why do cpu 2 in lga 2011 run hotter than cpu 1?

wildpig1234

Well-Known Member
Aug 22, 2016
1,824
273
83
45
Anyone knows the reason why this is? it seems to be the case in both 2011 v1/v2 as well as v3....
 

wildpig1234

Well-Known Member
Aug 22, 2016
1,824
273
83
45
i don't think that's the whole answer to it.... the heatsink over the cpu 2 chipset is also hotter.
 

wildpig1234

Well-Known Member
Aug 22, 2016
1,824
273
83
45
i have a z9pe-d16 and a z10pe-d16ws and also intel c2600cp.... identical coolers on both sockets... cpu 2 is always hotter in all 3 set ups.... i actually run mine on the bench not inside a case....
 

Mishka

Member
Apr 30, 2017
90
18
8
London, UK
Pretty much all dual socket systems the second CPU will run warmer due to being further inside the case, further from the air intake etc.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,528
975
113
artofserver.com
i don't think that's the whole answer to it.... the heatsink over the cpu 2 chipset is also hotter.
i'm tending to agree with you...

Pretty much all dual socket systems the second CPU will run warmer due to being further inside the case, further from the air intake etc.
i don't think that's the whole answer... because:

1) on HP Z820 systems where they have air ducts to keep the "air channel" for each CPU separated, i've seen the same thing.

2) on my daughter's computer with Supermicro X9DAi motherboard, which is installed in a full tower gaming case, I configured the CPU coolers to blow up, instead of out the back; exactly to avoid blowing hot air from CPU1 to CPU2. The top of the case has several large fans exhausting the hot air out the top. And yet, same issue... one CPU is always hotter...
 

wildpig1234

Well-Known Member
Aug 22, 2016
1,824
273
83
45
i'm tending to agree with you...


i don't think that's the whole answer... because:

1) on HP Z820 systems where they have air ducts to keep the "air channel" for each CPU separated, i've seen the same thing.

2) on my daughter's computer with Supermicro X9DAi motherboard, which is installed in a full tower gaming case, I configured the CPU coolers to blow up, instead of out the back; exactly to avoid blowing hot air from CPU1 to CPU2. The top of the case has several large fans exhausting the hot air out the top. And yet, same issue... one CPU is always hotter...
Yeah, i kinda wonder why for a long time but haven't gotten a good answer. didn't know if this is also the same with epyc, and xeon gold platinium...

yr daughter got a dual cpu game rig? lol... lucky for her.....:) isn't that a little too many cores for a game rig?
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,528
975
113
artofserver.com
yr daughter got a dual cpu game rig? lol... lucky for her.....:) isn't that a little too many cores for a game rig?
not a gaming rig... just a gaming tower case. it's got 2x E5-2670v2 and 1080TI in it too... but it's more of a machine for her to learn. I've got VMs setup for various OSes for her to experience, though the host/base OS is Fedora Linux. I'm teaching her to program on it too... once she gets fluent in some languages, I'll teach her about algorithms and performance optimization, which is where I think that machine will shine. Although, I have wondered if there's any value in her experiencing working on a large C project that takes days to compile... LOL.
 
  • Like
Reactions: Tha_14

Evan

Well-Known Member
Jan 6, 2016
3,128
522
113
You see I could imagine with the most used PCIe lanes being on CPU1 that it would run hotter, the logic for CPU2 hotter I have no idea.
 

i386

Well-Known Member
Mar 18, 2016
2,046
542
113
31
Germany
You see I could imagine with the most used PCIe lanes being on CPU1 that it would run hotter, the logic for CPU2 hotter I have no idea.
Shouldn't the devices being connected to cpu1 do the work and generate the heat?
 

Evan

Well-Known Member
Jan 6, 2016
3,128
522
113
Shouldn't the devices being connected to cpu1 do the work and generate the heat?
That’s what I am trying to say, logic says if any cpu is hotter it would be the 1st one as all the PCI lanes used to drive on board network and storage etc would be on that one.
 

William

Well-Known Member
May 7, 2015
785
250
63
63
Not sure if you are using stock TIM that come with the heat sinks ?
Unmount the fans and reapply TIM and remount.
Its possible you have a bad mount with the CPU in question.
 

WANg

Well-Known Member
Jun 10, 2018
873
503
93
It really depend on how the machine is glued together.

If you look at a majority of SuperMicro dual processor servers, you'll notice on their schematics (look at the motherboard manual) the PCH (the SATA ports, the USB and the IPMI BMC) is attached to one of the 2 CPUs, and it's usually CPU1. CPU2 ends up talking to the outside world via CPU1 (which means that it has to handle its own RAM, its own PCIe slots and then I/O requests from the other CPU via QPI). CPU1 constantly have to deal with interrupts, as opposed to CPU2, which is more isolated.
My guess is that the I/O work is keeping the CPU1 socket from sustained Turboboosting (so less heat buildup), unlike CPU2.
 

mstone

Active Member
Mar 11, 2015
505
117
43
42
Pretty much all dual socket systems the second CPU will run warmer due to being further inside the case, further from the air intake etc.
I've seen this phenomenon on rack mount systems where the CPUs are side by side.

That’s what I am trying to say, logic says if any cpu is hotter it would be the 1st one as all the PCI lanes used to drive on board network and storage etc would be on that one.
That all depends on how the OS schedules the work; it may be that I/O-intensive processes (which spend much of their time waiting) get scheduled on the CPU with the I/O, while CPU-intensive processes get scheduled on the other CPU.
 

wildpig1234

Well-Known Member
Aug 22, 2016
1,824
273
83
45
lol... still no consistent explanation. anyone can devise a test of their hypothesis? ;)..... so is this the same on epyc and xeon gold and platinium? anyone knows?
 

alex_stief

Active Member
May 31, 2016
666
211
43
35
Why do we have to devise a test? Nobody knows what kind of workload you run when you observe this behavior on your systems. Or which kind of add-in cards you use in which slot.
On all the dual-socket systems I have at hand, CPU2 runs hotter than CPU1 if it sits in the hot exhaust air of CPU1. Same for idle and CPU-intensive applications. This remains my #1 hypothesis.
 
  • Like
Reactions: William

wildpig1234

Well-Known Member
Aug 22, 2016
1,824
273
83
45
Why do we have to devise a test? Nobody knows what kind of workload you run when you observe this behavior on your systems. Or which kind of add-in cards you use in which slot.
On all the dual-socket systems I have at hand, CPU2 runs hotter than CPU1 if it sits in the hot exhaust air of CPU1. Same for idle and CPU-intensive applications. This remains my #1 hypothesis.
as was stated, even when the fans are not facing in serial, it still shows cpu 2 hotter consistently.