That was going to be my suggestion. Better to be mining and proving the rest of the system works.
Not sure if you have a windows PC that can boot via USB/Hive to do more testing on the question mark 6700XT.
fwiw: Does your other rig always run such high load average? Are you CPU mining?
Thank you for the suggestion I shall give it a try.
I use Mac (I have windows 10 installed on another ssd - nvme) maybe tomorrow after work shall copy the current HiveOS to another SSD-sata, but then I have to remove everything that works now on this system, which I don’t like. There is another one option to go to someone (company that may do that for me on a test system and to check the problematic GPU)… What must be done on windows 10, how to check it there.
The other rig is mining also with the CPU RTM (Raptoreum) that is why it has a load average.
Or, to leave the system for now it might become stable and if reboot or stops until comes the time before go to work to stop and remove the quest mark 6700xt for testing.
Could you please advice what test to do on a windows system so to be sure that the gpu is ok or not.
In my opinion, you want to validate the new system is stable for a time before dealing with a GPU that appears to have something adverse.
Pull the 6700XT out, let the rest run a day or so then dig in on the questionable GPU with fresh risers, new slot, etc. If that does not work, RMA it.
If the Windows option is not easy or native radeon drivers, skip it.
I totally agree with you. Before your reply I did the following:
My guesses are:
Shall leave all three gpus without the first problematic and shall run the system till risers arrives and then shall check p.2 and p.3 and p.4
In my case,my reference and sapphire work well on lower V and W,but Asrock rx6800xt and Gigabyte need more power to run, may this be just silicon lottery or vary by manufactures?
Which memory temperature is reasonable for RX 6800?
can you please detail the OC window for one of your RX 6800? Thanks a lot.
Both.
Thanks，My reference RX5700/XT RX6800 seems have better efficiency in my rigs，I wish I have more of them.
Did You change bios settings for mining?
Such as - “Set Bios mode to UEFI which may be at 2 places which was at my motherboard. So make sure both set to UEFI. Disable CSM. Set PCIE to Gen 2 (or any place required). Enable Above 4G decoding. Select Integrated Graphics for boot and force. Disable HD Audio (optional)” etc.
Also try not to use PCI_E4. I also recomend bios update.
Thank you very much your reply and given advices.
I did the required changes in the BIOS PCIE to Gen 2; Enable Above 4G decoding and everything else. Integrated Graphics for boot and force is not applicable for me as I have Ryzen 5 3600 (no graphic). Disabled HD Audio, and anything else as I have checked many comments, requirements and advices on this forum and many other for UNIX and Linux based OS’s.
Unfortunately, under UEFI changed in both places (which is very odd to me - so many time, and so many CMOS I did that I have learned this BIOS by heart) I can boot, but can not load HiveOS even with the latest stable version (which of course made me make my BIOS according the requirements for mining under LEGACY mode), not only on the screen but also in the hiveon. I have very stable rig with 6x RX 67000 XT under LEGACY mode, with exactly the same system configuration and parts.
Do you advice me not to use PCI_E4 (if I understand you correctly it is number 4 in yellow in the image:
). If yes, what is the reason any explanation that you have (in my other rig with the same motherboard I have no issues, maybe just a luck). I have red somewhere that this might be applicable for msi x470 Mobo, which has a different chipset.Currently (with most stable riser I have) I decided to check all available gpus I have (each and every one working alone under hiveos for 10 - 12 hours) to check their stability because one gpu I have returned (as previously explained I had bad luck with one PSU, one CPU, one motherboard). I am still waiting for new risers as a have high doubt on them.
About the BIOS, could you please advice which one you use and how via flash or via clean flash (no cpu, ram, and anything else attached). I installed the one that I use in my stable rig (7B86vH6).
Key feature is CSM disabled and uefi boot.
https://imgur.com/a/KQl3rLb bios setting i usualy set.
I found some info here https://www.reddit.com/r/NiceHash/comments/mwm5mn/msi_b450_gaming_plus_max_6th_gpu_connection_issue/
and maybe it is your case.
Via usual flash, no need to remove all hardware. Also i always use latest bios.
And why dont you try other OS just for test? RaveOS for example.
Thank you for the provided links. I know them have read them and tried them. I think it is very good that you post them here not only for me but for everyone whoo read here, such as me.
I shall update then the BIOS to the latest stable version (not beta), when the test for all my GPU pass and the new risers arrives, when I shall give a new try to all mentioned here and not only under HiveOS and shall post what I did and what is the result.
Before starting with the gpu tests under HiveOS, I striated the rig with another nvme SSD under windows 10 with the latest amd drivers to check them (one by one) in device manager for issues, everything with each gpu was fine and no errors.
If I see any issues I shall give a try with other possible OS based on Linux. I use MacOS for personal use and have some very basic knowledge on UNIX based OS’s when tasks and code should be read under terminal.
So, if I decide to increase the number of my rig for more than 5 gpu (6 or more) then I have not to use PCI_E4 anymore. How you solve this. What you use.
Editet: My problem with UEFI post is not with all gpu but when I try to with only one gpu attached with riser on PCI_E1, I forgot to tell.
My last try under Hiveos was only with two gpu attached to PCI_E1 and PCI_E2 and HiveOS saw only one gpu even after my third reboot, that is why I decide to do tests for all my gpu.
I have no such problem on my 2 rigs now. Bios update and settings helped me to launch 7 gpus on my MSI B450-a pro max. 6 in pcie slots + 1 in nvme slot.
I hope you not use sata power connectors and adapters on risers. Just to be sure.
I do use sata power from the PSU with 6 pin connectors and adapters with UBIT risers, I did a quality undervolt for my working rig and no issues as for now. I think that maybe all my current issues are for bad risers, I shall see when the new one arrives. What type of risers you could advise to check for availability.
Migrate away from this. The SATA connectors are rated below PCI spec and just another single point of failure, fire, etc.
The under voltage process is great when everything is working, but you run into a number of conditions where those clocks are NOT applied.
Motherboard PCIe slot (what the riser tried to replicate): 75W/13A
PCIe 6 Pin: 75W/13A - perfect match, weird
Molex: 132W/11A
SATA: 54W/4.5A
Never use sata cables or adapters for GPU - they are not safe. Use 6pin or 4pin Molex.
