Many people are having problems with the 6800 XT version, myself included.
Hope this is fixed with some driver update!
The models of my cards are XFX. 
I made my own modified script.upp.sh
It should only affect RX 6800s, but it might have some problems with AMD APUs because I don’t have a lot of sample data for them.
If anyone uses this, please let me know how it goes.
4 Likes
It’s gpu4 for you
I am running into an issue where it runs the script but times out. Anyone run into this issue before? Suggestions to resolve? I’m trying to get this to work on the 6800XT.
Perfect work! Sorry but I don’t have time this week to make this I think we have good script now. So now we can create full script to check if “UPP” is not installed to install it automatically (with python3) and then create new script (with your version) and put in on startup.
That’s above my skill level, so I will have to leave that to someone else.
__
I’ve had some people test it on 6600xt and 6800xt. It didn’t make much of a difference, either at Fclk=1550 or whatever the max value was under amd-info
Has anyone seen any effect on other RX 6000 cards?
I was all ready to post about doing a UPP dump of the pp_table for my 6800 and not seeing the FClk entry. Got distracted moving cards and motherboards around and for some reason…it works now. Confirmed the script worked in the journal but the best part is the 6800 is now in the 60’s instead of the mid 50’s. Power is a bit higher but I think that is more to do with the version of Hive I needed to run on this rig (6.193). For some reason, the Vegas (these XFX non-ref models have always given me issues) won’t work on any higher versions.
Thanks again, everyone.
undervolting for big navi wasnt supported until the 0.6-205 system image
the numbers on the dashboard aren’t actually being applied to your 6800 if you check amd-info
One way to improve the script for detecting GPUs would be to use the amd-oc command, at the end it shows the correct number for each one:
For the installation I included the parameter --assume-yes, to install without asking me something:
sudo apt-get --assume-yes install libgtk-3-dev build-essential python3 python3-pip
pip3 install upp
After installation I run each command to change the frequency followed by amd-oc command to apply:
upp -p /sys/class/drm/card4/device/pp_table set smc_pptable/FreqTableFclk/0=1550 --write
upp -p /sys/class/drm/card5/device/pp_table set smc_pptable/FreqTableFclk/0=1550 --write
amd-oc
2 Likes
Hello All
I’m new to mining
I have problem with my RX 6800 i try all over clock i can find in this page but not working for me,
I’m kind know the problem but i don’t know how to fix it,
i have 13X RX 6800 biostar i manage to get them all work at 760 MH,
the problem i face that most of the card when i increase the memory clock more than 1050 or 1055 they will crash or will drop in hash rat to 30 MH, what ever i did or what i change when it come to memory clock they have red line at 1050, can i please have any help regarding this problem i have my rig from 2 weeks now and i don’t start mining because i cant find any solution for the memory clock.
my best card work at 61.16 MH and worst work at 54.71, under this number.
Hi there
I have following issue on my new RX6800
From scratch I was able to run it at Win10 I use nbminer with Radeon Soft with 62MH/s
Settings was more less 1300Mhz / 2140Mhz, voltage went down on its own in the Radeon tool to 763mV
After 2 days of working one of the cards crashes and than the issues started…
It wont get up unless I take it out teh power supply cords and connect back.
Than one of the cards run oveclocked normally and stable the second one I left with factory settings.
It was more less ok, but any attempts to OC sooner or later crashes one of the cards.
Last symptom which pop up is NO RGB (so the card is dark - no lights) on one of the cards (but still mining). Once again take out the power cords solve temporarily the issue.
So … questions as follow:
- what causes teh issues - are there any harware issues?
- what software for OC you guys use - doesn’t look like Radeon one
- any other suggestions?
Regards and thanx in advance
You are running in the right range, and I hard set the voltage at 900 which tends to run less than 800mV as you see.
1200 in window for core is enough for 62.5 to 63.5 in my experience.
2138-2150 memory in Windows with Fast Timings enabled is enough for 62.5-63.5 in my experience.
The killer is to ensure zero fan is disabled and you get enough cooling by default with 50% fan minimums. Radeon software lets the memory get too hot in my experience.
If you have not run the fclock scripts and have not gotten the 32-33Amp settings I would try these ranges and work up:
Note: TeamRedMiner on HiveOS 210826
Note2: Ignore the DPM 3 artifact it is left over and I have not taken time to get it out of there.
I really appreciate your help but i m new to mining and new to Linux and hiveos
i really don’t know where to go and what to write i search about script in google and YouTube and no one talk about how to use them i will be so grateful if you can guide me where to go and what to write.
thank you for you replay.
Thank you for response.
Any idea why card crashes?
Do you use Radeon tool or anything else in Win?
Before jumping into the Linux commands see if your cards will run stable using the HiveOS interface.
I’d start by copying GPU2 and GPU3 settings in picture I added. these are “low” and even the worst 6800 GPU’s will run stable with them.
Assuming the correct driver and OS match: In my experience nearly always temperature, pushing overclocking attempts, or a faulty GPU. Again, ignore the DPM 3 setting, but I don’t believe you’ll be presented it as an option with your cards anyway.
I try them i crash as i told you already if i but memory clock more than 1050 or 1055 i crashed snd this is my problem if I know how to get memory up will be easy to reach 60 to 61 in all of them
I try every thing to solve memory problem but not working i flush bios from the good card and the same
But the script or play in bios oc that what I didn’t do
From the name, they appear to be AMD Reference cards.
The BIOS does not match any of my AMD Reference cards.
What brand of GPU are they?
Which BIOS are they running, stock, from a 6800 XT, or?
These (2) are formerly 57MH/s AMD Reference cards on stock BIOS that have had the Linux @frxbg script run against them. The directions to make it work are in a couple of threads now, but I do not consider it “easy” for a Windows only user. I am new to linux myself but had to learn to edit things for the VII GPUs and was able to get the 6800’s done in an <hour on my test rig below:
Good luck and test on one or two in a separate rig if you can. May as well have the rest mining while you are learning!
Here is the post I used for “directions”, there is a folder you need to create which is not clearly listed in a step, but you can get the script to run without it manually: