RTX Pro 6000 Blackwell Workstation Edition : Not Detected : Multiple Systems : Two Separate Units

Gaming and Visualization Technologies General Topics & Other SDKs GPU - Hardware
, , ,
1

Hello!

Having issues getting a RTX Pro Blackwell Workstation edition GPU to be detected and boot in the following system;

Gigabyte TRX50 Aero-D Rev2.1 (latest bios) + Threadripper 7960X + 1x64G DDR5-5600 ECC MTC40F2046S1RC56BR + Crucial 1TB P510 + Crucial 4TB T700

In this configuration the system boot-cycles, rebooting every time the EZ-Debug lights hit the VGA light in the boot cycle.

Swapping the PSU from the Phanteks listed above to a MSI A1000PL (also ATX3.1 w/ 600w 12v-2x6 connector) changes the behavior slightly, now not cycling but getting stuck on the VGA LED on the motherboard and never posting.

Attempting to test the same GPU on a B850M Gaming Plus WiFi + 7800X3D system that’s otherwise working fine with a RTX5070 and the system boots, yet does not detect the RTX PRO 6000 GPU at all and only provides display-out via the iGPU on the CPU.

Have tried reseating several times, different 12v-2x6 power cables, the two above power supplies plus a Silverstone 1000W ATX3.0 unit with 12vHPWR cable. No luck.

Tinkered with BIOS settings including disabling Re-sizable BAR, 4G decoding, setting the PCIe_1 lane to Gen4 or Gen5 instead of just auto, disabling / enabling CSM mode. Tried different PCIe ports on the Aero-D motherboard, no change.

After all this, we RMA’d the original unit, received a replacement, exact same issues. Unit is different (confirmed by SN) so now we’re reasonably convinced the fault is not with the GPU per-se, and we’re overlooking something obvious in terms of compatibility / requirements.

Having no luck with google / gemini or these forums in figuring it out. Any nudge in the right direction would be extremely appreciated!

edit : (solved - sort of) tried the 4x (6+2) to 1x (12v-6x2) adapter cable that shipped with the card and it worked immediately. Something to do with the 12v-6x2 connectors on the power supplies we tested with (sense pins?) or something… but they work perfectly fine with other high end GPUs and we tested 3 sets of cables. Did we just get very unlucky, is this common place, or are higher end higher wattage PSUs generally provided with tighter spec 600w 12v-6x2 cables that have a better chance of working with these cards?

2

Hello @milestm and welcome to the NVIDIA developer forums.

I tried to find more information internally that I could share, but no luck I am afraid.

From what I know you really might be just unlucky with your existing cables.

This GPU does count as a 600W card and as such has tighter specs. But whether that can be seen with the sense pins or not I cannot say. In general NVIDIA recommends the provided adapter or specifically certified ones. If you have one of the latter, I would RMA those for sure.

3

I had a very similar issue today with a new Blackwell 6000 WE. I purchased a 1200 Watt Corsair power supply at the same time and when I started it up I could barely get a POST. I tried the 4x - 1x adapter as well and it made it slightly farther but still ended up killing the power of the whole machine after a few seconds. I thought the power supply was defective and got a different one (MSI brand) of the same power. Its dedicated 6x2 12V cable makes it past POST but instantly kills the computer right when Ubuntu is bringing up the graphical login screen. I didn’t try the 4x to 1x adapter with this new PSU but will try that after seeing your experience. Very oddly, even putting my older 200-watt GPU back in and powering through the dedicated 6x2 12V cable led to new instability. It would boot but power off when I typed my password or used the computer for 3-5 minutes. Frustrated, I put the original 850 Watt PSU back in with the original low-power GPU and everything was stable again. This is all on an old AM4 X570 motherboard so I had assumed there was some incompatibility there and was planning a proper upgrade, but the story doesn’t quite fit yet.