Topic: Help Please - Vega 56 Mining Rig Troubleshooting (Read 4849 times)
Guys what power suplys u use for rig with 6xVega56??
I just found using castXMR instead of xmr-stak, then the problem fixed!!
Excuse me, what's the reg file?
and according to your words, just set the IGPU in the very beggining right?
What I faced is- I must use vega as the primay output instead of IGPU, otherwise one card would be 400 hs lower (I have two cards).
I don't know what's going on.
and if HDMI as the primary setting, it goes up to normal (~3600 hs), but I cannot do anything.
THanks for your share!!
David
and according to your words, just set the IGPU in the very beggining right?
What I faced is- I must use vega as the primay output instead of IGPU, otherwise one card would be 400 hs lower (I have two cards).
I don't know what's going on.
and if HDMI as the primary setting, it goes up to normal (~3600 hs), but I cannot do anything.
THanks for your share!!
David
This is my setup.
I run my monitor via the onboard Intergrated Graphics dvi port. I made the adjustments in the mobo BIOS to route everything to integrated - it showed up in two places. This way one of the vegas doesn't have to do dual duty.
Install windows. Disable all the "sleep and monitor dimming" settings.
Install AMD blockchain drivers. I installed the driver only. Unchecked everything else.
I did NOT flash my 56 to 64, still not sure if that is going to do much.
Did the reg-fix to all 4 cards.
Reboot.
On boot, disable all vegas. Then re-enable all vegas.
Run the script to apply the Overdrive settings. Fans kick in to 3000 rpms.
Run the miner (make sure you've updated the config for the miner so it can access dual mining on each vega). All four vegas running at 1950 h/s and about 150-170W. I haven't tried to adjust the mV to lower that draw.
I tried to install GPUz and HWinfo but I think that caused problems so uninstalled them. I do leave HWMonitor running on screen.
If I run into issues (like after I installed GPUz) I simply reinstall the AMD Blockchain drivers (drivers only again). Shut down. Power up and repeat the disable/enable -> overdrive script. I disable and re-enable the vegas on every boot up. I haven't had to reinstall windows yet - simply the AMD drivers. I think not having that crappy Wattman tool on the computer is half the battle.
I run my monitor via the onboard Intergrated Graphics dvi port. I made the adjustments in the mobo BIOS to route everything to integrated - it showed up in two places. This way one of the vegas doesn't have to do dual duty.
Install windows. Disable all the "sleep and monitor dimming" settings.
Install AMD blockchain drivers. I installed the driver only. Unchecked everything else.
I did NOT flash my 56 to 64, still not sure if that is going to do much.
Did the reg-fix to all 4 cards.
Reboot.
On boot, disable all vegas. Then re-enable all vegas.
Run the script to apply the Overdrive settings. Fans kick in to 3000 rpms.
Run the miner (make sure you've updated the config for the miner so it can access dual mining on each vega). All four vegas running at 1950 h/s and about 150-170W. I haven't tried to adjust the mV to lower that draw.
I tried to install GPUz and HWinfo but I think that caused problems so uninstalled them. I do leave HWMonitor running on screen.
If I run into issues (like after I installed GPUz) I simply reinstall the AMD Blockchain drivers (drivers only again). Shut down. Power up and repeat the disable/enable -> overdrive script. I disable and re-enable the vegas on every boot up. I haven't had to reinstall windows yet - simply the AMD drivers. I think not having that crappy Wattman tool on the computer is half the battle.
Excuse me, what's the reg file?
and according to your words, just set the IGPU in the very beggining right?
What I faced is- I must use vega as the primay output instead of IGPU, otherwise one card would be 400 hs lower (I have two cards).
I don't know what's going on.
and if HDMI as the primary setting, it goes up to normal (~3600 hs), but I cannot do anything.
THanks for your share!!
David
and according to your words, just set the IGPU in the very beggining right?
What I faced is- I must use vega as the primay output instead of IGPU, otherwise one card would be 400 hs lower (I have two cards).
I don't know what's going on.
and if HDMI as the primary setting, it goes up to normal (~3600 hs), but I cannot do anything.
THanks for your share!!
David
This is my setup.
I run my monitor via the onboard Intergrated Graphics dvi port. I made the adjustments in the mobo BIOS to route everything to integrated - it showed up in two places. This way one of the vegas doesn't have to do dual duty.
Install windows. Disable all the "sleep and monitor dimming" settings.
Install AMD blockchain drivers. I installed the driver only. Unchecked everything else.
I did NOT flash my 56 to 64, still not sure if that is going to do much.
Did the reg-fix to all 4 cards.
Reboot.
On boot, disable all vegas. Then re-enable all vegas.
Run the script to apply the Overdrive settings. Fans kick in to 3000 rpms.
