Hi all,
I’m running an ASUS Ascent GX10 and the firmware update has been on the progress bar for more than 6 hours. I’d like advice before I touch anything.
What I did (all on 2026-10-18, times are EEST):
- sudo apt full-upgrade: DGX OS 7.4.0 → 7.6.0 (dgx-spark-ota-update-meta 26.09.2), kernel 6.17.0-1008 → 7.0.0-1019-nvidia, driver 580.126 → 580.178.04, nvidia-container-toolkit 1.20.1. After reboot everything worked: nvidia-smi shows GB10, Docker containers can see the GPU.
- sudo fwupdmgr update:
- Embedded Controller 0x02000002 → 0x02000007 (Release ID 149881)
- UEFI Device Firmware / GX10 SoC FW 0x03000002 → 0x03000008 (Release ID 149880)
Both reported “Successfully installed firmware”. I confirmed the reboot at 15:09.
What happened next:
- 15:14 ASUS logo
- 15:28 green progress bar (fast), then a second bar that sat at ~50% for about 20 min, filled up around 15:56, then back to the logo
- From ~16:06 a new bar that hides and reappears at the same position, crawling:
~40% at 16:13, ~55% at 16:27-17:11, ~60% at 17:52, ~75% at 18:53
- At 21:30 it is still around 75%, with very slow or no visible movement
State: power LED on, unit on a UPS, nothing pressed or unplugged. No network (expected during capsule update).
Questions:
- Is a SoC/EC capsule update taking 6+ hours on GX10 a known behaviour, or is it stuck?
- If it is stuck, is the recommended recovery a power drain (hold power 10s, unplug, hold power 60s, re-plug, boot)? Is that safe at this point, or could it brick the unit?
- If the power drain fails, what is the official recovery path for GX10 (recovery USB / capsule re-flash)?
Thanks!
Update: the unit recovered by itself after a power drain. EC updated, SoC rolled back.
After about 8 hours on the progress bar (stuck around 75-77%) I did a power drain at 23:18:
- held the power button until it went off
- unplugged the power brick from the wall/UPS and the USB-C cable from the GX10
- held the power button for 60 s, waited about 3 min
- reconnected (USB-C to the GX10 first, then the brick to the wall) and pressed power once
What happened on boot:
- fans at full speed, black screen for about 3 min, then the ASUS logo with no bar for about 15 min
- then the message “Copy early boot firmware from recovery image to primary image.” with a green progress bar (also slow)
- I left it overnight. In the morning it had booted normally into DGX OS 7.6.0 (kernel 7.0.0-1019-nvidia), reachable over SSH
Result (fwupdmgr get-history):
- Embedded Controller: 0x02000002 → 0x02000007, Update State: Success
- UEFI Device Firmware (GX10 SoC FW): Update State: Failed, “failed to run update on reboot: expected 0x03000008 and got 0x03000002”. The unit is back on 0x03000002.
- BIOS version (dmidecode): GX10DGX.0101.2025.1014.1350
- nvidia-smi and Docker (–gpus all) both see the GB10. Everything works.
- spark-ota-check torn-score: OTA2607, torn = 8
To prevent another automatic attempt:
- the failed SoC capsule (fwupd-f1392323-…cap, 30 MB) was still in /boot/efi/EFI/UpdateCapsule/. I moved it out to a backup folder.
- sudo apt-mark hold dgx-spark-ota-update-meta
- sudo systemctl disable --now fwupd-refresh.timer
Takeaways for others: be patient, and if it really is stuck for hours, a full power drain (brick unplugged from the wall) let the unit restore its primary firmware from the recovery image by itself.
Remaining questions:
- Any idea why the SoC capsule 0x03000002 → 0x03000008 failed and took hours? Could the jump from BIOS 0101 straight to SoC 3.0.8 be the issue? Do intermediate versions (0103/0104/0105) need to be applied first?
- What is the recommended safe way to retry: fwupdmgr again, or the ASUS 0106 package with capsule_update.sh?
- Is it OK to stay on SoC 0x03000002 with DGX OS 7.6.0 / driver 580.178 for now?
Thanks!
Unfortunately, since EC firmware on ASUS GX10 is proprietary, I recommend you reach out to ASUS customer support for further help on the firmware update issue.