Nehalem Mac Pro systems suffer audio-based performance issues

That is a bummer - expect performance and get less. I have been reading about this for at least a couple months now i.e., not really news .

The first thread I saw about this issue is four months old now Luckily I grabbed a 2008 model last September

July 2009 Macbook Pro 15" nothing to report here.

AppleCare is claiming that it is not an issue, that is is within 'normal'. I would think that they wouldn't make sure a claim out of their butts. and a Pro tower user is just as likely to hit a store instead of making a call, or in addition to the call so one would think if there was an issue it would be in the files. course that's also why there are fans. if things were getting that warm and the fans were a no go, I"d be worried.

course I"m the one with the system that gets several degrees hotter every time I am running anything flash for more than ten minutes, but Adobe assures me that there are no bugs with their software

It wouldn't have surprised me if the numbers and OSs were reversed, where Windows was having the CPU usage shoot up when using audio. But the fact that Win 7 on the same machine doesn't produce the same results is embarrassing for Apple.

It wouldn't be nearly as embarrassing if Apple didn't make the computer, but they do and tie their OS solely to the hardware. They should have figured out their drivers before this since they know their hardware so well. Add on top of this, its a Mac Pro, their top model computer used in many audio recording houses.

Something going on in EFI isn't going to show up in OS X Activity Monitor or any perhaps software running under OS X if it relies upon Apple's sensors if Apple is trying to hide something.

Since Windows doesn't use EFI...


I wonder if it's communicating with Apple, verifying the content and owner of the audio being played?


This would be my first suspect, they have been talking about implementing something like this though EFI for years now.

Oh Windows 7 has it's power issues too..

http://www.computerworld.com/s/artic...taxonomyId=125

Oh god, wouldn't put it past the bastards ay.

Even some kind of file diagnostic info, about how itunes is used or something. It'll be collecting and/or sending something to Apple.

Strange. I have two Macpros from 2009. They seem to be running fine. We monitor the stats and "benchmark" new hardware before we implement it and I am sure the CPU temps aren't doubling. (?) We don't test audio functions per se, but we do allot of audio work in pro-tools and I've never noticed fans blasting or any other performance issue. My freelancers are always the first to tell me there's a problem even if I can't find it.

The temperature fluctuation makes this report sound dubious to me. Off the top of my head average CPU temp floats around 120+ degrees after running half a day or so. So they are saying that the temperatures almost reached 250+ degrees? That just sounds inaccurate. They are used exclusively for Ptools HD or FCP though. I wonder how many machines they tested.

Ok I read the report sounds like a bunch of computers were tested. Guess we're just lucky.

Sure thing -- but I don't understand what all the fuss is about. My Mac Pro (4-core Early 2009 2.66 GHz Nehalem model) when at idle will have its 4 cores at around 137 deg F. With an ambient temp of around 84 deg F when playing iTunes music it's typical for all 4 cores to kick up to 165 deg F. So what? This isn't an issue and the CPU does not overheat and my Exhaust, Intake and Power supply fans hover at their idle speed of 600 RPM. The Booster and Fan expansion slot fans hover around their idle speeds of around 800 to 850 RPM.

The Marcel Bresink's Hardware monitor shows an Upper Limit for the core temps to be 212 deg F.

CPU's do get hot when used -- that's all I can say.

I do not see this as an issue on my 2.66 GHz Mac Pro 4-core Nehalem.

My idle power consumption sits at around 145 watts and when iTunes is playing music it kicks up to around 217 watts. This translates to around 246 BTU/hour.

I guess the question being asked is the rise in CPU temp and power increase warranted for the likes of iTunes playing a music file? I dunno!

What I can tell you is that if I run all 8 core threads at 100% cpu while iTunes is playing a music file the core temps will all kick up to around 207 deg F and power consumption runs at around 270 watts and fans ramp up another 100 RPM from idle speeds.

If I now pause the audio from iTunes the core temps and power draw remain at the levels of 207 deg F and 270 watts.

