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

jsurpless

macrumors member
Original poster
Aug 6, 2009
88
0
Hi everyone

I'd really like to get some thoughts on this, particularly interested in what others who have the same machine (Mac Pro 1,1 @ 2.66 Xeon 5150) are experiencing...

When I first bought my Mac Pro, I was so ecstatic at how cool it ran (coming from a Quad 2.5 G5)... last year, I noticed it was running much warmer... ultimately, I tracked it down to the fact that the CPUs used to draw 7-10W each @ idle with OS X 10.5.2 or lower and JUMPS to 30W each @ idle with OS X 10.5.3 or later...

It seems to be purely a software thing because toggling back and forth between the two revisions (10.5.2 vs 10.5.3 or later) clearly affects the power draw...

I find this pretty frustrating because my CPUs never drop below 125F and this equates a LOT of extra heat being exhausted, as well as unnecessary electric cost...

Several others on this forum have said that their machines still idle at the lower power use, despite running even Snow Leopard... the only conclusion I can draw is that something is wrong with my machine? but what?

Any thoughts?

-Justin
 
You could try reapplying the thermal paste with something better.

That's what I thought might be the issue when I first noticed that the temps were so high... had a service shop look into that...

When I got it back, I stumbled on the power draw... that's why the temps are so high, because the CPU is drawing the same power at idle, as it does when under load... and THAT is not right, I think... no?
 
Did you try SMC reset already? It's first to do when you experiencing power related issues.
 
Did you try SMC reset already? It's first to do when you experiencing power related issues.

Pretty sure I'd done this in the past but I just did it again... unfortunately, nothing... still drawing the crazy amount of power per CPU...

Thanks anyway

-Justin
 
One Mac Pro 1,1 owner still using 5150s said his power draw was much lower...

What about others?
 
Take a look at these two screenshots... one is under 10.5.2 and you can clearly see how much LOWER the power draw and consequently the temps are... the other is 10.6.4...

One thing I just noticed is that under 10.5.2, the CPU frequency periodically drops to 2000 (is this SpeedStep?)... whereas under 10.6.4, it's constant at 2667... I can't imagine this doesn't have an impact, although it's so short...

Thoughts?
 

Attachments

  • OS_X_10.5.2_Sensors.jpg
    OS_X_10.5.2_Sensors.jpg
    688.5 KB · Views: 243
  • OS_X_10.6.4_Sensors.jpg
    OS_X_10.6.4_Sensors.jpg
    685.6 KB · Views: 213
Take a look at these two screenshots... one is under 10.5.2 and you can clearly see how much LOWER the power draw and consequently the temps are... the other is 10.6.4...

One thing I just noticed is that under 10.5.2, the CPU frequency periodically drops to 2000 (is this SpeedStep?)... whereas under 10.6.4, it's constant at 2667... I can't imagine this doesn't have an impact, although it's so short...

Thoughts?

Have you measured the power draw with a power meter or something? I don't really trust software in these kinds of things, it can simply be a bug or something
 
Have you measured the power draw with a power meter or something? I don't really trust software in these kinds of things, it can simply be a bug or something

I have not - I'll have to see about it at work on Monday... though, one question - where would I measure the draw?

I understand the concern over a possible bug but I don't think it's the case here... the air coming out of MP is noticeably warmer in >= 10.5.3... in fact, that's what first drew my attention to it...
 
I have not - I'll have to see about it at work on Monday... though, one question - where would I measure the draw?

I understand the concern over a possible bug but I don't think it's the case here... the air coming out of MP is noticeably warmer in >= 10.5.3... in fact, that's what first drew my attention to it...

Apple says it shouldn't consume more than 171W when idle. You can't measure anything else but the total system's power consumption, you need something like this (make sure the plug is right as some countries have their own). As I said earlier, you should try reapplying the thermal paste, it should at least cool it down
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
This has echos of what the 2009 Mac Pro went through regarding the silly power draw at idle.

Can you do benchmarks in 10.5.2 and 10.6.4 to see if there is any change?

The CPU is either no being put into a full idle state - or - it's being used for something and thus not going into a fully idle state.
 
Apple says it shouldn't consume more than 171W when idle. You can't measure anything else but the total system's power consumption, you need something like this (make sure the plug is right as some countries have their own). As I said earlier, you should try reapplying the thermal paste, it should at least cool it down

Bear in mind Apple's test was made with only 1GB of RAM (2 modules). The more modules you have on that Mac Pro, the more it will consume. Mine with 8 modules overs around 200w give or take 10w.
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Apple says it shouldn't consume more than 171W when idle. You can't measure anything else but the total system's power consumption, you need something like this (make sure the plug is right as some countries have their own). As I said earlier, you should try reapplying the thermal paste, it should at least cool it down

Thanks for the link on the power usage monitor...

As for the thermal paste - a service tech removed the CPUs and re-applied some thermal paste last May... didn't have any impact...
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
This has echos of what the 2009 Mac Pro went through regarding the silly power draw at idle.

Can you do benchmarks in 10.5.2 and 10.6.4 to see if there is any change?

The CPU is either no being put into a full idle state - or - it's being used for something and thus not going into a fully idle state.

By benchmarks - any particular program you recommend?

