I’d suggest being methodical, one change at a time. Again, your overclock of 1075 is at the max, not all GPUs will tolerate that value. Then check your risers, etc.
1 Like
I am currently at these parameters and the hash rate becomes very low.
Doing everything step by step and after an hour or two shall see.
My concern is still on problem 3 (what it might be). I googled the issue and it says: " As for “the monitored program dumped core” message, that means that the development environment you’re using ran your program, but your program crashed . That, in turn, means that your program has a bug."
Suspecting your rig ran more than the prior 15 minutes?
Your suspect is correct.
Previously (Yesterday), I had 26 hours working rig with 2x RX 6700 XT and 2x RX 6800 XT., which stopped by itself without any issues and/or error stated. I reboot it and after 3 hours it was an error for two of the gpus: detected DEAD.
I just installed all of the gpus and this is the current status on the following image:
My first rig (x 6700 xt that is very stable for long time) in the following image is with older kernel and different drivers
Good news and glad your 6800XTs are working too.
I’d let them all run a bit, with the plan to increase your core and mem rates a bit and decrease your voltage a bit.
Not sure you’ll be well served with all your 6700’s at 1075, but obviously some fair well.
If it is not broke right now, don’t fix it. Eventually you’ll have a feature/maintenance window/miner update/or a crash that opens an upgrade opportunity.
I hope not to have any problems with the rig 6700’s, the gpus are new and clean and in place with enough place with fresh air.
The mixed one (6700’s and 6800’s) has no issues for more then one day. I shall increase core and mem rates a bit and decrease voltage, but may be after 30 min to see. Currently my fans run noisy and one thing take my attention. On top of one 6800xt where is is the RGB light (lighting of RADEON sign) is blinking, not stable light as the other one 6800xt.
Two and three figures are normal, they are range related.
In my opinion the dump event should be researched in an isolated environment, aka test rig, if it is proven to follow the GPU. Swap the two you have in that rig, just the GPUs, using the slots, risers, cables, and everything else in the position. If that jump still has the dump issue, assume it is the GPU.
I followed your instructions. Swap the two 6700xt’s using the same risers, slots, cables, and everything else. When booted and loaded on the screen the problematic card wasn’t found and when run amd-info the dump thing appaired on one of the 6800 xt’s.
When rebooted and reloaded again none of the gpu’s were recognised.
On my third reboot it reload all gpus, and the problematic card appear again with the same DUMP issue as in the fig.
Intermittently missing well known GPUs in reboots, timing out of well known commands, on a single rig which has been showing something amiss since it was (2) 6700XTs.
Thoughts?
I don’t know, maybe:
-
Stop the system and removing the problematic 6700xt and restart the system to see how long it will work without the removed one until the new risers arives in a few days;
-
Once the risers arrives change them all, check the system again (with and without the problematic 6700xt), if problem still exists to claim replacement for the problematic 6700xt.
-
Is it possible the CPU ot the motherboard to create all this, again
or -
Drivers are not installed properly
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:
- Removed the questions 6700xt and then after running “amd-info” same issue “dumped” appeared on 6800XT and stopped the system.
- Removed all gpu except the new problematic 6800xt with “dumped issue”, alone without other gpus after running “amd-info” no issue, nothing, everything was normal.
- I did double check. Changed the new problematic 6800xt with “dumped issue” (without issues p.2) with the second one 6800xt alone without other gpus after running again “amd-info” no issue, nothing, everything was normal.
My guesses are:
- bad risers;
- bad drivers;
- bad HiveOS installation, or
- bad motherboard
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.