Again, the CPUs can run very hot but the Mac Pro's cooling system handles it very well and with little to no discernible noise.

So the headline mentions a performance hit, but all I get from this is that power draw increases, so obviously CPU temp will as well. Is there an actual performance problem though???

As far as I know, iTunes is not multi-core enabled. However, the Nehalem architecture has Turbo feature that ramps up clock speed in certain instances when not all cores are active. Therefore, isn't it possible that increased power draw is nothing more than the CPU ramping clock speed on the single core being used by iTunes, thereby requiring increased power draw, and consequently, an increase in heat? Moreover, you wouldn't necessarily see increased CPU loads either since my understanding is that CPU load and CPU speed are not directly tied. You could ramp CPU speed to infinity without any corresponding increases in load, right?

Really you guys, you think a stealthy phone home routine is going to drive up CPU temps and inflict a 20% performance hit? Get a grip.

It happens even when you simply plug in a usb audio device.

Your idle temperatures sound pretty high to me.

Eh mate. I think they are talking about he mac pro tower which is priced okay but honestly, Apple towers used to be for the Pro and Prosumer for recording and the likes. Why they have to use Deon that take special ram so you cannot upgrade your own sticks, is beyond me.
More Apple lockdown if you look at it. And a 20% hit is huge, especially in Logic or Pro Tools.

Nehalem Macs are cursed. The iMacs have issues too. Maybe that's what Nehalem is - a Jewish swearword.

Bring back the golden age of Core 2 duo!

hahah very funny!

This is what is mean I guess when they say the mac pro is for power users!

More power to the user, then!

That's because the MacBook Pro released in June 2009 does not use Nehalem CPUs, and is not a Mac Pro.

Sounds like a driver issue. Does it in Mac OS X, but not not in Windows.


Doesn't EFI just run a BIOS emulation?

It seems like Apple's QA testing over the past couple years has gone into the toilet - all I know is they can have my pre-metal (and non-glossy) 24" iMac when they pry it out of my cold, dead hands!

Of course it has, they've had to use everyone for the maxipad, iphone, and ipods. It's becoming apparent more every day, Apple doesn't care about their computer side.

Either way, I have slightly noticed these issues. I have a 8 core 2.66, and I'm usually idling around 38deg Celsius (normal), after a full work day my computer is around 54deg, and if I get after it and max out the cpus from compiling or encoding I can get it up to 65deg. I play iTunes all day long.

I have a temp. monitor. With just iTunes playing I can get the northbridge up to a scorching 65deg Celsius. It seems the Northbridge heats up the most, so it's definitely more of a driver issue.

I don't see them ignoring this situation forever. The new cpus will be based on the same northbridge drivers, and they'll have to correct those. Perhaps a 10.6.3 update? Either way I have yet to update to 10.6.2.

Not to derail the thread, but the other issue they completely overlooked was itunes / quicktime crashing when returning the 2009 mac pro w/ ati 4780 back from sleep. I heard snow leopard corrects this, but still. Either way, I don't think the problem is as bad as these media sites are making it out to be.

I haven't seen this at all on my system (using Mac Pro 2.66Mhz dual xeon 2009) . Sounds very much like a CODEC issue with either mp3 or AAC files. Has anyone tried different encoded MP3 and AAC files
to see if there is any difference ? Does the issue happen in quicktime too ?

The problem doesn't seem to be defined clearly enough to solve.

No. EFI is a replacement for BIOS. On PC's EFI can emulate BIOS for OS's that don't support EFI directly, but Apple doesn't (as they have never used nor relied on BIOS).

http://en.wikipedia.org/wiki/Extensi...ware_Interface

Yes, because Apple is completely incapable of focusing on more than one device at a time. Never mind they all use the same core technologies and have benefited extensively from the cross-pollination between categories.



What a load of rubbish. I don't know why so many are threatened by Apple's success in area's other than the Mac - if you think Apple is going to abandon the Mac or make dramatic changes to the Mac experience you are seriously delusional.

