Could be CPU
max_pcie_lanes: 16
so you can control max max max …8 GPU - if he can stand on it
Meaning 2 lanes per GPU? I’m not really sure how it works. In my head, you need only one lane to keep it up and running.
Hi Guys, is there anything I can do to get more hash rates to the GPU with Samsung RAM. I thought Samsung RAM is better.
What is the best hash rate you get from your GPU’s? Did anyone able to get more than 29 MH? Is my temp’s OK?
TIA
recently I use teamblackminer and this brings out 30MH/s,
I see that same on the pool so maybe it is
Yes, to be taken for calculations - 2 lanes /GPU.
Now I don’t remember but it wasn’t all. …
I did send my Gigabyte Eagle back to NewEgg because I was not getting above 27Mh no matter what. Try core 950 | 700/700/1200 and 950 mem. If it does not wake up - send it back.
What motherboard/CPU you’ve got?
I did try to run only 8GPUs last night - same story. This should not be a problem. I do believe my CPU or something along the lines related to mobo/BIOS causing this issue. Waiting for my new 12 series mobo and CPU to test it out.
Don’t waste your time. I have tried TBM a number of times, seems to get better hash rates to start off, then starts barfing out invalid shares. Once you calculate total shares - invalid shares, TRM is better.
1 Like
Thanks for the info. That will save me time.
Anyways Gigabyte Eagle is sucks and if you have one - send it back before it’s too late 
This thread has helped me a bunch. Thanks everyone (especially Batfink) for the info.
Here is what I have that is (mostly) stable on TRM / ETH.
GPU #3 is a champ. The others undervolt decently but GPU #4 can’t run the 901mhz clock speed without at least a 720mv+ core for some reason… it can run lower voltage at 950mhz though at similar wattage lol. The other 2 use micron memory and run stable at a little bit higher wattage but still low enough to keep me happy.
- Question -
Which SoC frequencies and SoC undervolts are you guys using on ETH+TON mining?
When tuning SoC values for stability do you go up/down by ~5 on both the voltage and frequency? Only one of my cards can really do the 418/772 and keep stable. How big of a difference is 418/772 vs. 420/790 vs 500/850 vs 800/800 for example? Many of you seem to use wildly different values.
Thanks!
1 Like
30 mhs? I use TRM too but no way with decent power saving settings is possible to go as high. Do you use any special settings?
My bad, anyway TBM inflates hashrate in my tests and produces less shares despite showing higher hashrate compared to TRM
Anyone have info. on how the SoC values work?
I have mine set reasonably low, I do notice the hashrate starts to climb up on all cards then settles back down compared to no SoC values at all (max stable hashrate).
I’m guessing as the hash increases, the cards start pulling more power and these values tell it to tame down a bit?
Also, what are the default values for the 6600 cards? (and others if possible… hive only shows default 5700xt values)
@tlatch52, looks like you are doing well with your OCs, nice hashrate and low power consumption. SoC values are a bit of an extra power saving once everything else is stable and working fine and it looks like you have values that are pretty low already.
I usually get everything else stable then start dropping SoC Freq by 10 each time until it crashes or the hashrate starts to fall off a cliff, then just raise it up slightly (+5) to keep it stable, (a lower frequency should require less power), test for a bit (24hrs) then start dropping SoC VDDMax by 10 also in the same way. VDDMax is voltage so lowering it will reduce power consumption more directly but find out what SoC frequency it is happy with first, this will kinda dictate the voltage it will need. Drop VDDMax until it crashes or is unstable, then raise slightly (+5). All cards have a slightly different stability point (silicon lottery) so don’t expect all your cards to be the same, they won’t be
‘amd-info’ and ’ cat /var/log/amd-oc.log’ show a bit of extra SoC parameter info.
To find the default SoC values, remove your existing SoC values, let HiveOS apply the changes and then run ‘amd-info’ and it will show you the default values. You can then easily revert back to your original config.
SoC = System-on-chip. In a simplistic sense it is the ‘Operations Manger’ for the GPU, it makes sure everything runs smoothly on the board (and alot more). Mining is a pretty simple operation for a GPU so the SoC does not have much to do, so we can throttle it back to consume less power. Obviously there is a tipping point where it does not have enough frequency and voltage to do its job properly and then things go tits.
1 Like
@Dojo76, Hi buddy, sorry for the 2 month delay responding to you. Did you ever get anywhere with the 4-port splitter card in the USB (PCIE signalling) port? I can’t see any reason why this would not work but there may be motherboard limitations we are not aware of. I am very interested if you managed to get over the 11 GPU limit.
I did (kinda) fix my random crashing by upping a few more volts here and there
It does still randomly crash every week or so 
I purchased one of these about 2 months ago on Amaz@n to reboot the rig after several (failed-to-auto-reboot) crashes when I was away from home. Works really well, no random switching on and off as some reviewers on Amaz@n mentioned. My one is in direct line-of-sight of the wifi router. I can turn on and off the rig remotely via Internet with a downloadable app for Apple/Android. This model also has power consumption stats. Rock solid so far. Very useful if you have a rig that waits for you to leave the house then immediately crashes
The (phone) app is basic but does what it says on the tin (be a bit careful, it is quite easy to turn it off accidentally when the app it open).
Disclaimer: I dont work for tp-link and I have no affiliation with them.