AMD GPU not found

[Easy55]

I am new to Hive OS. I am migrating from Windows. I got my first rig up and running with one AMD 6600XT yesterday. All looked good, and I got the performance I expected.

Next I added a second 6600XT GPU. Turns out it was DOA (I debugged it down to the GPU).

I stopped the miner and shutdown the rig from the HiveOS app. Went to the physical rig. The rig was still up and running it did not stop. The rig would not respond to the Power switch. I turned power off and removed the bad GPU. I turned power back on. I went back to my laptop and HiveOS said the rig was back up and running. BUT, the known good GPU is no longer found.

I have tried rebooting several times. No luck. In Windows, I would uninstall the AMD drivers and re-install in this case. I don’t know what to do in HiveOS. I searched the forum and found nothing that would help.

Again, HiveOS says the rig is up and running. I can send commands and they execute. But no GPU found.

Do need to start over completely and re-flash the SSD?

Any suggestions are appreciated!

MB: Z590-P
CPU: i5-10400
Updated to latest BIOS
BIOS key for HiveOS settings:
VT-d = Disabled
IOMMU Pre-Boot = Disabled
Primary Display = CPU Graphics
PCIEx16_1 Link Speed = Gen1
PCIe Speed = Gen1
Above 4G Decoding = Enabled
Launch CSM = Disabled

[Easy55]

Just ran the upgrade to the latest HiveOS posted today and rebooted. HiveOS says the upgrade worked on my rig. But the one single AMD known good 6600XT card still is not found.

I then ran an amdmeminfo command:

AMDMemInfo for Hive OS v2.1.16
original code by Zuikkis and Yann St.Arnaud

CL_DEVICE_TYPE_GPU Failed: Unable to get the number of OpenCL devices.

[keaton_hiveon]

if you hook a monitor to the amd card does it display video output? are you able to get into the bios and such through that? if not then its not a hive issue, if so then more troubleshooting is needed.

[kaundabits]

I have had same issue when i added an newly bought 6600xt.

Found a forum post telling me to “downgrade” Hiveos to last stable version and reboot.
Then my gpu was found and i could upgrade Hiveos to latest.
Seems its a minor bug.

Hope this can help you.

[Easy55]

I can get to the BIOS through the motherboard HDMI port on reboot. After that the HiveOS blue screen flashes up for maybe 2 seconds, and then disappears. No command prompt is available from the physical rig. The video goes away. But that is just a different problem I am assuming. Replaced risers and cables with all known good, no help.

[keaton_hiveon]

can you get into the bios through the card? using the built in motherboard hdmi port wont tell you if the gpu is working as intended.

[Easy55]

@kaundabits
Thanks for you input. Since there is no video / command line provided at the physical rig, where do I run “hive-replace -y --stable” from ?

[kaundabits]

I did it remote from my pc, on the bothersome rig.
If you can acess your rig from other pc, you might benefit from this post.

troubleshooting-downgrade

Thats the info i used, so wont copy/paste them here.

[Easy55]

My particular problem with AMD GPUs not be recognized was due to a “No Temps” error at start-up. I removed all the AMD 6600XT overclocks, rebooted the rig, and then started mining with the default AMD 6600XT settings. The GPU came up and ran no problem (no “No temps” error). Then while the rig was running I applied my overclocks again, but a little less aggressive this time: core 956, core mV 762, mem cntlr mV 850, mem mV 1350, mem MHz 1100. At 7 hours up time, all is looking good and very stable.

Next I added a second rig, this one with six 6600XTs. I got the same “No Temps” error on this rig. But when I used the same sequence of starting with no OC’s set, and then adding the OC’s once all was working, again all six GPUs are mining and stable. I did need to set the fan speed on this rig because it started to overheat. But at 75% fan each GPU is drawing only 52W to 53W. And max temp is 56C. And I followed the same HiveOS install process for a third 6600XT rig, again good results so far.

Also, note on the missing video output my rig: my Z590-P motherboards don’t provide an HDMI output when running HiveOS, but my Z390-P motherboards do provide video and HiveOS / Miner logging. Haven’t figured out the missing video on Z590’s but that is for another thread!

By the way the AMD driver “No temps” error is nicely reported by HiveOS, but it does bring the rig down. It would be nice if this was “caught” somehow so you could still take actions on the rig.

Thanks everyone for all the good ideas.