Audio Decoding KILLS MacPro (2009) green factor

[nanofrog]

i was under the impression that the MP heatsyncs/cooling systems were pretty high quality and efficient! i guess not

They're certainly not the low-ish end stock units that Intel ships with thier retail packaged processors (though the current one's aren't horrible; IIRC, people have hit 3.8GHz with an i7-920 on them ).

But the airflow is different than any other system, and they certainly don't go to the hassle of lapping the copper surface of the cooler either. Too expensive. Enthusiasts have to do that by hand (I've seen a vendor or two that will do it for a fee, and they quit at 1000grit). It does help. But a mistake can mean you have to get a new cooler.

But as I've just discovered on the '08's, they had two different coolers (one P/N for the 2.8 and 3.0GHz models), and a different unit for the 3.2GHz units. But the lower clocked units were E series running at 80W, while the 3.2 was an X series, running at 150W. Quite a difference, and even meant a different logic board as well (more robust Voltage Regs it seems).

 

[eemzah]

Hi,

My first post here. Been combing the net for my load temp issues on my Mac Pro Quad Nehalem. Boy I am glad I am not alone. I have talked to my local mac 'hero' and he claims the 09' MacPros were designed to run hot.

Anyway, I can confirm with the rest here that listening to music or watching video alone will spike the load temp to 61~65 deg C and will drop back to 42~48 deg C when stopped. The only work around right now is to turn off two cores if I am just surfing the net and listening to music.

I find it rather strange that listening to music can generate so much heat on the 09' MacPro CPU. I did other much more CPU intensive work like stitching a 300MP image, processing 700 images with Lightroom, doing intricate Illustrator and InDesign layouts all the same time only generates up to 56 deg C max.

Confused.

 

[nanofrog]

I just made a couple of posts in this thread (i5 iMac is in hand of at least one person).

The first is requesting help on determining if it's a problem with software (shows up in the Nehalem iMacs or not, and hopefully soon to be followed by the i7 units), or if it's specific to the '09 MP's. The second is more of an explaination of what's going on.

Hopefully, it will make them aware of the issue might exist on their systems, and help us sort out if it's in fact software or not.

eemzah, you're not alone. Oh wait.... that should be. ()

 

[gugucom]

The only work around right now is to turn off two cores if I am just surfing the net and listening to music.

How do you do that? I'm clueless about switching cores on or off in OS X. I would love to do this when I run EyeTV in standby or recording stuff. Most of the time My MP4,1 is just doing that and I don't need the fan noise and the waste of power.

 

[smacman]

Hi,

My first post here. Been combing the net for my load temp issues on my Mac Pro Quad Nehalem. Boy I am glad I am not alone. I have talked to my local mac 'hero' and he claims the 09' MacPros were designed to run hot.

Anyway, I can confirm with the rest here that listening to music or watching video alone will spike the load temp to 61~65 deg C and will drop back to 42~48 deg C when stopped. The only work around right now is to turn off two cores if I am just surfing the net and listening to music.

I find it rather strange that listening to music can generate so much heat on the 09' MacPro CPU. I did other much more CPU intensive work like stitching a 300MP image, processing 700 images with Lightroom, doing intricate Illustrator and InDesign layouts all the same time only generates up to 56 deg C max.

Confused.

Call Applecare and report this. Refer them to this thread and these threads:

http://discussions.apple.com/thread.jspa?threadID=2112526&start=0&tstart=0

http://discussions.apple.com/thread.jspa?messageID=10550378&#10550378

This is a fixable issue, but Apple will only do something about it if they have to.

 

[nanofrog]

How do you do that? I'm clueless about switching cores on or off in OS X. I would love to do this when I run EyeTV in standby or recording stuff. Most of the time My MP4,1 is just doing that and I don't need the fan noise and the waste of power.

Switching cores on and off. I figured it was just me that wasn't aware of it, as I don't spend lots of time in OS X. 99% of the time I'm in Windows though.

 

[nanofrog]

I just got a response about the i5 iMac (here).

There's still no idle temp on this system yet, so it's hard to say what the differential is.

Ok ran exactly as you asked for over 20 mins this is what I got.

With all running at the same time:
cpu usage never went over 7%
Temp steady at 47 to 50 celsius
Fan 938 rpm
Itunes on its own uses 23% of the total cpu usage
Crazy heavy flash no effect

Just incase I wasn't understanding correctly even run on their own no apparent change in temps or cpu usage.

 

[smacman]

Interesting. So like you say it's a little early to come to a conclusion but it looks like the iMacs either don't have this issue, or they are more efficient at cooling the cpu.

