MacPro4,1 w/5,1FW 2 x x5675 strange temperatures

[tpluth]

Ever since I upgraded my MP4,1 from dual 2.26QC to dual x5675 hex cores I've seen strange temperature readings. At idle, I'm seeing 65c - 70c on CPUA and 70c - 75c on CPUB, at idle, but when I load it up (GB3, for example) the temperatures will drop down into the 40's, 50's and low 60's. Once the load subsides, they slowly return to the higher temperatures.

I've pulled the heatsinks and re-doped the CPU's with Arctic Cool MX4 (one thin stripe down the center of the IHS) and it didn't really change anything. Fans are running at lowest speeds and don't really even speed up during the stress test.

Any ideas or should I just not worry about it?

System - MP4,1 wih 5,1 firmware, dual x5675, 24GB DDR3 10600R, Powercolor R7 250, WD blue 1TB, WD green 2TB, WD red 3TB, 4x240GB SSD in RAID0 on 2 88SE9230 x2 controllers, Yosemite 10.10.1

 

[h9826790]

Can you provide some screenshots?

 

[tpluth]

Can you provide some screenshots?

Idle

GB3 Started

GB3

More GB3

1 minute after completed

 

[Denis Ahrens]

Hi

I assume that the tool simply uses the wrong smc key. I made changes to several smc utilities like smcFanControl, HWSensor and OSXFanControl.

the correct keys are ICAC and ICBC for cpu A and B.

What you see is simply the difference to 104C which is the point where the cpu would begin to throttle down.

Denis

PS: smcFanControl used the right key in older versions but they changed to something wrong in the latest version. HWSensors has my changes on github, but there is no new binary package yet.

 

[InsertCatchyNic]

The firmware in the SMC does not get updated when you run the 4,1 to 5,1 hack. At least the ars technica artile I read this morning on the FW upgrade stated as much

So you can't rely on your 4,1 modified machine to keep things running cool. So you will probably have to use a secondary controller to keep your machine running at reasonable temperatures.

I actually do this on my 3,1 and just modify the fan speeds via iStats.

 

[shindom]

Idle
View attachment 515188
GB3 Started
View attachment 515189
GB3
View attachment 515190
More GB3
View attachment 515191
1 minute after completed
View attachment 515192

I see absolutely nothing wrong with your temps .. they actually look ideal ..I am quite jealous of your Northbridge temp (mine is at 76C).

It seems you are reading the temps wrong for your processors.
The temp CPU A core from PCECI (and respectively CPU B) is actually the difference between the core temp and the max allowed for the processor.
So the higher this number, the further you are from the max, the better.

This is why you see it decrease during a Geekbench test as the actual temps of the proc are rising and getting closer to their max. Does that make sense ?

Your temps are great !

 

[tpluth]

I see absolutely nothing wrong with your temps .. they actually look ideal ..I am quite jealous of your Northbridge temp (mine is at 76C).

It seems you are reading the temps wrong for your processors.
The temp CPU A core from PCECI (and respectively CPU B) is actually the difference between the core temp and the max allowed for the processor.
So the higher this number, the further you are from the max, the better.

This is why you see it decrease during a Geekbench test as the actual temps of the proc are rising and getting closer to their max. Does that make sense ?

Your temps are great !

Hopefully, the developers of Macs Fan Control and/or HWMonitor will change their programs to correctly interpret the values in the future.

My northbridge was around 72c until I blew the dust out of it...

 

[h9826790]

I am quite jealous of your Northbridge temp (mine is at 76C).

If you increase your booster fan speed to 1100, you should have that Northbridge temp as well.

 

[Spacedust]

Please see my results from Mac Pro 4,1@5,1 with Apple branded 2x X5570's 2.9 GHz (top 2009 model) when idle. I'm using a bit lower speeds on PCI and PS fans because I'm using GTX680 which makes them noisy.