Unsolved
1 Rookie
•
7 Posts
0
87
R740xd + Nvidia A100 = fans at Max?
Hello, I have a Dell R740xd and have installed a GPU enablement kit into it, and then an Nvidia A100 into PCIe slot 8. Everything is working great, I can see the GPU in my OS (Proxmox) and work and interact with it normally. The only issue is that the fans are operating at full blast even though everything's at room temperature. I know that the system will run at full blast for unrecognized 3rd party cards but the idrac does not recognize the card as 3rd party so I don't think this is the issue. I am suspicious that it might be because the system can't get a temperature for the GPU "System Board GPU8 Temp" in screenshots. Is there any way I can more reasonable behavior here? The server needs to be near equipment so I need it in a lab, where the noise will be a big bother.
DELL-Young E
Moderator
Moderator
•
3.7K Posts
0
January 8th, 2024 05:20
Hello thanks for choosing Dell. This may help;
https://dell.to/3NUKmoB page 41
￭ Using a GPU card: This results in an increase in overall system acoustics
aquirdTurtle
1 Rookie
1 Rookie
•
7 Posts
0
January 8th, 2024 20:17
Thanks for the reference. I understand that in general I'd expect it to be louder, but it's surprising and undesirable that it'd be so loud immediately after boot with no yet load on anything, and with everything still very cool. I'd expect the server to increase the fan speed as temperatures rise, as normal computers do, and as the server was doing before the GPU installation. I'm trying to understand if this is "normal" behavior or a bug with the way I've configured things. I've been suspicious of the latter, especially since the fans are literally at 100% and the iDRAC is not reporting a GPU temperature.
DELL-Chris H
Moderator
Moderator
•
8.5K Posts
0
January 8th, 2024 21:11
AquirdTurtle,
While it could be the amount of hardware installed, I would first make sure the sever is up to date on BIOS, iDrac, Raid controller, gpu, etc. Essentially just make certain the server is completely up to date, I ask this as the iDrac is what controls the fans, and even something else being out of date can misreport its details and cause the fans to spin up. So my first step would be completely updating the server to current and then see how the fans are afterwards.
Let me know if thie helps.
aquirdTurtle
1 Rookie
1 Rookie
•
7 Posts
0
January 13th, 2024 02:23
Hi Chris, thanks for the suggestion. I have just updated all of the drivers you mentioned but unfortunately it did not change any of the fan behavior. Looking for next steps.
DELL-Young E
Moderator
Moderator
•
3.7K Posts
0
January 15th, 2024 03:06
Hello
just to double check- have you followed the installing guide on page 120?
https://dell.to/3O7zQu3
Respectfully,
aquirdTurtle
1 Rookie
1 Rookie
•
7 Posts
0
January 16th, 2024 18:37
Hello, yes I followed this carefully, including the installation of all of the gpu enablement kit hardware such as the high-performance fans (visible in the screenshots above) and gpu shroud.
DELL-Chris H
Moderator
Moderator
•
8.5K Posts
0
January 16th, 2024 19:33
It may be that due to the level of hardware installed, it is requiring the fan speed to maintain the temp, another thought is that the specific GPU isn't a supported kit for the server.
Lastly, you may want to confirm your configuration to the table on page 10 here, as it states that under some configurations the Nvidia A series aren't supported in those configs.
(edited)