Unable to update firmware on K1 in Dell 720

We have been running our K1 in a dell 720 running esx5.1 for the past few years. I am now planning on running vGPU in esxi 6 and understand I need to upgrade the firmware on the K1 card to avoid the PSOD. I am trying to apply
Nvidia High Resolution (Dual-Link) Multi-Monitor Virtual Machine Enablement for K1 (Dell PN R8RGR) and K2 (Dell PN JW9YC and 98RCK) cards Nvidia Multi-Monitor Virtual Machine Enablement for K1 and K2- Adds High Resolution Multi-Monitor support for Nvidia K1 (Dell PN R8RGR) and K2 (Dell PN JW9YC and 98RCK) cards in Virtual machine configurations

  • To enable high resolution multi-monitor support the video BIOS (VBIOS) needs to be updated to the following versions via the provided self-run executable.
  • This is in addition to upgrading the NVIDIA Display Driver to Version R320.86 or later.
    80.07.DC.00.05/06/07/08 (NVIDIA GRID K1 P2401-502)
    80.04.F5.00.03/04 (NVIDIA GRID K2 P2055-552)
    The instructions say I need to boot with Linux to perform the update. I followed the instructions from Upgrading the Nvidia Grid Card Firmware | Jan Hendriks Blog but cannot get the server to boot into Linux with the grid card installed. If I remove the K1 I can boot from the USB to the Linux desktop just fine. With the card in Linux begins to load and then just hangs almost like it is trying to load the K1 card but cannot. This has been super frustrating. I am using ubuntu-16.04.2-desktop-amd64 but have tried other Linux distros with the same result. I have also tried putting the card in a R730 with the same results. Any ideas would be appreciated.
1 Like

I’m currently facing the same issue - I’m not able to find firmware compatibility matrix for our K1 cards against ESXi 6.0 (or even 6.5). We are using 80.07.be.00.xx. I noticed that for 80.07.DC.00.xx the release notes are from 2013 - for me personally this is totally unacceptable…
I managed to upgrade the DELL R720 ESXi hosts to 6.0 U2 2 weeks ago and since then some VDIs are just loosing connection and unable to reconnect. If i reboot the hosts it works for some time. The firmwares for the BIOS, NICs, storage controllers have been upgraded/confirmed to be compatible with ESXi 6.0 U2.
Did you manage to find more information? Any help will be greatly appreciated…

1 Like

I have the same issue, and do not see this solved anywhere. Its looking like maybe the server needs to be set to boot in BIOS mode and cannot boot in UEFI mode without throwing PCIe “training” errors. This issue goes away in BIOS mode provided you have set the “Memory Mapped I/O above 4GB” BIOS setting enabled:

As well as the associated PCIe slot enabled? For my R720 the valid slots would be 4 OR 6:

The firmware hasn’t been updated in a very long time (presumably due to the inability to monetize additional licensing fees based on how much use you actually get out of these cards…) so I am going to load up my Ubuntu LiveCD on a USB drive OR Remote Console “Virtual Media” in BIOS mode (already installed in UEFI mode, but the Grid Cards don’t let me boot in UEFI mode for whatever reason…) and reinstall Linux, then attempt to reflash the VBIOS of the NVidia card.

I see there is an “URGENT!” update covering these and the Tesla cards in 2016 I can try next if that doesn’t work…

This is pretty shady from a support perspective NVidia. Tisk tisk. You have been shamed.