vBIOS Upgrade for RTX Ada R6000?

Graphics / Linux Linux
1

I have two RTX Ada R6000 GPUs that are stuck in Hardware Power Brake Slowdown mode previously mentioned here " HW Power Brake Slowdown on a pair of RTX Ada R6000 ", I am exploring whether a vBIOS update could resolve this issue.

Both GPUs are running vBIOS version 95.02.3A.00.01 (build date 2022-12-06). With my system integrator support, I have already attempted troubleshooting by replacing the PSU, verifying power connections, checking temperatures, and testing each GPU individually.

Based on TechPowerUP’s VBIOS collection, there are newer vBIOS versions available for the exact same GPU model, such as version 95.02.59.00.09 from 2023-05-04. Since vBIOS controls power management logic and thermal thresholds, I am wondering if the newer firmware might contain fixes for the power braking behavior that my GPUs are exhibiting.

Before proceeding with a vBIOS flash, I would appreciate any help on whether this approach is recommended for the RTX Ada R6000. Specifically, I need to know if there are official NVIDIA vBIOS updates available for RTX Ada R6000, what the compatibility requirements are, and whether flashing a newer vBIOS from TechPowerUP’s collection carries any risks for workstation-grade hardware.

So far I only know nvflash for linux, just not sure if it will work on workstation cards. NVIDIA NVFlash 5.867 Download | TechPowerUp

2

Hi @xkm121,

Looking at older issues like this it seems either an RMA or a VBIOS update might help you. Both of these have to go through your GPU vendor.

While the RTX R6000 Ada GPUs might be reference designs, they are manufactured by OEMs and as such, they are the first ones to contact for either solution. NVIDIA does not provide public VBIOS updates for these GPUs. I also suggest to not try to use third party downloads of the flash utility. Worst case scenario is that you brick your GPU.

Hope that helps.

3

Great advice. I will reach out to my SI and go from there!

4

Reporting back that I have successfully upgraded BIOS for both GPUs using SI provided flashing tools. Unfortunately the GPU power braking is still present. Interestingly the GPUs functions perfectly in a Windows workstation, pulling full 300watt under load. In the end it looks like a motherboard issue