 

[DoFoT9]

Interesting. So like you say it's a little early to come to a conclusion but it looks like the iMacs either don't have this issue, or they are more efficient at cooling the cpu.

or it only effects the i7 and not the i5?

 

[DoFoT9]

from the same thread nano linked to before, the OP has stated the i5 idles around 30°C. she also reported playing iTunes music results in the computer hitting ~50°C.

this doesnt look good.

edit: so after discussing it a tiny bit further it seems that the temperature did not go above ~38°C whilst playing both video and audio on the iMac. i am presuming that 38°C was recorded when watching videos. better then the 50°C+ you MP users are seeing!

so what on earth is the cause!?

 

[smacman]

from the same thread nano linked to before, the OP has stated the i5 idles around 30°C. she also reported playing iTunes music results in the computer hitting ~50°C.

this doesnt look good.

Hmmmmmm.. Very interesting...

 

[DoFoT9]

Hmmmmmm.. Very interesting... The fact that it stabilizes at 50ish is somewhat positive. I would have thought it would cook itself more so than a MP. I can only hope that if it is indeed going to be a problem for the iMac clan as well, that a fix will come sooner.

please read my post edit.

 

[smacman]

So based on the thread in the iMac forum, the quad core iMacs are NOT having the same issue. They are purring along nicely at 30C while playing audio. Now this gets even more perplexing!

 

[nanofrog]

Interesting. So like you say it's a little early to come to a conclusion but it looks like the iMacs either don't have this issue, or they are more efficient at cooling the cpu.

It's definitely too early yet. We need more i5 and i7 iMac testers to get an accurate idea of what's going on with those systems.

or it only effects the i7 and not the i5?

No way to know yet.

There are some differences between the 3 CPU's, but the biggest, is the i5-750 and i7-860 are connected to a P55 chipset via DMI, not QPI (as the LGA1366 parts use). There are others differences as well, related to features such as HT, VT-d,... But most, if not all of them aren't part of audio or video decoding (save perhaps HT, and it's still part of the controller).

Note that the older Intel MP's use DMI as the connection methodology to the chipsets as well. But also keep in mind, that under the Windows side, we're not seeing the same high temps for the decoding aspect of audio or video (what the CPU has to deal with).

That still lends me to think it's the software, not the hardware, as if hardware alone were the cause, it would be prevalent under Windows as well on the same set (or similar architecture; i.e. QPI used) of CPU's used in MP's.

from the same thread nano linked to before, the OP has stated the i5 idles around 30°C. she also reported playing iTunes music results in the computer hitting ~50°C.

this doesnt look good.

edit: so after discussing it a tiny bit further it seems that the temperature did not go above ~38°C whilst playing both video and audio on the iMac. i am presuming that 38°C was recorded when watching videos. better then the 50°C+ you MP users are seeing!

so what on earth is the cause!?

I saw the posts, and the information could be a bit confusing. At first it seems the problem exists (idle = 30C, testing = 50C @ low CPU %), the second (where you made the edit from), seems to refute that (idle and test temps only differed by 8C, which is quite normal IMO, going by the Windows side).

We need more information, and we'll have to be patient (especially waiting for the i7 based systems to reach eager hands, and get through the "Geekbench" phase ).

 

[DoFoT9]

There are some differences between the 3 CPU's, but the biggest, is the i5-750 and i7-860 are connected to a P55 chipset via DMI, not QPI (as the LGA1366 parts use). There are others differences as well, related to features such as HT, VT-d,... But most, if not all of them aren't part of audio or video decoding (save perhaps HT, and it's still part of the controller).

Note that the older Intel MP's use DMI as the connection methodology to the chipsets as well. But also keep in mind, that under the Windows side, we're not seeing the same high temps for the decoding aspect of audio or video (what the CPU has to deal with).

That still lends me to think it's the software, not the hardware, as if hardware alone were the cause, it would be prevalent under Windows as well on the same set (or similar architecture; i.e. QPI used) of CPU's used in MP's.

yes, it seems to make sense that it is some soft of software based issue. the 750 has no HT as we all know. still waiting for a reply on the i7 thread to see if there is the itunes "problem".

I saw the posts, and the information could be a bit confusing. At first it seems the problem exists (idle = 30C, testing = 50C @ low CPU %), the second (where you made the edit from), seems to refute that (idle and test temps only differed by 8C, which is quite normal IMO, going by the Windows side).

yup originally it didnt look good, but then she said 8°C was the max deviation, very reasonable! so it doesnt seem to be a problem for the i5's at the moment. we cant rule this out on the i7's yet though as they have HT come into play.

