High PCH die temperature (Mac Pro 4,1)

[btjtaylor]

Hi all

I've recently bought a Mac Pro 4,1 and done the usual upgrades:

Mac Pro 4,1 with 5,1 firmware
2x Intel X5680 CPUs
48GB 1333MHz RAM
1TB Crucial MX500 SATA SSD
PowerColor AMD RX580 graphics card
KT4004 USB3.0 controller

The system is running great and is totally stable. However, I noticed the fan noise was quite loud when idle on the desktop.

I installed HWMonitor and checked the temps, it's telling me that the PCH Die temperature is 86c and the Booster B fan is running at 5000rpm which must be what's causing the noise. I did some reading and found some instances where people had a broken clip holding the heatsink onto the southbridge (the chip between the CPU sockets) which I guess is the same as the PCH die? I removed the CPU board checked and my clips are still holding the heatsink down, so I applied some fresh thermal paste, but the temperature is still high. The other temps are OK, for example the CPUs are idling at around 50c

Does anyone have any other ideas I could try to attempt to resolve this issue? Does anyone else have this problem?

Thanks in advance

 

[h9826790]

Hi all

I've recently bought a Mac Pro 4,1 and done the usual upgrades:

Mac Pro 4,1 with 5,1 firmware
2x Intel X5680 CPUs
48GB 1333MHz RAM
1TB Crucial MX500 SATA SSD
PowerColor AMD RX580 graphics card
KT4004 USB3.0 controller

The system is running great and is totally stable. However, I noticed the fan noise was quite loud when idle on the desktop.

I installed HWMonitor and checked the temps, it's telling me that the PCH Die temperature is 86c and the Booster B fan is running at 5000rpm which must be what's causing the noise. I did some reading and found some instances where people had a broken clip holding the heatsink onto the southbridge (the chip between the CPU sockets) which I guess is the same as the PCH die? I removed the CPU board checked and my clips are still holding the heatsink down, so I applied some fresh thermal paste, but the temperature is still high. The other temps are OK, for example the CPUs are idling at around 50c

Does anyone have any other ideas I could try to attempt to resolve this issue? Does anyone else have this problem?

Thanks in advance

Correct, PCH is the Northbridge.

86C is hot. If you already re-apply thermal paste and nothing change, that sounds a bit strange to me.

You may upload the screen capture of the HWMonitor readings so that we can have better big picture to help.

And I assume you already perform the PRAM and SMC reset, but nil help.

 

[btjtaylor]

Correct, PCH is the Northbridge.

86C is hot. If you already re-apply thermal paste and nothing change, that sounds a bit strange to me.

You may upload the screen capture of the HWMonitor readings so that we can have better big picture to help.

And I assume you already perform the PRAM and SMC reset, but nil help.

Just tried the PRAM and SMC reset now and it doesnt help

 

[Spacedust]

Can you check in Macs Fan Control?

 

[h9826790]

Just tried the PRAM and SMC reset now and it doesnt help

If you manually spin up the exhaust fan to 1100RPM, will the NB temperature and Booster B RPM go down?

 

[btjtaylor]

I installed "Macs Fan Control" and it said the control for BOOSTA was "Unknown!". So I set this to auto and boost A spun up to 2500rpm and now the PCH is 73c! I'm going to try to reboot but I think this issue is solved ... I have no idea what happened there, maybe someone in the past was messing with it? Or maybe I didn't clear the SMC properly?
[doublepost=1532726002][/doublepost]Problem solved I rebooted and left the machine idling on the desktop and the temps/fan speeds are now under control

Thanks for pointing me in the right direction!

 

[handheldgames]

I installed "Macs Fan Control" and it said the control for BOOSTA was "Unknown!". So I set this to auto and boost A spun up to 2500rpm and now the PCH is 73c! I'm going to try to reboot but I think this issue is solved ... I have no idea what happened there, maybe someone in the past was messing with it? Or maybe I didn't clear the SMC properly?
[doublepost=1532726002][/doublepost]Problem solved I rebooted and left the machine idling on the desktop and the temps/fan speeds are now under control

Thanks for pointing me in the right direction!

Your RAM slot temps are CRAZY high. 129c / 264f!! Yikes!!

 

[btjtaylor]

Your RAM slot temps are CRAZY high. 129c / 264f!! Yikes!!

There's no module in that slot (triple channel config 6x 8GB sticks) so it must be an error!

 

[h9826790]

Correct, that 129C is the default number when no reading available. Same as the GPU's 85C. That's also a pre-defined number.

 

[verdejt]

Funny thing I was just wondering about my temps and installed HWMonitor and mine show the same thing. I installed SMC fan control and ramped up my fan speeds as currently my office is a little warm and the temps came back down real quick. I figure it's about 78 in my office and not much air flow. I know my iMac that was in here always ran hot.

 

[handheldgames]

Is there a consensus to the expected temperature spread between IOH Diode and IOH Heatsink?

I'm currently seeing a 13 degree variance between the two IOH temps.

IOH HeatSink 37
IOH TDiode 50

 

[h9826790]

Is there a consensus to the expected temperature spread between IOH Diode and IOH Heatsink?

I'm currently seeing a 13 degree variance between the two IOH temps.

IOH HeatSink 37
IOH TDiode 50

I don't know any exact number, but 13 ∆°C seems at the normal range. For your reference, this is my temperature

 

[verdejt]

If I set my fans to normal speed my PCH Die temp climbs up pretty quickly. I set my fans up just a notch and the temp comes right back down. I bought my MacPro used and it's a 2010, as a precaution should I replace all the thermal paste in the processors?

 

[handheldgames]

If I set my fans to normal speed my PCH Die temp climbs up pretty quickly. I set my fans up just a notch and the temp comes right back down. I bought my MacPro used and it's a 2010, as a precaution should I replace all the thermal paste in the processors?

With a single core X5677 and Innovation Cooling Solutions Graphite reusable thermal interface material I’m seeing:

Idle, 33 diode / 31 heatsink
Under 100% extended load, 45 diode / 42 heatsink

If you have a large variance between cpu diode and cpu heatsink or you want to improve thermal transfer to the copper heat pipe in the cMP heatsink, this new material is legdit and easy to work with.

I’m using a 30mmx30mm graphite pad and it’s a bit too small for my liking. Whilst a 40mm was delivered today to replace it, Amazon damaged it by packing it with a heavy item.

 

[h9826790]

With a single core X5677 and Innovation Cooling Solutions Graphite reusable thermal interface material I’m seeing:

Idle, 33 diode / 31 heatsink
Under 100% extended load, 45 diode / 42 heatsink

If you have a large variance between cpu diode and cpu heatsink or you want to improve thermal transfer to the copper heat pipe in the cMP heatsink, this new material is legdit and easy to work with.

I’m using a 30mmx30mm graphite pad and it’s a bit too small for my liking. Whilst a 40mm was delivered today to replace it, Amazon damaged it by packing it with a heavy item.

I think you better state your fan speed and ambient temperature as well. 45C under 100% loading is exceptionally low.

 

[handheldgames]

I think you better state your fan speed and ambient temperature as well. 45C under 100% loading is exceptionally low.

INTAKE & EXHAUST are based on DIMM Proximity 3. 30c / 45c setting in MacsFanControl.
In a 100% CPU load LuxMark CPU Stress Test, fan speed is under 950 RPM.
These two fans are responsible for Cool air intake and HOT Air exhaust. Running them too slow leads to heat buildup in RAM, CPU and Northbridge. Heat also slows RAMDisk performance.

BOOSTA is based on CPU Diode A. 30c / 52c setting in MacsFanControl
In a 100% CPU load LuxMark CPU Stress Test, fan tops out at 2259 at 45c diode 41c heatsink
Idling at 32c diode / 31c heatsink after the stress test.

 

[MriX]

INTAKE & EXHAUST are based on DIMM Proximity 3. 30c / 45c setting in MacsFanControl.
In a 100% CPU load LuxMark CPU Stress Test, fan speed is under 950 RPM.
These two fans are responsible for Cool air intake and HOT Air exhaust. Running them too slow leads to heat buildup in RAM, CPU and Northbridge. Heat also slows RAMDisk performance.

BOOSTA is based on CPU Diode A. 30c / 52c setting in MacsFanControl
In a 100% CPU load LuxMark CPU Stress Test, fan tops out at 2259 at 45c diode 41c heatsink
Idling at 32c diode / 31c heatsink after the stress test.

View attachment 773250

What was the Northbridge temperature for you before upgrading to the Graphite thermal pad from Innovation Cooling?

I use good thermal paste and my chip runs at 73 degrees Celsius with the stock cooler and default fan speeds.

 

[handheldgames]

What was the Northbridge temperature for you before upgrading to the Graphite thermal pad from Innovation Cooling?

I use good thermal paste and my chip runs at 73 degrees Celsius with the stock cooler and default fan speeds.

On stock fan settings, my single CPU x58 IOH idles at 56c DIODE and 41c Heatsink. 73c seems high for a single or dual cpu setup.

My single CPU 2009's x58 Northbridge is still running with the original thermal paste. Whilst I had considered going though the effort required to replace the thermal paste on the Northbridge, IOH DIODE temps are hovering < 50c while at load with Macs Fan Control, which apparently is a good temp.

If the IOH temps were too high, or the variance between it's heatsink and diode was out of the norm, I would have gone through the cleanup process.

 

[MriX]

On stock fan settings, my single CPU x58 IOH idles at 56c DIODE and 41c Heatsink. 73c seems high for a single or dual cpu setup.

