% CPU usage in Top. at 1 min: 46/39. 3 mins: 49/38
As render nears end and more "Parts" completed, fewer cores needed to continue rendering, temps begin dropping. Render finished at 3 mins 45 seconds
I believe that this covers it, as far as I have applications to test. I know that some people did multiple demanding tasks in conjunction with audio, but it's clear that playing audio causes the temp spikes. At no time did I hear the fans rev up. Not very scientific, I admit, but my license for Hardware Monitor hasn't arrived, so...sound is my only gauge of fan speed at this time. Most telling is that no test other than the Blender render caused anything near even 1 core to max or even to spike fully. Blender by orders of magnitude was the most demanding test I ran.
________________________________________
I also observed a 65-watt spike in Power drawn while playing music from iTunes.
Called Apple Care yesterday, and related the experiments above, that I thought there was a bug, etc, and was told that Apple knows about this issue and is working on a fix...that it would be a "firmware-like" fix, not tied to the OS, as it was determined to be a hardware issue, not an OS issue (their words).
I was also told that I would be called on phone when the fix was ready, and that the time frame would be "two weeks max." Also, their words.
This is all very positive, of course, but I feel that we should continue our efforts until after the fix is known to be working, as several have commented that Apple's ears need to be opened a bit and the whole relationship between users and Apple needs to be more responsive and respectful.
My little ears nearly burst into flames as I read account after account of Mac Pro users given the brush off on this issue...either getting the Deny All Knowledge, or "Within specifications" brush-off.
That this issue has been known about by users far and wide for the best part of 1 year, and that the problem can be easily duplicated by any 2009 Mac Pro User, and that Apple did nothing for almost 1 year is beyond the pale.
In the meantime, how much extra juice have we all burned through? Extra electric bills? Potential damage to our expensive rigs?
When we call Apple with issues, especially ones as grave as this one, they need to sit up straight and pay attention, and not reach for the "Placate and get rid of caller" cue card.
That is full-on malarkey.
Thanks is due to several posters who have sweated it out for 6 months or more, and to the sites like Ars and others who deemed this to be news worthy.
cheers!
phzr
I am still happy with my Pro. Sooner or later all this will be behind us.
At least apple knows that we know and we are as mad as hell.
They will release a fix, be it hardware or software. It will probably be both.
Thanks to all who did so much to get the issue pushed through.
I'll be even happier if I can drop a six core in as a replacement.
At least apple knows that we know and we are as mad as hell.
They will release a fix, be it hardware or software. It will probably be both.
Thanks to all who did so much to get the issue pushed through.
I'll be even happier if I can drop a six core in as a replacement.
Article in Macintosh Performance Guide
This is a really good article. I found in yesterday, and havent noticed it mentioned yet.
Mac Pro power usage spike playing music
http://macperformanceguide.com/Reviews-MacProNehalem-SoundAndPower.html
--Jess
This is a really good article. I found in yesterday, and havent noticed it mentioned yet.
Mac Pro power usage spike playing music
http://macperformanceguide.com/Reviews-MacProNehalem-SoundAndPower.html
--Jess
So do I. We can all crank up fan speeds with apps that already exist...it's what I've been doing since learning of this issue.
Would Apple be that lame as to simply crank the fan speeds without addressing the root cause?
phz
Getting the fix/update
Question to anyone... How do you think this "fix" is gonna be distributed? Will everyone get it via Software Update, or will we have to call Apple Care and specifically ask for it? Someone said that they were told they would be "called" in a few weeks when the fix is available.
--Jess
Question to anyone... How do you think this "fix" is gonna be distributed? Will everyone get it via Software Update, or will we have to call Apple Care and specifically ask for it? Someone said that they were told they would be "called" in a few weeks when the fix is available.
--Jess
pretty much since march but definitely since May i had been listening to music any time that my computer wasn't being used for video editing literally all day and most of the night, consistently. my electric bill is typically in the $170-180 range, and was a little higher in the summer when we were running the AC off an on.
since noticing this bug almost two months ago, i've been charging my iphone from the wall and playing audio through my iphone while i work rather than through my computer... last month i noticed a $40 drop in my in electric bill and i'm looking forward to this month's to see if it's also abnormally low.
i still use my computer as much as ever, but now i avoid firewire and music whenever i can.
i was going to say that it's the only thing that has changed about my electricity usage... but actually, now that i think about it, i also purchased an HDTV and have been using it pretty consistently for the past month so if anything my electric bill should be higher than average.
since noticing this bug almost two months ago, i've been charging my iphone from the wall and playing audio through my iphone while i work rather than through my computer... last month i noticed a $40 drop in my in electric bill and i'm looking forward to this month's to see if it's also abnormally low.
i still use my computer as much as ever, but now i avoid firewire and music whenever i can.
i was going to say that it's the only thing that has changed about my electricity usage... but actually, now that i think about it, i also purchased an HDTV and have been using it pretty consistently for the past month so if anything my electric bill should be higher than average.
