Hello thanks for choosing Dell and welcome to our community!

https://dell.to/4bMsrul
These are the things we need to check:
-Maximum 30°C for CPU > 240 W
-DELL POWEREDGE SERVER EXTENDED AMBIENT 40°C OPERATING SUPPORT (ASHRAE A3 COMPLIANT)
A2 GPU is not supported
 
What CPU do you use?

Respectfully,

Thanks for welcoming me!

My cpu model is the AMD EPYC 9124 16-Core Processor

I have 2 in the server.

Best regards

https://www.amd.com/fr/products/cpu/amd-epyc-9124

this seems compatible with the A2

cpu temperatures are currently above 30° !

Best regards

Hello could you take a look at page 15 and see which slot you are using?
https://dell.to/48u1nNI

 

Respectfully,

Hello
I'm coming back to you after doing some tests.
I lowered the temperature of the server room 
My graphic card is now recognized and I have placed it in riser 3 but it increases the speed of the fans to 100%!
I've tried all the thermal profiles!
I searched on the net and saw that this problem is recurrent on pci cards.

With racadm:

get system.thermalsettings
[Key=system.Embedded.1#ThermalSettings.1]
#AirExhaustTempSupport=Not Supported
#AirTemperatureRiseLimitSupport=Not Supported
#CurrentSystemProfileValue=Maximum Performance
DriveTemperaturePolling=Default
#FanSpeedHighOffsetVal=75
#FanSpeedLowOffsetVal=25
#FanSpeedMaxOffsetVal=100
#FanSpeedMediumOffsetVal=50
FanSpeedOffset=Off
#MaximumPCIeInletTemperatureLimitSupport=Not Supported
#MFSMaximumLimit=100
#MFSMinimumLimit=26
MinimumFanSpeed=255
#PCIeSlotLFMSupport=Not Supported
#SoundCapState=Not Applicable
#SystemCFMSupport=Not Licensed
#SystemExhaustTemperature=21
#SystemInletTemperature=13
#SystemInletTemperatureSupportLimitPerConfiguration=35
#TargetExhaustTemperatureLimit=70
ThermalProfile=Maximum Performance

racadm>>racadm get System.PCIESlotLFM.3 
[Key=System.Embedded.1#PCIeSlotLFM.3] 
#3rdPartyCard=No 
#CardType=Video 
LFMMode=Automatic 
#MaxLFM=1030 
#PCIeInletTemperature=20 
#SlotState=Defined 
#TargetLFM=Airflow Controlled

I tried to change the LFMMode but the graphics card is not recognized as a 3rdPartyCard. Apparently it's not supported.
I've tried to take control with ipmitool but I get a privilege error when I'm root.

root@server:~#ipmitool -I lanplus -H ip-idrac -U root -P password raw 0x30 0x30 0x01 0x00
Unable to send RAW command (channel=0x0 netfn=0x30 lun=0x0 cmd=0x30 rsp=0xd4): Insufficient privilege level

root@server:~#ipmitool -I lanplus -H ip-idrac -U root -P password  user  list
ID  Name             Callin  Link Auth  IPMI Msg   Channel Priv Limit
1                    true    false      false      NO ACCESS
2   root             true    true       true       ADMINISTRATOR

I've made progress but I'm really stuck because everything I've been able to find doesn't work! I hope you can help me!

Best regards

When I put the pci card:

I'd like to point out that when I remove the graphics card, the server has an average fan speed of 27.13% PWM, which is what I'd like with the graphics card.

Best Regards

INFOLOGEEK,

 

 

Would you confirm a couple details for me?

Is the server up to date on BIOS, iDrac, etc?

What is the specific part number for the Nvidias you have installed?

If the server isn't up to date it may cause the fans speed to increase, and if it has an unsupported version of the card it can do the same. 

 

Let us know.

 

Hello
I'm coming back to you after doing some tests.

I update my bios to the latest [ 1.6.8 to 1.7.1 ], IDRAC is up to date.

I ran another test with the updated bios but I still have the fans running at 100%.
This graphic card is offered in the R6625 configuration.

https://www.dell.com/en-us/shop/dell-poweredge-servers/poweredge-r6625-rack-server/spd/poweredge-r6625/pe_r6625_tm_vi_vp_sb

Many thanks for your  support.

Hello, I’m thinking maybe installing KUDA on OS might be an idea.

Respectfully,

Hello
I installed cuda on the bone but unfortunately it didn't change anything.

I tried all the solutions:
- Thermal profile
- The card is not recognized as a warm card so I can't modify the LFM
racadm>>racadm set System.PCIESlotLFM.3.LFMMode 1
ERROR: PCI3022: Unable to adjust the Airflow Linear Flow per Minute (LFM) settings of the PCIe slot
       because the LFM settings can be adjusted only for installed cards that are identified as third party cards.

- ipmi mode raw no longer supported in idrac9
- I've installed the cuda driver 
- I never thought I'd have so many problems with a graphics card!
- The server is new, I'll call support and hopefully come back with a solution.

Thank you very much for your help 
Best regards