We need more information, and we'll have to be patient (especially waiting for the i7 based systems to reach eager hands, and get through the "Geekbench" phase ).

the geekbench are quite reasonable, i7's get ~9500 in 64-bit mode compared to ~8500 in 32-bit mode, representing about half the total power of the higher end MP. im going to be happy with my i7

 

[nanofrog]

yes, it seems to make sense that it is some soft of software based issue. the 750 has no HT as we all know. still waiting for a reply on the i7 thread to see if there is the itunes "problem".

It could be HT, or something else. There's just not enough info to narrow it down yet.

yup originally it didnt look good, but then she said 8°C was the max deviation, very reasonable! so it doesnt seem to be a problem for the i5's at the moment. we cant rule this out on the i7's yet though as they have HT come into play.

We can't base it on one person's report though. We need a larger sample size, as it's possible the testing wasn't carried out in a manner to generate the results we actually need.

the geekbench are quite reasonable, i7's get ~9500 in 64-bit mode compared to ~8500 in 32-bit mode, representing about half the total power of the higher end MP. im going to be happy with my i7

I'm not the biggest fan of Geekbench, but the comment was to do with where the interest currently lies. Our issue is back-burner yet in comparison, and will possibly continue to be that way (many other topics to pop up, such as "Is it 10.6.2 on the system?" before they really get into the systems). I expect it will repeat when the i7 based units start to show up.

So we'll have to be patient. Not only for more to get their hands on them (including the i7's), but to get past the initial topics. It was the same with the MP's too, but not to the same extent, as it's a smaller segment, and MP owners are really intersted in the "nuts-and-bolts" operation of it. Different markets = different interstests it seems. I.e. how fast does the "What temps do you get?" threads in the MP side.

 

[DoFoT9]

It could be HT, or something else. There's just not enough info to narrow it down yet.

We can't base it on one person's report though. We need a larger sample size, as it's possible the testing wasn't carried out in a manner to generate the results we actually need.

true, looks like ill just have to order mine so that we can get some solid numbers haha.

I'm not the biggest fan of Geekbench, but the comment was to do with where the interest currently lies. Our issue is back-burner yet in comparison, and will possibly continue to be that way (many other topics to pop up, such as "Is it 10.6.2 on the system?" before they really get into the systems). I expect it will repeat when the i7 based units start to show up.

geekbench does seem over-rated, it doesnt test efficiently. we need a sticky that says "it comes with 10.6.2". there, solved! who gives a rat!?

So we'll have to be patient. Not only for more to get their hands on them (including the i7's), but to get past the initial topics. It was the same with the MP's too, but not to the same extent, as it's a smaller segment, and MP owners are really intersted in the "nuts-and-bolts" operation of it. Different markets = different interstests it seems. I.e. how fast does the "What temps do you get?" threads in the MP side.

yup the MP owners i guess are more technical and would have already put in their research for results etc.

update: interesting result.. CPU sits at ~47°C with just itunes. computer had been running for a while. i7

 

[nanofrog]

true, looks like ill just have to order mine so that we can get some solid numbers haha.

I thought you were going to stick with a hack. Makes more sense given your financial situation as a student, and have the ability to do it (time + previous experience). You also get the benefit of upgrading hardware you can't on an iMac.

But if you go with the iMac, I'll hold you to playing guinea pig with testing.

yup the MP owners i guess are more technical and would have already put in their research for results etc.

update: interesting result.. CPU sits at ~47°C with just itunes. computer had been running for a while. i7

It has more of a practical reasoning behind it (well, some of it anyway ).

 

[DoFoT9]

I thought you were going to stick with a hack. Makes more sense given your financial situation as a student, and have the ability to do it (time + previous experience). You also get the benefit of upgrading hardware you can't on an iMac.

well, there are a number of considerations that i included with my decision. the fact that the imac has an incredible IPS 27" panel is pretty much the breaking deal for me. to build an equivalent hackintosh with such a good screen is basically the same price. sure the hack can be OC'd and whatnot, but the iMac can have its CPU and GPU upgraded - a MR user has reported that its possible (GPU would be expensive to do).

other factors include power, desk space requirements and ease of configuration (hack would require me to download a hell of a lot of stuff and still wouldnt be guaranteed to work). the hack might fail after an upgrade to 10.6.3. who knows.

that was my main reasoning i guess.

