CPU2 hotter than CPU1...by 13 degrees!

[TommyL]

Hi!

Just finished put together a homelab with Supermicro CSE-826 chassis, X9DRI-LNF4+ mainboard and 2x E5-2650v2. 64GB RDIMM.

I've been running memtest for some hours, and CPU2 is considerably hotter than CPU1. CPU1=40 degrees, CPU2=53 degrees.

Just wondering if this is normal, or should i try to re-seat the cooler on CPU2?

Regards

Tommy

 

[Monoman]

Does cpu1 partially vent into cpu2? I'd have to check the MB. I made a channel on mine out of gift cards so it no longer blows hot air into the I take on cpu2.

 

[kapone]

Seems normal.

 

[TommyL]

Does cpu1 partially vent into cpu2? I'd have to check the MB. I made a channel on mine out of gift cards so it no longer blows hot air into the I take on cpu2.

Yes it does.

Seems normal.

I just had a read around, and it seems theres many with the same issue. So i guess youre right.

 

[nthu9280]

If both have active heatsinks, you could also rotate the cpu2 heat sink by 180° and also change the fan direction to pull and exhaust to the rear.
Normal SM configuration is both push. So the fan on CPU2 heatsink is grabbing the warm air coming out of CPU1 heatsink and limited distance between the two.
Default:
Fan1 -> HSk1 -> Fan2-> HSK2 ->

Alt:
Fan1 -> HSk1 -> Hsk2-> Fan2 ->

This gives additional space for the air.

 

[TommyL]

If both have active heatsinks, you could also rotate the cpu2 heat sink by 180° and also change the fan direction to pull and exhaust to the rear.
Normal SM configuration is both push. So the fan on CPU2 heatsink is grabbing the warm air coming out of CPU1 heatsink and limited distance between the two.
Default:
Fan1 -> HSk1 -> Fan2-> HSK2 ->

Alt:
Fan1 -> HSk1 -> Hsk2-> Fan2 ->

This gives additional space for the air.

Both are passive, sorry for not stating that before. But if it turns out to be a problem later, i will certainly consider your suggestion.

 

[BLinux]

that's pretty typical since that board design stacks the airflow from CPU1 into CPU2. temps are more even if you had a X9DRD board:

on a X9DRD-LN4F+ with 2x E5-2680v2 that runs 24/7:

coretemp-isa-0000
Adapter: ISA adapter
Physical id 0:  +41.0°C  (high = +80.0°C, crit = +90.0°C)
Core 0:         +40.0°C  (high = +80.0°C, crit = +90.0°C)
Core 1:         +40.0°C  (high = +80.0°C, crit = +90.0°C)
Core 2:         +37.0°C  (high = +80.0°C, crit = +90.0°C)
Core 3:         +37.0°C  (high = +80.0°C, crit = +90.0°C)
Core 4:         +37.0°C  (high = +80.0°C, crit = +90.0°C)
Core 8:         +37.0°C  (high = +80.0°C, crit = +90.0°C)
Core 9:         +35.0°C  (high = +80.0°C, crit = +90.0°C)
Core 10:        +37.0°C  (high = +80.0°C, crit = +90.0°C)
Core 11:        +41.0°C  (high = +80.0°C, crit = +90.0°C)
Core 12:        +36.0°C  (high = +80.0°C, crit = +90.0°C)
coretemp-isa-0001
Adapter: ISA adapter
Physical id 1:  +43.0°C  (high = +80.0°C, crit = +90.0°C)
Core 0:         +39.0°C  (high = +80.0°C, crit = +90.0°C)
Core 1:         +38.0°C  (high = +80.0°C, crit = +90.0°C)
Core 2:         +40.0°C  (high = +80.0°C, crit = +90.0°C)
Core 3:         +41.0°C  (high = +80.0°C, crit = +90.0°C)
Core 4:         +39.0°C  (high = +80.0°C, crit = +90.0°C)
Core 8:         +39.0°C  (high = +80.0°C, crit = +90.0°C)
Core 9:         +36.0°C  (high = +80.0°C, crit = +90.0°C)
Core 10:        +40.0°C  (high = +80.0°C, crit = +90.0°C)
Core 11:        +40.0°C  (high = +80.0°C, crit = +90.0°C)
Core 12:        +42.0°C  (high = +80.0°C, crit = +90.0°C)

 

[Markess]

After reading a similar thread here last year, I've checked the difference in CPU temps on every dual CPU system I've worked on. Even in cases where the motherboard was in an open air test bench, CPU 2 has been hotter. Not by 13 degrees, but hotter, usually 3-6 degrees.

 

[TommyL]

After reading a similar thread here last year, I've checked the difference in CPU temps on every dual CPU system I've worked on. Even in cases where the motherboard was in an open air test bench, CPU 2 has been hotter. Not by 13 degrees, but hotter, usually 3-6 degrees.

3-6 degrees...ok. I might just re seat the cpu2 cooler since mine runs 12-13 degrees hotter then.

 

[vudu]

Is it possible the thermal compound or its application may have a bearing?

 

[TommyL]

Is it possible the thermal compound or its application may have a bearing?

Should be equal, but im gonna look at cpu2 when i come home. Maybe the seating is slightly skewed, or theres a problem with the cooler. Or i was too stingy on the compound on both

 

[Dawg10]

Swap the cpu's and see what the readings are.

 

[TommyL]

Swap the cpu's and see what the readings are.

Good suggestion. Will do.

 

[svtkobra7]

on a X9DRD-LN4F+ with 2x E5-2680v2 that runs 24/7:

A pair of 2680 v2 and 2690 v2 are relatively close in temps ... [OPEN SPOILERS]

Spoiler: 1 HOUR - RRDTOOL

Motherboard: X9DRH-7TF | CPUs: 2 x INTL Xeon E5-2690 v2 | Fans: CPU - 2 x SNK-P0048AP4

Motherboard: X9DRH-7TF | CPUs: 2 x INTL Xeon E5-2680 v2 | Fans: CPU - 2 x SNK-P0048AP4

Spoiler: 30 DAY - RRDTOOL

Motherboard: X9DRH-7TF | CPUs: 2 x INTL Xeon E5-2690 v2 | Fans: CPU - 2 x SNK-P0048AP4

Motherboard: X9DRH-7TF | CPUs: 2 x INTL Xeon E5-2680 v2 | Fans: CPU - 2 x SNK-P0048AP4

Spoiler: AIRFLOW

The only time I've only had a variance that "concerned" me - I should have been concerned. CPU2 was smokin hot and with good reason. Epic fail on the paste install. In my defense, I wanted to try applying in a square on the heatsink (not IHS) as the SNK-P0048AP4 ships from SMCI with TP already applied. That didn't work out so well as you can see =>

 

[BLinux]

@svtkobra7 your temps are often in the high 60C?

 

[svtkobra7]

@svtkobra7 your temps are often in the high 60C?

• In the 30 day view, you might infer the CPUs are pegged (at close to 100% utilization)
  
• I make those space heaters earn their keep!
  
• Marginal Expense to Mine [1] $ < Revenue From Mining $
• [1] (a) Electrical Cost @ 100% CPU Utilization less (b) Average Electrical Cost, w/o Mining

(While we are talking about an offset of pennies, I like the principal of being able to offset some of the infra cost)

O/c, there is the much more difficult to quantify expense to consider of the impact of running the rigs harder than need be ...