The interaction of which you're thinking of is drivers, not firmware. Firmware is rather simple for the affected areas of the system (i.e. boot capablility for USB and FW, and Thermal Profile data for the PWM controller the fans attach to <part of the SMC>).
It may not be appropriate, but it will be the case if they're looking for a quick fix for a more complicated problem.
Generally speaking, it's likely in such situations, as it's cheaper. Corporations are all about the bottom line, and some are more aggressive about it than others, which translates into things like poor customer service, lack of support,... Sound familiar?
The patch is here
Well, its a software patch...
http://support.apple.com/kb/HT4034
Its working for me sofar. No detectable change in cpu a and b wattage before or after audio playback.
Before 3.2W/3.5W After/during about 10mins of the Neville Brothers, its 3.9/3.8W.
Im gonna test with the firewire transfer thing now and report back. Oh yeah, the temps are 33/29 deg during audio playback now! Looks good sofar.
--Jesse
Well, its a software patch...
http://support.apple.com/kb/HT4034
Its working for me sofar. No detectable change in cpu a and b wattage before or after audio playback.
Before 3.2W/3.5W After/during about 10mins of the Neville Brothers, its 3.9/3.8W.
Im gonna test with the firewire transfer thing now and report back. Oh yeah, the temps are 33/29 deg during audio playback now! Looks good sofar.
--Jesse
Well, its a software patch...
http://support.apple.com/kb/HT4034
Its working for me sofar. No detectable change in cpu a and b wattage before or after audio playback.
Before 3.2W/3.5W After/during about 10mins of the Neville Brothers, its 3.9/3.8W.
Im gonna test with the firewire transfer thing now and report back. Oh yeah, the temps are 33/29 deg during audio playback now! Looks good sofar.
--Jesse
Excellent news! Can't wait to try this when I get back to my Mac Pro in a few hours... Keeping my fingers crossed!!!
Just thanked Chris from Ars Technica. Only after his article we started hearing from AppleCare Techs. Also thanks to everyone here who raised attention towards the problem. Still it is just half the way as the fix requires 10.6.2 to install meaning that people like me running on 10.5.X wont be able to install it. But again thank you very much all of you.
Added in Edit:
It works. Actually on the mac pro i tested the patch on after the initial spike when playing music the temp goes down by 1C lol
Im gonna miss coming to this thread to vent lol.
Added in Edit:
It works
. Actually on the mac pro i tested the patch on after the initial spike when playing music the temp goes down by 1C lolIm gonna miss coming to this thread to vent lol.
FW file transfer now causes small power increase.
Copied 40 Gigs from one my my internal drives to an external FW drive. About 5 mins into it and I'm only seeing about a 4W per CPU increase in power consumption with a 1 deg C increase in temp.
I am still seeing no change in wattage or temperature when playing music via iTunes.
Just to reiterate. The problem seems to be fixed. Looks like it was certainly a software patch as most have speculate it would be.
--Jesse
Copied 40 Gigs from one my my internal drives to an external FW drive. About 5 mins into it and I'm only seeing about a 4W per CPU increase in power consumption with a 1 deg C increase in temp.
I am still seeing no change in wattage or temperature when playing music via iTunes.
Just to reiterate. The problem seems to be fixed. Looks like it was certainly a software patch as most have speculate it would be.
--Jesse
No performance degradation on Cinebench tests.
Featured on ars already:
http://arstechnica.com/apple/news/2...r-audio-related-mac-pro-performance-issue.ars
Featured on ars already:
http://arstechnica.com/apple/news/2...r-audio-related-mac-pro-performance-issue.ars
Confirmation of FW transfer issues
Can someone please confirm the fix is in for firewire file transfer?
--Jesse
Can someone please confirm the fix is in for firewire file transfer?
--Jesse
Can someone please tell them that there is no 'Late 2009' Mac Pro?
Only because the Quad got a new Processor it is NO new Mac Pro.
Last iteration is still the Early 2009.
The fix seems to be working for me with on-board audio. I also plugged in my FireWire audio interface and I saw only a small increase in temperature after 20 minutes or so.
We should continue to keep an eye on our temps just in case Apple slowed down the problem rather than eliminating it completely but . . . I'm cautiously optimistic.
So what was that . . . 6 months to stonewall and 8 days to fix it once they were finally forced to admit the bug? Sure am glad I paid for that premium Apple experience!
We should continue to keep an eye on our temps just in case Apple slowed down the problem rather than eliminating it completely but . . . I'm cautiously optimistic.
So what was that . . . 6 months to stonewall and 8 days to fix it once they were finally forced to admit the bug? Sure am glad I paid for that premium Apple experience!