I've given thought to the CPU being used for something but Activity Monitor doesn't show any use - the load is certainly very near to 0%...
 
Bear in mind Apple's test was made with only 1GB of RAM (2 modules). The more modules you have on that Mac Pro, the more it will consume. Mine with 8 modules overs around 200w give or take 10w.

Definitely something to keep in mind, but that would only affect the power drawn by the memory banks... Hardware Monitor differentiates, as you can see in the images I uploaded - only CPU power varies between them...
 
Stupid question: do you use that same version of HW Monitor with Leo and SL?
Try iStat - recent version - to confirm those reads. Maybe it's only software bug/incompatility like Hellhammer did mention...
 
Stupid question: do you use that same version of HW Monitor with Leo and SL?
Try iStat - recent version - to confirm those reads. Maybe it's only software bug/incompatility like Hellhammer did mention...

Good question but yes, I am using the same version - 4.7 between both...

Tried iStat Pro - apparently, it's only capable of monitoring temps and fan speeds (and only the CPU heatsinks at that) so I can't be certain... however, the CPU heatsink temps are within 1-2 degrees...
 
I have a Mac Pro 1,1 purchased in August 2006.
It's currently running 10.6.4 and is fully updated.
Hardware wise mine currently has 2 x 2.66GHz Xeon 5150's, 4 hard drives, an XFX ATi 4870 1GB graphics card, 6GB RAM (2x2GB + 4x512MB, as my other 2x2GB has just gone bad), dual SATA optical drives and Bluetooth + WiFi.

I haven't noticed any difference in temperatures or the power draw over the life of my machine, except when I upgraded the hardware such as the graphics card.

I've run Hardware Monitor (version 4.9), and here are my results. I haven't modified the fan speed or changed any settings regarding the hardware. The only reading which you can't see on that which go off the bottom are: Power Supply Fan = 599 RPM, Graphics Processor = ~5%, CPU Clock Frequency = 2667 MHz.

Oh also the state of the machine when those reading were taken was practically idle, though with lots of programs open.
 

Attachments

  • temps.png
    temps.png
    202 KB · Views: 122
By benchmarks - any particular program you recommend?

I've given thought to the CPU being used for something but Activity Monitor doesn't show any use - the load is certainly very near to 0%...


Anything, really. XBench showed the problem on the 2009 Mac Pro.

Make sure NOTHING is running in the background, so no MSN, iTunes, etc.

[On a side note, this is at complete idle, no apps running (bar hardware monitor?)]
 
I have a Mac Pro 1,1 purchased in August 2006.
It's currently running 10.6.4 and is fully updated.
Hardware wise mine currently has 2 x 2.66GHz Xeon 5150's, 4 hard drives, an XFX ATi 4870 1GB graphics card, 6GB RAM (2x2GB + 4x512MB, as my other 2x2GB has just gone bad), dual SATA optical drives and Bluetooth + WiFi.

I haven't noticed any difference in temperatures or the power draw over the life of my machine, except when I upgraded the hardware such as the graphics card.

I've run Hardware Monitor (version 4.9), and here are my results. I haven't modified the fan speed or changed any settings regarding the hardware. The only reading which you can't see on that which go off the bottom are: Power Supply Fan = 599 RPM, Graphics Processor = ~5%, CPU Clock Frequency = 2667 MHz.

Oh also the state of the machine when those reading were taken was practically idle, though with lots of programs open.

Alfihar - thanks for posting this... much appreciated... your CPU power readings are right where mine are supposed to be under 10.6.4 (and are under 10.5.2)... so it's definitely indicating to me that something is amiss with mine...

Now - what I wonder is what could survive between OS installations, even to a brand-new user directory...
 
Anything, really. XBench showed the problem on the 2009 Mac Pro.

Make sure NOTHING is running in the background, so no MSN, iTunes, etc.

[On a side note, this is at complete idle, no apps running (bar hardware monitor?)]

Not sure exactly what I'm supposed to be looking for but here are the results...
 

Attachments

  • XBench_Results.zip
    6.9 KB · Views: 57
On a whim, I decided to replace the 10.6.4 'AppleIntelCPUPowerManagement' extension (making a backup first) with the corresponding one from 10.5.2... I also removed the 'AppleIntelCPUPowerManagementClient' extension and then rebooted...

Upon reboot, HW Monitor revealed that my CPU power draw was now where it was supposed to be, down around 10W per CPU...

My question now falls to Alfihar - can you check what version of the 'AppleIntelCPUPowerManagement' extension you have under 10.6.4? Mine is '105.10.0' vs '9.33.0' under 10.5.2...

****EDITED****

OK, now I am even more confused... in a further attempt to isolate the problem, I restored the 10.6.4 version of 'AppleIntelCPUPowerManagement' and rebooted... power usage remained low... so then I was like, hmmm... I guess it's the other extension... so I restored that one too... power usage REMAINS LOW?!?!?!

In short, I copied 10.5.2 extensions, rebooted, problem went away, restored machine to original 10.6.4 extensions AND problem remains away?!?!?

What in the world?!?! Thoughts?
 
Well I have version '105.10.0' of the extension.

I wonder if changing the extensions caused it to rebuild the extensions, thus correcting whatever the problem was.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.