My single CPU 2009's x58 Northbridge is still running with the original thermal paste. Whilst I had considered going though the effort required to replace the thermal paste on the Northbridge, IOH DIODE temps are hovering < 50c while at load with Macs Fan Control, which apparently is a good temp.

If the IOH temps were too high, or the variance between it's heatsink and diode was out of the norm, I would have gone through the cleanup process.

I have a MacPro4,1 upgraded to 5,1 with dual Xeon X5680 and I thought 70c at idle is pretty normal and not bad, because Intel states that the chip is specified for max. 106c.
Usually the heat sink runs 10c cooler than the chip itself.

So the thermal pad isn’t that much better for you than thermal paste? Otherwise I would consider upgrading also to that thermal pad

 

[handheldgames]

I have a MacPro4,1 upgraded to 5,1 with dual Xeon X5680 and I thought 70c at idle is pretty normal and not bad, because Intel states that the chip is specified for max. 106c.
Usually the heat sink runs 10c cooler than the chip itself.

So the thermal pad isn’t that much better for you than thermal paste? Otherwise I would consider upgrading also to that thermal pad

With the thermal pad installede with a x5677, I'm seeing better numbers than thermal paste at Idle and under load. There is also a notable decrease in the variance between CPU DIODE and HEATSINK indicating a better cooling with increased thermal transfer.

Pulling the x5677, replacing it with the 990x, it's idling at 32c diode 31c heatsink with the same thermal material. Under full load, 50c diode and 41c heatsink. Adding 2 cores increased temps by 5c on load. The 990x is running cooler on idle than the x5677.

As crazy as it sounds, you just may see better temps with this specific brand of thermal pads.

 

[h9826790]

What was the Northbridge temperature for you before upgrading to the Graphite thermal pad from Innovation Cooling?

I use good thermal paste and my chip runs at 73 degrees Celsius with the stock cooler and default fan speeds.

73C should be very normal for dual processor cMP.

His ambient was quite low, 26C. And fan speed is way higher than normal.

 

[handheldgames]

73C should be very normal for dual processor cMP.

His ambient was quite low, 26C. And fan speed is way higher than normal.

Thanks for sharing. My ambient was 28c. (look closer at the image)

At 100% load,

• Intake/Exhaust runs around 930, which is 330 rpm higher than stock speeds at idle. This does not seem like a major jump in fan speed.
• BOOTSTA runs at 2259, 56% of max fan speed @ 100% load.

What fan speed is considered normal for BOOSTA at 100% load?

For comparison, a non-actively cooled x58 chipset at 57c on a Gigabyte Mobo is running in the same room, sitting next to an AV receiver, on the top shelf / heat trap of a mostly closed closet, running a media center pc that's constantly pushing data through it. The ambient temp is > 30c

 

[h9826790]

Thanks for sharing. My ambient was 28c. (look closer at the image)

At 100% load,

• Intake/Exhaust runs around 930, which is 330 rpm higher than stock speeds at idle. This does not seem like a major jump in fan speed.
• BOOTSTA runs at 2259, 56% of max fan speed @ 100% load.

What fan speed is considered normal for BOOSTA at 100% load?

For comparison, a non-actively cooled x58 chipset at 57c running on a Gigabyte Mobo is running in the same room, sitting next to an AV receiver, on the top shelf / heat trap of a mostly closed closet.

28 was the PCIe ambient, system ambient was 26C, you room temperature will be even lower.

I agree that both intake / exhaust was just a bit above normal. But Booster was really way way above normal (even at 100% load)

Even I use my own fan profile (more conservative than the Apple native profile), with system ambient 30C, and keep the CPU temperature at around 78% at 100% load (85C for Apple native setting), my Booster fan still just around 1200 RPM.

 

[handheldgames]

28 was the PCIe ambient, system ambient was 26C, you room temperature will be even lower.

I agree that both intake / exhaust was just a bit above normal. But Booster was really way way above normal (even at 100% load)

Even I use my own fan profile (more conservative than the Apple native profile), with system ambient 30C, and keep the CPU temperature at around 78% at 100% load (85C for Apple native setting), my Booster fan still just around 1200 RPM.
View attachment 773422

Thanks for sharing highly respected specifics of your cMP Dual CPU install. The architecture of the dual processor board clearly presents cooling challenges for the 5500 chipset, cooling it with the air that's cooling the copper base of the heatpipe. CPU B is also effected by heat buildup from RAM, the chipset and CPU A. A slight increase to the base levels of intake and exhaust should provide excellent results with minimal sound effects.

FWIW.... The cpu bay / heatsink / chipset cooling design of the single core 4,1/5,1 apparently is the best air-based cooling design I've had or worked with.

 

[DPUser]

My Dual CPU 3.33 5,1 Northbridge typically runs bout 73 degrees Celsius, too.