Run the miner (make sure you've updated the config for the miner so it can access dual mining on each vega). All four vegas running at 1950 h/s and about 150-170W. I haven't tried to adjust the mV to lower that draw.
I tried to install GPUz and HWinfo but I think that caused problems so uninstalled them. I do leave HWMonitor running on screen.
If I run into issues (like after I installed GPUz) I simply reinstall the AMD Blockchain drivers (drivers only again). Shut down. Power up and repeat the disable/enable -> overdrive script. I disable and re-enable the vegas on every boot up. I haven't had to reinstall windows yet - simply the AMD drivers. I think not having that crappy Wattman tool on the computer is half the battle.
I run my monitor via the onboard Intergrated Graphics dvi port. I made the adjustments in the mobo BIOS to route everything to integrated - it showed up in two places. This way one of the vegas doesn't have to do dual duty.
Install windows. Disable all the "sleep and monitor dimming" settings.
Install AMD blockchain drivers. I installed the driver only. Unchecked everything else.
I did NOT flash my 56 to 64, still not sure if that is going to do much.
Did the reg-fix to all 4 cards.
Reboot.
On boot, disable all vegas. Then re-enable all vegas.
Run the script to apply the Overdrive settings. Fans kick in to 3000 rpms.
Run the miner (make sure you've updated the config for the miner so it can access dual mining on each vega). All four vegas running at 1950 h/s and about 150-170W. I haven't tried to adjust the mV to lower that draw.
I tried to install GPUz and HWinfo but I think that caused problems so uninstalled them. I do leave HWMonitor running on screen.
If I run into issues (like after I installed GPUz) I simply reinstall the AMD Blockchain drivers (drivers only again). Shut down. Power up and repeat the disable/enable -> overdrive script. I disable and re-enable the vegas on every boot up. I haven't had to reinstall windows yet - simply the AMD drivers. I think not having that crappy Wattman tool on the computer is half the battle.
Gents, I'm having a problem adding a second PSU to run card #5(and eventually 5-8). I have 4 Vega 64s powered by a corsair hx1200i, getting 8k (monero) hashes total @ 780 watts. I set up the rig using this guide http://vega.miningguides.com/ and it's been working smoothly. I'm using an add2PSU to connect two corsair hx1200i's. Whenever I add my 5th card using my secondary PSU, it goes unrecognized in windows. The 'Radeon' lights up, but there is no red light near the power cables.
To troubleshoot, I changed all connections to my second PSU so it acts as primary and it powers 4 cards fine. I've also added 5 cards to each PSU and they all are recognized in windows, and work fine w/ 10K total. One time I switched two cards to the secondary psu - one card had all red lights on, the other just a green light but neither were recognized.
Dried DDU in safe mode then reinstalling blockchain drivers. Crossfire is disabled, tried a different riser, and all risers are powered by primary psu. Also tried to power the 5th card's riser off secondary psu. I originally setup the 4 cards with that guide and it's been smooth. Again, I can get 5 working just fine if all are plugged into primary psu. It's only when it's plugged into secondary that it won't show up in windows.
The problem lies with trying to get any cards powered to the secondary psu to be recognized in windows. I'm using a Biostar Tb-250 pro. Anyone have any suggestions?
To troubleshoot, I changed all connections to my second PSU so it acts as primary and it powers 4 cards fine. I've also added 5 cards to each PSU and they all are recognized in windows, and work fine w/ 10K total. One time I switched two cards to the secondary psu - one card had all red lights on, the other just a green light but neither were recognized.
Dried DDU in safe mode then reinstalling blockchain drivers. Crossfire is disabled, tried a different riser, and all risers are powered by primary psu. Also tried to power the 5th card's riser off secondary psu. I originally setup the 4 cards with that guide and it's been smooth. Again, I can get 5 working just fine if all are plugged into primary psu. It's only when it's plugged into secondary that it won't show up in windows.
The problem lies with trying to get any cards powered to the secondary psu to be recognized in windows. I'm using a Biostar Tb-250 pro. Anyone have any suggestions?
if your getting crashes/hangs your pushing too hard with the clocks vs voltages
you can adjust in overdrivent while minning. start at 1200/900 and work your way up to 1466/960... with the memory you can go upto to 1105/910 (clock/voltage)
it depends on the individual card - silicon lottery
if your only getting 1400hs you need to enable/disable each card... you can use device manager or much better is to use a devcon script.. if your getting crashes when enable or disable cards check the riser usb, power and pcie cables for that card
use the onboard video connector and virtual memory needs to be 10gb per card
using castxmr you should easily get 1900+ per card
you can adjust in overdrivent while minning. start at 1200/900 and work your way up to 1466/960... with the memory you can go upto to 1105/910 (clock/voltage)
it depends on the individual card - silicon lottery
if your only getting 1400hs you need to enable/disable each card... you can use device manager or much better is to use a devcon script.. if your getting crashes when enable or disable cards check the riser usb, power and pcie cables for that card
use the onboard video connector and virtual memory needs to be 10gb per card
using castxmr you should easily get 1900+ per card
Random question. When adjusting settings in OverdriveNTool do you do it while the miner is turned off or while it is running? I feel like if I adjust the settings while the miner is running sometimes it causes it to hang and ultimately cause a restart.
