I am currently trying to overclock my cards, doing this from work on teamviewer lol. I am confused as to why it only lets me drop my GPU frequency to -4% on wattman? Doesn't make any sense to me as the post I am following says to drop it to -30%. Also I dropped my Power Limit to -30% but only saw a 20w reduction... not sure if its because I set my fan at minimum 3000 RPM or whats going on. I have my memory up to 950 right now.
Also, I am only able to set about 20gb of virtual memory because I only have a 60 gb ssd. If I plug in a second oldschool HDD into sata slot 2 and set the virtual memory on that drive to say 60 GB would that work or does it all have to be on the same drive?
I had the same problem when i tried a "normal" driver release, i could only underclock -4% , installed the official amd compute workload driver 23aug and there was the -30% underclock!
I find it alot better and stable to undervolt and set clocks with softpowerplay, OverdriveNTool 0.2.1 has a built in editor.
Have 2 vega rigs working with this settings for some days now, it looks quite stable except 1 card needs 910-920mV or it crashes sooner or later.
https://imgur.com/a/KxdIwThats the problem though is it seems line 17.30 is unstable for me for some reason. On 17.11.2 I am atleast stable with 4 cards.
Have five running steady now on 17.30 with 9977 h/s at 1300w. For some reason some of the cards didn't like certian risers. Had to switch them around until I found a match.
I tried mining with 17.11.2 and couldn't get over 1300 a card. I'm over 2k on each card now. I'm making more on five cards now then when running 6 cards on the new driver update.
So you got 17.30 stable? And it was risers that were your issue? I might order some new risers then
It seems to be risers that were making things difficult. I was about to order some new ones and then I remembered someone saying they just switched them around and got them working. Here's what I did:
1. clean install of windows in UFEI bios mode with all 6-cards plugged in
2. updated windows and all components including gpu drivers
3. paused windows update
4. used amd cleanup utility
5. used DD clean in safe mode
6. installed blockchain driver one card at a time in 16xpcie slot (two restarts per card)
a. if a card did not behave, this is where I tried another pcie riser until the card worked.
b. i then used what i thought was a faulty riser board with another card and all worked just fine (can't make heads or tails of this, but it is what it is)
7. I then plugged in each card, let it install and rebooted a couple of times.
a. this worked up to five cards. when installing the 6th and final card i get the BSOD.
It was stable running Cast-xmr and xmr-cpu, both overclocked and undervolted overnight. I shut it down this morning to work on card six.