Unsolved
2 Posts
1
8162
T640 and Nvidia RTX2080Ti GPU
Has anybody else been running a PowerEdge T640 with the new Nvidia RTX2080Ti GPU? We've been running them for a few months on Ubuntu 18.04, and everything seems fine, other than once in a while, we get a "A fatal error was detected on a component at bus 217 device 0 function 2." error in our iDrac logs. These errors don't seem to affect the server when in use. As of posting this message, all server firmware is up to date, and we're running the latest Nvidia drivers (using an NGC container).
GrishaLevit
1 Message
0
December 4th, 2018 17:00
How many GPUs are you running at once?
Crater Labs
2 Posts
0
December 7th, 2018 17:00
iiee
1 Message
0
December 20th, 2018 18:00
Hey- I just set up a system with a 2070 and it seems to work, but I’m getting the same errors on boot. Have you figured out a way around this such that I don’t need to physically hit F1 on reboot to move past the error?
davidbar2
11 Posts
0
March 5th, 2019 05:00
GPUille
1 Message
0
May 10th, 2019 01:00
Yes you need a kit 6pin PCI Express to 2 X PCIe 8 (6+2) pin like this https://www.amazon.fr/gp/product/B0762H6K5C/
rhelGPU
1 Message
0
June 17th, 2019 10:00
I'm looking to purchase the T640 with the GPU Kit (I assume that includes the GPU PIB), and was hoping to run a combination of GTX 1080 Ti and RTX 2080 Ti GPUs. I would feel a lot better if I knew that these cards are showing NO compatibility issues.
Is the GPU Kit even necessary to operate with two consumer GPUs, or does this require extra components?
Thanks in advance!
rptechsoft
1 Message
0
July 9th, 2019 23:00
Hi, I am interested to know if you got 4 x RTX2080Ti GPU working in T640? Thanks
SteveHart_JHU
3 Posts
0
May 29th, 2020 12:00
Just pinging this to see if there was any resolution to this issue. I have a T640 that we're trying to run four RTX2080Ti GPUs in and even with only one plugged in, we see the same issues. We're seeing fatal PCI bus error as well as errors related to "insufficient power from the PSU". I can't believe the PSU's cannot provide the necessary power. We bought the T640 with the dual (redundant) PSUs. With no cards, the server boots normally. On each reboot, the server stops and waits for a manual command to continue normal boot. Extremely annoying having to trudge over to the server room each time.
DELL-Josh Cr
Moderator
Moderator
•
8.5K Posts
1
May 29th, 2020 14:00
Hi,
It is not a validated card. It may not work properly.
I would try updating the BIOS. What is the wattage of the power supplies? You need at least 1100W. Is redundancy for the power supplies disabled in BIOS? Do you have dual processors?
SteveHart_JHU
3 Posts
0
May 29th, 2020 15:00
Dual 1600W power supplies, Redundancy enabled. Dual processors. I'll double check but I believe the BIOS is updated fully.
DELL-Josh Cr
Moderator
Moderator
•
8.5K Posts
0
May 29th, 2020 15:00
Try turning redundancy off so it can use more power.
williambyrne
11 Posts
0
July 13th, 2020 18:00
@DELL-Josh Cr
Thanks for checking in. Just following this thread as I’m planning to buy one of these systems. The plan is to run a KVM hypervisor with a decent GPU available to the clients.
Please follow up with any possible configs.
DELL-Joey C
Moderator
Moderator
•
3.1K Posts
0
July 13th, 2020 20:00
Hi,
I would probably best suggest that you should contact the sales team on your request, as there are certain configurations and hardware on GPU installation that needs to be confirmed during point of sales.
I would only be able to provide you documents on supported GPU for the system: https://dell.to/2CuIQre
SteveHart_JHU
3 Posts
0
July 14th, 2020 05:00
BIOS fully updated. PSU redundancy disabled. No change in behavior.
I could contact support but you know they will fall back on the same old mantra. "Do not use commercial based RTX cards".
I went so far as to install only one single known good card and the issue remains.
These systems were purchased before I started at this position. At this point, I think it's best to re-purpose these T640s and give up on them being useful for GPU processing nodes.