P

Supermicro X9/X10/X11 Fan Speed Control

Daniels25143

New Member
Nov 5, 2016
11
1
3
38
Very interesting chart. So is there a way to edit "TEMPERATURE _TARGET" so that the motherboard will start ramping up the CPU fans a little bit earlier? I don't have my computer in front of me, but does the "CPU temperature" max variable that can be altered via IPMItool the same as the "TEMPERATURE_TARGET" variable?
 

Patrik Dufresne

New Member
Oct 19, 2016
16
1
3
34
BTW, the value "3" is not skipped. It's available on some board like mine: X9DRT
It's mean: Set Fan to PUE(Power Utilization Effectiveness) Optimal Speed
 

MikeWebb

Member
Jan 28, 2018
87
21
8
50
Awesome. Working on a X11SPH-NCTF. I just changed out the fans on a Roswell RSV-R4000 case with Noctua PWM fans and couldn't get the CPU fan to stop pulsing. When ipmi fan mode was set to optimal or standard the system was having a hissy fit when a fan dropped below the threshold value. Setting the mode to full and controlling speeds with ipmitool was instant relief both on the ears and also knowing I didn't make a mistake with buying new PWM fans. So thank you

If I can add one thing to all this. If I set my "Zone 1" values too low (120mm fans on the disks) and the speed drops below the threshold, all my good work is undone and IPMI takes over and everything is running at full speed. To overcome this I lowered the IPMI threshold values for the individual fans and now I can sit next to my server peacefully. I would like to say in silence but now I can hear my 8 SAS drives.

Lower the IPMI fan speed thresholds
"sensor thresh <id> lower <lnr> <lcr> <lnc>"

I.E.

Code:
root@pve:~# ipmitool sensor thresh FANA lower 100 225 300
Locating sensor record 'FANA'...
Setting sensor "FANA" Lower Non-Recoverable threshold to 100.000
Setting sensor "FANA" Lower Critical threshold to 225.000
Setting sensor "FANA" Lower Non-Critical threshold to 300.000
root@pve:~# ipmitool sensor thresh FANB lower 100 225 300
.....etc.....
 
Last edited:

VirtualBacon

Member
Aug 21, 2017
32
5
8
26
Very interesting chart. So is there a way to edit "TEMPERATURE _TARGET" so that the motherboard will start ramping up the CPU fans a little bit earlier? I don't have my computer in front of me, but does the "CPU temperature" max variable that can be altered via IPMItool the same as the "TEMPERATURE_TARGET" variable?
Did you ever figure this out? I'd love to change that
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,519
964
113
artofserver.com
Anyone had any luck getting this working on a X9DRD-7LN4F-JBOD? Control of the mode works, which is what I'm using now. Trying to manipulate a zone to a specific speed returns an error though.

Code:
root@hyperion:~# ipmitool raw 0x30 0x70 0x66 0x01 0x00 0x32
Unable to send RAW command (channel=0x0 netfn=0x30 lun=0x0 cmd=0x70 rsp=0xcc): Invalid data field in request
I'm having the same issue. Anyone figure this one out yet?
 

Sonner

New Member
Jan 20, 2016
9
7
3
45
These two work on my x9srw that has IPMI fw 3.39.
ipmitool raw 0x30 0x91 0x5A 0x3 0x10 0x7f
ipmitool raw 0x30 0x91 0x5A 0x3 0x11 0x7f

0xff is 100% duty cycle. 0x7f is 50%.

I also have a x9dri-ln4f+ and a x9drw-f. Neither the above two commands nor
ipmitool raw 0x30 0x70 0x66 0x01 0x00 0x32
work. I would be eternally grateful if anybody could point me to the right IPMI fw version and right raw commands for these boards.
 

nthu9280

Well-Known Member
Feb 3, 2016
1,588
441
83
San Antonio, TX
I agree. None of the intermediate speed commands work on X9DRW-iF. Just the Standard, Optimum, Heavy I/O and Full.
I've set it to optimal and gave up. I have Nidec 0.3A fans and run about 2000 rpm. Can't recall the SM part# just swapped out the housing. I've gotten accustomed to the noise. Interestingly same fans ran at a higher speed on X10DRI-T4+ board.

