Dell PowerEdge R730XD + NVIDIA Tesla P100 (PCIe) power cabling behavior

Hi,

 

My point of view is that, it is not safe in terms of hardware voltage. You may run into fail-safe error that can lead to full mainboard failure. I ran a check on the available GPU cable for R730/R730XD, the only part number turned out is N08NH. I also found on Reddit, someone posted a question about cable available and N08NH was mentioned. 

@DELL-Joey C​

Hi DELL-Joey C,

Thank you for your response. I actually purchased the Gintai - Cavo di Alimentazione... https://www.amazon.it/dp/B0881NDT89?ref=ppx_pop_mob_ap_share. However, when I connect this cable directly to the Tesla P100, the power supply immediately goes to amber/fault state and the server won’t power on.

When I use a splitter (connecting two 8-pin cables from separate PCU outputs into the splitter, then to the GPU), everything works perfectly - the GPU powers up and is properly detected.

My question is: can the single N08NH cable actually handle the full 250W power draw of the Tesla P100, or is the dual-cable splitter configuration the correct approach for high-wattage GPUs in the R730XD?

Thank you for your help.

Yes — the dual‑cable (two 8‑pin feeds from separate PSU/riser outputs into a splitter to the GPU) is the correct and safe approach for a 250 W Tesla P100 in a PowerEdge R730XD.

Reason (concise):

• The P100 (~250 W) will likely exceed the current capacity or rail distribution expectations of a single 8‑pin feed on the R730XD. Feeding it from one PSU/riser triggers PSU protection and the amber fault you saw.
• Supplying the card from two separate 8‑pin feeds (ideally each coming from different PSU/riser outputs) balances the load across rails and avoids per‑cable/rail overcurrent — which is why your splitter arrangement works.
• Generic single‑cable adapters (like the N08NH you bought) may be rated for high current in some contexts, but in the R730XD chassis the server/PSU expects dual‑path distribution for high‑wattage PCIe cards; using a single feed can trip PSU protections even if the cable itself is nominally rated.

Recommendations:

• Keep the two 8‑pin → GPU setup with each 8‑pin coming from separate PSU/riser outputs (as you have working now).
• Wherever possible use Dell‑approved cabling/adapters or verified equivalents rated for this server model.
• Verify both PSUs and risers are configured and healthy in iDRAC/BIOS, and keep firmware updated.
• If you need official confirmation for warranty/support, contact Dell support with your chassis/PSU/GPU exact SKUs.

If you want, tell me your PSU wattage and exact PSU part numbers and I’ll check typical Dell guidance for that R730XD PSU model.

@Dell-Martin S​ 

I’m powering a Tesla P100 (~250 W) on a PowerEdge R730XD and I need clarification on the correct power distribution.

Currently, I’m using two 8‑pin connectors coming from two separate PSU/riser outputs, which feed into the GPU splitter. However, I saw this video where someone draws all the power from a single PSU output: https://youtu.be/fgTUHqCqFG8?si=NRywKYELrHtNnlDd

From an electrical and R730XD PSU/backplane protection standpoint:

• Is it preferable to distribute the load across two separate PSU rails/paths to avoid overcurrent on a single rail, PSU protections tripping, or amber faults?
• Or is using splitter cables from a single PSU output, which theoretically supports the GPU’s current draw, sufficient and safe in this server context?

I want to understand the official/technical best practice to ensure stability, load balancing, and PSU longevity in this high-power scenario.

Hi,

 

Based on the video, the user showed that the split cable is converted back to a single connector to connect to the GPU card. 

 

Distributing the load from 2 PSU seems to be a good direction to have proper load, but I don't have the technical schematics to refer if the power drawn will not affect the mainboard. I only able to provide is the power cable which searched for the server R730XD: N08NH

 

 

@Dell-Martin S​ @DELL-Joey C

Hi Giuseppe, thank you for the clarification.

Just to confirm: if there are no issues from a motherboard or backplane protection standpoint, I will continue with my current configuration, as shown in the photo below, which is powering the Tesla P100 using two separate GPU power outputs (i.e. two different GPU power connectors from the system feeding the GPU splitter, rather than drawing the full load from a single output.

I will of course use Dell-certified GPU power cables (such as N08NH) for the connections.

Please confirm that this configuration is acceptable and does not pose any risk to the system.

Thank you.
.

Hello,

 

R730 has these cables:

J30DG - ASSY,CBL,PWR,GRPHC,R730

ASSY Graphics Power Cable for x8 Backplane to System Board (3.5" Chassis), (Poweredge R730)

Or this one is the same just a version 2:

N08NH - ASSY,CBL,PWR,GRPHC,R730,V2

ASSY Graphics Power Cable

 

This one you have is for the R720:

Part #:9H6FV

Part Desc.:ASSY,CBL,PWR,GRPHC,R720

ASSY Graphics Power Cable for Riser to GPGPU, (Poweredge R720, Female, Tin Plating)

 

The R730 GPU kit comes with low profile heatsink and two cables.

 

I couldn't find a good cable routing for the R730XD, but this is for the R730: