So what is the method to update kernel only?
^this
Ok thanks will test later.
I have been running a 10* 5700 rig for a year and cannot go past 0.6-203. Sometimes it stays up for a week, sometimes only a day, but that is good enough for me considering the BIOS mods.
With Samsung memory I can’t go over 900, but on the plus side power usage is way lower, so it is very efficient.
With Micron memory it can go up to 940 but power usage is much higher, to the point of being inefficient.
start with a fresh latest stable image and tune oc from there from scratch
try 5.10.83 with AMD 20.40 that doesn’t have memory leak like 20.30 … And check your memory Samsung 885 Micron 900-930
I have given up at this point with anything above 203, ive been tuning like a madman and still i get gpu dead errors and invalid shares, i have tried all solutions suggested also, nothing worked including new kernel and drivers, so here i am back on 203 now and everything is running super smooth and stable again.
so yes i dont know whats up with above 203 and 5700s but you are not alone, hopefully one day this will be resolved and then i can reliably dual mine some ton also but until then i guess we are stuck with 203
what clocks and voltages are you running? can you post a screenshot of your worker overview screen?
Glad to see I’m not the only one with this issue. My 5700 rig will only work on 0.6-203. Any newer hiveos version or newer linux kernel will result in PCI I/O errors.
Weirdly enough, I have the same problem with my RDNA2 rig. It will only work with the 0.6-203 kernel (5.4.0-hiveos #108), but any hiveos version will work (on 0.6-217 now). Any kernel newer than that will result in PCI I/O errors, like my 5700 rig.
Still thinking it has more to do with your guys’ settings/voltages than it does with software versions. Myself and many many other are running lots of 5700s on much newer versions without issue, just flashed the latest stable and got a slight bump in hashrate even.
Try starting from scratch with your ocs, flash the latest stable and then find the the best settings for your unique card(s)
It’s definitely not the settings. Even on default clocks and voltages and booted with maintenance mode on and drivers loaded, hiveos will throw PCI I/O errors, miner not even running. I’ve attached my hiveos dashboard. Load average is high because the cpu is mining monero. Even with cpu mining turned off and a fresh hiveos image flashed on the ssd, I cannot go any higher than 0.6-203. Been trying almost every new version ever since 0.6-204 came out last year to no avail on my 5700 rig. Maybe it is the motherboard.
Default clocks/voltages aren’t good for mining at all.
Can you post the output of amd-info?
Yes I am aware. My point was that even without any parameters entered into the hiveos dashboard, on any version past 203, during boot, the problem that prevents hiveos from even loading properly is a PCI I/O error. It won’t reach the hiveos landing screen where I can type in commands.
Wonder if the higher core clock is helping mining TON. I got in contact with bitby and I flashed the latest stable kernel and then downgraded as AMD 20.40 seems to be the sweet spot for drivers so I am running
Kernel: 5.10.0-hiveos
0.6-202@220422
Running fine now.
Why is your cpu load usage so high? Do you cpu mine on a 6700?
I updated yesterday to take advantage of the new teamredminer R-mode, and I removed the eth_config=B and eth_big_mode_adjust=64 settings from the config, and the rigs seemed to run stable for awhile. Then I had multiple reboots on both of my 5700 rigs and my Vega rig with 1 5700. I downgraded to 0-6.203 (still using R-mode) and haven’t had a single reboot for 12 hours. I’ve brought this issue to the attention of hiveos and they tell me that it’s my OC settings, which it’s not.
Just an update. The motherboard that ran my 5700 rig died. It was a 6th gen intel motherboard with an i7 6700. The new motherboard is a 9th gen motherboard with an i5 9600. I’m having different issues such as none of the old clocks are stable anymore.
However, I am able to install and run the latest version of HiveOS, 0.6-219@221104. I suspect the issue with some of our rigs not being able to go past 0.6-203 is because the chipset is too old, just a guess.
This topic was automatically closed 416 days after the last reply. New replies are no longer allowed.