I'm on the latest IPMI FW 2.74. The listed version for this motherbord is 2.59. I think I was having issue with IPMIView iKVM console and SMC suport provided the newer firmware. If anyone needs it, search for SMM_X9 and grab the 2.74. As you can see there are two variants SMM & SMT for IPMI. I've asked SMC support If I can use the latest on SMM or SMT for a specific board, they said they have to test in their lab on respective board before releasing it to public. So technically if your board uses SMT_??, you should be able to use the higher version on this page. But since the BMC can be finicky, I didn't want to update without their blessing.

SMC X9 BIOS-IPMI
 
  • Like
Reactions: Sonner

BLinux

cat lover server enthusiast
Jul 7, 2016
2,519
964
113
artofserver.com
regarding ipmi raw fan control commands on the X9DRD-7LN4F, i sent Supermicro support an email and this was their reply:

We only officially support the 4 modes mentioned, standard, full, optimal, and heavy IO.
I believe the controls you are looking for have been removed.
I'm on IPMI/BMC firmware 3.39. Anyone with a X9DRD-7LN4F on an older version of firmware? If so, can you see if the fan control ipmi commands work on older versions?
 

superempie

Member
Sep 25, 2015
30
5
8
119
I solved my noisy build by changing the 140mm fans I have on the side of my build. It's a X9DRG-QF in a Lian Li PC-A75X case. CPU coolers are from SM (probably the ones that came with the board initially) and case fans are the ones that came with the Lian Li case.
First changed the fan speed mode in the IPMI interface from full speed to standard mode. After that I heard the fans spin up and down constantly.
A few minutes ago I changed the Noctua NF-A14 PWM 140mm's to Noctua NF-A14 industrialPPC-3000 PWM 140mm's (FANC and FAND). It's now way more silent. No headphones needed anymore when powered on. :)
Looks like the problem is the minimum RPM on the NF-A14 PWM's, which is 300 RPM. The NF-A14 industrialPPC-3000 PWM's are 500 RPM minimum. Read that somewhere on the internet and just gave it a go. Haven't done any extensive testing yet though.

Edit: just saw IPMI saying they run at 675 RPM @ idle system
 
Last edited:

canta

Well-Known Member
Nov 26, 2014
1,028
198
63
39
I solved my noisy build by changing the 140mm fans I have on the side of my build. It's a X9DRG-QF in a Lian Li PC-A75X case. CPU coolers are from SM (probably the ones that came with the board initially) and case fans are the ones that came with the Lian Li case.
First changed the fan speed mode in the IPMI interface from full speed to standard mode. After that I heard the fans spin up and down constantly.
A few minutes ago I changed the Noctua NF-A14 PWM 140mm's to Noctua NF-A14 industrialPPC-3000 PWM 140mm's (FANC and FAND). It's now way more silent. No headphones needed anymore when powered on. :)
Looks like the problem is the minimum RPM on the NF-A14 PWM's, which is 300 RPM. The NF-A14 industrialPPC-3000 PWM's are 500 RPM minimum. Read that somewhere on the internet and just gave it a go. Haven't done any extensive testing yet though.

Edit: just saw IPMI saying they run at 675 RPM @ idle system
noctua is luxurious fans. I usually use artic F12/F14 PWM PST CO, dual ball bearing low noise. the price is not expensive.

does IPMI trigger Fan Warning due to <800RPM? my Tyan board triggers the warning due to 140mm fan is running ~700RPM,
 

superempie

Member
Sep 25, 2015
30
5
8
119
Noctua is indeed not cheap, but my experience with them are good.