Having said this, it is annoying that simply playing iTunes can spike my Mac Pro's CPU temperature - but it doesn't seem to be dangerous or adversely affect the CPU's. I have thought about getting something like SMC fan control and bumping up the low speed thresholds for the fan's, but honestly it hasn't shown itself to be a significant issue.

I do want to google around and see if there is an explanation why CPU A is routinely about 20% hotter than CPU B. Probably has to do with it's placement in the case and airflow since the CPU's are in-line. That was a nice thing about the G5's - they were staggard - but then again if they weren't they probably would have melted

First of all, it was widely known that leopard was delayed because of the iPhone. Same technologies does not mean the same thing. Just means the same basis.


It's already happened on more than one occasion. If their market took a shift towards smaller devices, that is where their attention will go. Many companies do this and often paint themselves into a corner. Apple will go where the money is.

BTW that line that you quoted, was supposed to be half joking. But it is apparent they have been ignoring major issues on their computing side (or all sides) for many years. Security holes take forever to get patched, OS problems take forever to get fixed, seems each computer they release has some sort of issues that go wrong with it (small or major) and they take their sweet time getting to it (if they ever do).


I will point out something I haven't seen very many sites point out. It's the northbridge that gets hit hard with the heat. Seems the northbridge almost goes up in heat immediately and takes cpu A a while to get there. I think this is more of a northbridge problem than anything else.

OK -- here's my test results for running on my Mac Pro (early 2009 model) 4-core 2.66 GHz Nehalem computer.

I ran two tests.

Test 1. Ran HandBrake to encode two BBC Rough Dimand DVD series to Apple TV format using HandBrake defaults.

Test 2. Same as 1 except I had iTunes playing audi files during the HandBrake execution.

The HandBrake execution was consuming 70% of all my core's CPU cycles and core temps all climbed to around 200 to 206 deg F. Power consumption was at around 95 watts for the CPU/Processor.

Test 1 completed in 3,685 wall time seconds

Test 2 completed in 4,272 wall time seconds

Difference in walltimes was 587 seconds.

Test 2 was slower than Test 1 by approximately 16%

I should run Test 1 several times to understand what its walltime variation would be -- but at this point it does seem that iTunes playing audio files is somehow influencing the likes of HandBrake execution performance.

The HandBrake is very parallel and will consume as many threads as the computer provides and it's very CPU intensive.

The iTunes CPU use is minimal so should not impact HandBrake running alongside it and in addition HandBrake leaves around 30% of my CPU cycles for other processes to use. So on the face of it the iTunes and HandBrake, from a CPU use standpoint, should not interfere with each other. That leaves at least two other resources or features that could be influencing the interaction between iTunes and HandBrake.

One would be memory bandwidth contention. Is iTunes and HandBrake fighting over the aggregate memory bandwidth and slowing each other down ?

Another thought which has been voiced/mentioned elsewhere is that somehow the Intel/Nehalem turbo boost is not being allowed to accelerate Handbrake's use of the cores but disallowing the cycle times to be increased from 2.66 GHz to 2.793 GHz, then to 2.926 GHz and finally 3.059 GHz.

I find it of interest that the increase from 2.66 GHz to 3.059 GHz is 15%.

The difference between Test 1 and Test 2 is 15.93%

So, could the HandBrake degraded performance be supported by iTunes not allowing the turbo boost feature to favor/help HandBrake complete faster?

At this point I'm apt to say this issue is not bogus and Apple needs/should in all fairness explain to Mac Pro (2009 models) users what is causing this. I suspect it may be a fact of life and we have to live with it -- but then on the other hand Apple may have a fix for this.

Hmmmm ---- interesting...

looks like this affects me, too. I tried out the scenario they describe and I get exactly the same result.

It certainly is weird and I hope that they fix it, but I can't say I've ever even noticed the issue prior to seeing this story, so I'm not really suffering here.

I have a much bigger gripe involving what appears to be a bug in the Finder that was introduced sometime after 10.6. It basically prevents me from using utilities like rsync, and that really sucks.

That's because they want you to use iSync instead