But if you go with the iMac, I'll hold you to playing guinea pig with testing.

hmm id rather be a rabbit. more of a pest

 

[nanofrog]

well, there are a number of considerations that i included with my decision. the fact that the imac has an incredible IPS 27" panel is pretty much the breaking deal for me. to build an equivalent hackintosh with such a good screen is basically the same price. sure the hack can be OC'd and whatnot, but the iMac can have its CPU and GPU upgraded - a MR user has reported that its possible (GPU would be expensive to do).

other factors include power, desk space requirements and ease of configuration (hack would require me to download a hell of a lot of stuff and still wouldnt be guaranteed to work). the hack might fail after an upgrade to 10.6.3. who knows.

that was my main reasoning i guess.

What a lazy bum!

hmm id rather be a rabbit. more of a pest

No you wouldn't.

My recollection of rabbits has to do with an old medical experiment while in college. We called them "The Humpback Rabbits". Think white bunny with a silcone breast implant on it's back. The part I dealt with was getting the bunny back in liquid form (various organs) for testing in an FTIR.

 

[DoFoT9]

What a lazy bum!

haha thanks for the positive reinforcement of my selection!

No you wouldn't.

My recollection of rabbits has to do with an old medical experiment while in college. We called them "The Humpback Rabbits". Think white bunny with a silcone breast implant on it's back. The part I dealt with was getting the bunny back in liquid form (various organs) for testing in an FTIR.

:| that sounds uhhh thrilling!

 

[nanofrog]

haha thanks for the positive reinforcement of my selection!

Nah. I figured you didn't really need that big a screen, and even more so on the financial end. Budgeting $2k - 2.2k seems a bit tough from other posts and emails.

:| that sounds uhhh thrilling!

It wasn't that bad really. I like animals, but it was medical research. Justifiable I guess when comparing a non-endangered species vs. a human for testing. Financially speaking, the critters are cheaper, and gets the job done (human volunteers are hard to come by with consistency as I understand it).

 

[DoFoT9]

Nah. I figured you didn't really need that big a screen, and even more so on the financial end. Budgeting $2k - 2.2k seems a bit tough from other posts and emails.

ssshh you will give away our status

i priced up a hack, it ended up being about $2.7k aus. the i7 imac is $2.9k aus including 3 year warranty (another thing the hack wont have).

financial status has changed slightly, i will be selling shares that i didnt really want to sell before (they were my savings for the long term HA). i have 3 more exams and a few jobs have presented themselves so i think i may be able to get my savings back if i work hard over the holidays. (email me if you want deets haha )

It wasn't that bad really. I like animals, but it was medical research. Justifiable I guess when comparing a non-endangered species vs. a human for testing. Financially speaking, the critters are cheaper, and gets the job done (human volunteers are hard to come by with consistency as I understand it).

if it gives out money, i will do the tests! i didnt know you also did the medical side of things! you do everything!

 

[nanofrog]

ssshh you will give away our status

i priced up a hack, it ended up being about $2.7k aus. the i7 imac is $2.9k aus including 3 year warranty (another thing the hack wont have).

financial status has changed slightly, i will be selling shares that i didnt really want to sell before (they were my savings for the long term HA). i have 3 more exams and a few jobs have presented themselves so i think i may be able to get my savings back if i work hard over the holidays. (email me if you want deets haha )

Using xe.com for conversion, $2.9k AUD = $2,683.65 USD

I know prices are higher there (sticker and the taxes on top of it), but I presumed it's essentially the 27" IPS display (or similar), but what's going on?

if it gives out money, i will do the tests! i didnt know you also did the medical side of things! you do everything!

It was a Uni job while a student in the Materials Eng. Dept.. They used students for low cost technical personnel, we got experience as well as a few bucks to help cover costs.

 

[DoFoT9]

Using xe.com for conversion, $2.9k AUD = $2,683.65 USD

how does that compare to a similarly specc'd US imac? (i7, 1TB, 4GB RAM, education discount and full warranty are all the changed aspects)

I know prices are higher there (sticker and the taxes on top of it), but I presumed it's essentially the 27" IPS display (or similar), but what's going on?

tariff on IT coming into australia sucks. the ISP panel would be a good $1k i reckon. i donno where all the rest of it comes from. CPU/RAM/HD = ~$600 all up. GPU would be a good $500 (its the mobility GPU i think).

It was a Uni job while a student in the Materials Eng. Dept.. They used students for low cost technical personnel, we got experience as well as a few bucks to help cover costs.

HA! payment methods for helping others in our uni generally revolve around beer, or cartons of it.