I can't recall if it gave me a warning in IPMI. I don't think so. Just did some searches on the Internet for this issue and the 500 RPM minimum seemed a good explanation. Gambled on that and it worked. Yesterday I did some testing and the system is now way more silent, even under load. Tested with BOINC with 2x GPU and 2x Xeon at max for over 2 hours. The new fans reached a maximum of 1350 RPM according to IPMI.
 
  • Like
Reactions: canta

Tooms

New Member
Nov 17, 2018
29
16
3
I have changed the limits on my x11 and here is an video with the details for it.
 

EffrafaxOfWug

Radioactive Member
Feb 12, 2015
1,218
412
83
does IPMI trigger Fan Warning due to <800RPM? my Tyan board triggers the warning due to 140mm fan is running ~700RPM,
It depends on the a) what the lower thresholds are set to and b) the specific fan I think; my 120mm noctua had a minimum speed at ~300rpm, but I ran into a problem where I couldn't stop them spinning down further than that (since the noctua has its own internal management IC), which meant the IPMI kept thinking they'd failed as they read 0rpm (cue ramping up all PWM to max duty cycle, resulting in the "surging" problem). If your fans are still reading 700 you just need to tweak the thresholds (not sure if Tyan's will work the same as the SMs though).

It seems to be an intrinsic problem with some noctua's, to the extent that I had to swap out the fans in this particular system.
 
  • Like
Reactions: canta

canta

Well-Known Member
Nov 26, 2014
1,028
198
63
39
Noctua is indeed not cheap, but my experience with them are good.

I can't recall if it gave me a warning in IPMI. I don't think so. Just did some searches on the Internet for this issue and the 500 RPM minimum seemed a good explanation. Gambled on that and it worked. Yesterday I did some testing and the system is now way more silent, even under load. Tested with BOINC with 2x GPU and 2x Xeon at max for over 2 hours. The new fans reached a maximum of 1350 RPM according to IPMI.

Tyan S5533 has only CR Field, others are NA, this is totally different with supermicro mobo that I own,,

I did:
ipmitool sensor thres SYS_FAN_1 lcr 560 (front fan)
pmitool sensor thres SYS_FAN_2 lcr 560 (back fan)


thanks guys, remind me to back to STH after hibernating for almost 3 years :)
 

james23

Active Member
Nov 18, 2014
409
75
28
48
this is great info. THANKS!

For those that DO want the normal SM IPMI fan curve, but simply want it at a REDUCED curve (ie you still want IPMI to gradually up/down the fan speeds based upon temperatures its reads) there is a hardware option.
https://www.amazon.com/gp/product/B072M2HKSN/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

this noctura fan controller is EXCELLENT. (as usual with nocturnal). it allows you to have a physical dial to set your 4-pin pwm fan's speed. but whats unique about it- it also can take a 4pin fan connector (from MB) as INPUT, and then adjust fans connected to IT (the noctura) based upon the speeds the MB is telling that fan header to spin at. So in this configuration, the physical dial on this nocturnal fan controller acts as an OFFSET for your IPMI's fan curve.

Im using this exact setup on 2x , 2u x10 DRU-i+ setups where for some reason IPMI has the fans idle at 4600rpm (too loud), even though temps are low/normal. its working great to keep the idle fan speeds low/quite , but if needed for heat they can spin up (if MB/IPMI tells them to).

(ofcouse you can do this same thing with a script + ipmitool, as a few great ones have been written, but i was looking for something a bit more "permanent/solid" as ESXi does not have ipmitool, nor did i want to worry about a remote VM running a script against this remote box).

on caviet - it does seem to have a injection molded sata power connector (a quick google search for Sata power connector fire , will show why these types of injecton molded sata connectors can be dangerious). however i have been assured by noctura support, that they tested their connectors extensively (FWIW). I personally still soldered and heatshrink'd a normal molex connector onto the 4x of these units i have running in servers.

(one note on this reference doc, while this is stated, imo its not very clear- if you manually set/override the actual fan speeds (via first setting fan mode to full), then the ipmi controller will not adjust the speeds based on the temperature at all (ie no fan curve will be used, or rather the fan curve will be a flat line at whatever you set the PWM % to). I tried to test if you pass the high alert CPU temp threshold (ie ~95c , ) would reset back to the fan curve and kick the fans back to 100% , but i wasnt able to get my cpu hot enough. I just wanted to add this info incase others read it. (again this was stated in the article, so maybe it was just me)
 
  • Like
Reactions: canta

canta

Well-Known Member
Nov 26, 2014
1,028
198
63
39
this is great info. THANKS!

For those that DO want the normal SM IPMI fan curve, but simply want it at a REDUCED curve (ie you still want IPMI to gradually up/down the fan speeds based upon temperatures its reads) there is a hardware option.
https://www.amazon.com/gp/product/B072M2HKSN/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

this noctura fan controller is EXCELLENT. (as usual with nocturnal). it allows you to have a physical dial to set your 4-pin pwm fan's speed. but whats unique about it- it also can take a 4pin fan connector (from MB) as INPUT, and then adjust fans connected to IT (the noctura) based upon the speeds the MB is telling that fan header to spin at. So in this configuration, the physical dial on this nocturnal fan controller acts as an OFFSET for your IPMI's fan curve.

Im using this exact setup on 2x , 2u x10 DRU-i+ setups where for some reason IPMI has the fans idle at 4600rpm (too loud), even though temps are low/normal. its working great to keep the idle fan speeds low/quite , but if needed for heat they can spin up (if MB/IPMI tells them to).

(ofcouse you can do this same thing with a script + ipmitool, as a few great ones have been written, but i was looking for something a bit more "permanent/solid" as ESXi does not have ipmitool, nor did i want to worry about a remote VM running a script against this remote box).

on caviet - it does seem to have a injection molded sata power connector (a quick google search for Sata power connector fire , will show why these types of injecton molded sata connectors can be dangerious). however i have been assured by noctura support, that they tested their connectors extensively (FWIW). I personally still soldered and heatshrink'd a normal molex connector onto the 4x of these units i have running in servers.

(one note on this reference doc, while this is stated, imo its not very clear- if you manually set/override the actual fan speeds (via first setting fan mode to full), then the ipmi controller will not adjust the speeds based on the temperature at all (ie no fan curve will be used, or rather the fan curve will be a flat line at whatever you set the PWM % to). I tried to test if you pass the high alert CPU temp threshold (ie ~95c , ) would reset back to the fan curve and kick the fans back to 100% , but i wasnt able to get my cpu hot enough. I just wanted to add this info incase others read it. (again this was stated in the article, so maybe it was just me)
That is neat quick solution and very useful to whom needs quick solution and do not mind paying $$
This only lengthen PWM through potentiometer (to let the fan running slower). and "no stop" is monitoring minimum 300RPM for the fan when is enabled


For DIY and thinkering:
mini arduino clone with minimal parts can do this, but you need to program. :-D for cheap. monitoring RPM can be looped back via USB serial from clone arduino.
I did some PWM controller in the past, and start working in my spare time to control and monitor fans for TS440 (fans rpm do not show in lm-sensors and no way to control RPM). monitoring and controlling ( fan speed via PWM) can be accessed via usb serial ( similar like arduino serial monitor)
 
  • Like
Reactions: dswartz

dswartz

Active Member
Jul 14, 2011
393
33
28
Very nice! I have an x10 MB in a 4U SM case, with two side-by-side 80mm fans. It's a little noisers than I like - they run at about 3200RPM. The CPU is running nice and cool, so I'm thinking I could use this to cut the noise quite a bit!
 

laujuth

New Member
Feb 14, 2016
16
0
1
42
I've literally been looking for this info for several years now.
After having looked for weeks now to stop the BMC from revving up all fans after it fails to properly read RPM values (every x minutes, a random FAN was at 0 RPM, which is a big fat lie). Now that I've set fan speed to full and set speeds similar to yours, I finally have a silent NAS in my living room. A BIG thank you from me! :)