You can identify a serial number next to the PCI slot, which will probably say 1 or 2. You may also want to try uninstalling the drivers and reinstalling. http://vega.miningguides.com/p/faq.html#GPUorder
I did a fresh install of windows a couple days ago to see if it would fix the issue with the disable/enable. As for identifying a serial number what does that correspond to in the properties window of the gpu? I am on a H110 BTC+ motherboard.
The below link is a display Drive Uninstaller you might want to run. After you do that, try reinstalling the drivers. Good luck!
http://www.guru3d.com/files-details/display-driver-uninstaller-download.html
I have been running steady for about a week now. However, the problem that persists is I have 2 cards when I disable/reenable them 1 gives me error code 43 and wont mine. The other gives artifacts and restarts the computer. When the computer comes back on it is disabled I reenable it and it mines without hbcc memory. I jave been running both of these without hbcc memory and pulling around 1400 h/s each But all my other cards are 1900+ so I should have additional 100 h/s but can't figure out why these won't work. I am also not sure how to identify which card is which on the machine versus in device manager.
You can identify a serial number next to the PCI slot, which will probably say 1 or 2. You may also want to try uninstalling the drivers and reinstalling. http://vega.miningguides.com/p/faq.html#GPUorder
I did a fresh install of windows a couple days ago to see if it would fix the issue with the disable/enable. As for identifying a serial number what does that correspond to in the properties window of the gpu? I am on a H110 BTC+ motherboard.
I have been running steady for about a week now. However, the problem that persists is I have 2 cards when I disable/reenable them 1 gives me error code 43 and wont mine. The other gives artifacts and restarts the computer. When the computer comes back on it is disabled I reenable it and it mines without hbcc memory. I jave been running both of these without hbcc memory and pulling around 1400 h/s each But all my other cards are 1900+ so I should have additional 100 h/s but can't figure out why these won't work. I am also not sure how to identify which card is which on the machine versus in device manager.
You can identify a serial number next to the PCI slot, which will probably say 1 or 2. You may also want to try uninstalling the drivers and reinstalling. http://vega.miningguides.com/p/faq.html#GPUorder
I have been running steady for about a week now. However, the problem that persists is I have 2 cards when I disable/reenable them 1 gives me error code 43 and wont mine. The other gives artifacts and restarts the computer. When the computer comes back on it is disabled I reenable it and it mines without hbcc memory. I jave been running both of these without hbcc memory and pulling around 1400 h/s each But all my other cards are 1900+ so I should have additional 100 h/s but can't figure out why these won't work. I am also not sure how to identify which card is which on the machine versus in device manager.
using the b110 BTC+ im at 8 vegas
What are your P7 / P3 states & mv settings?
What's the total hashing rate for all 8 cards after the miner has run for a while? Just curious.
P7 is 1466/960
memory P3 1105/905
15855 H/s over 24 hours
Something new happened. When I start Stak it loads all the gpu threads then for each one it comes up and says hwloc memory pinned. Anyone know what that means and if its bad?
https://www.reddit.com/r/NiceHash/comments/75an32/xmrstak_hwloc_memory_pinned/
Thank you!
Something new happened. When I start Stak it loads all the gpu threads then for each one it comes up and says hwloc memory pinned. Anyone know what that means and if its bad?
https://www.reddit.com/r/NiceHash/comments/75an32/xmrstak_hwloc_memory_pinned/
Something new happened. When I start Stak it loads all the gpu threads then for each one it comes up and says hwloc memory pinned. Anyone know what that means and if its bad?
using the b110 BTC+ im at 8 vegas
What are your P7 / P3 states & mv settings?
What's the total hashing rate for all 8 cards after the miner has run for a while? Just curious.
Interesting predicament. I am running all 6 vega 56 at GPU P7 1408 MHz 1000 mV and memory P3 1100 MHz 950mV. Where it seems to be fairly stable. I did get 1 hang in GPU0 which caused a restart. However, when I undervolt the cards down to GPU 841 mV and memory 805 mV GPU0 that hangs in openCL or something like that. It is always GPU0 and that confuses the hell out of me as these are all Sapphire RX Vega 56 and all have samsung memory. The only difference is that GPU0 is in the actual PCIE slot on the motherboard and the others are all in risers. It does have a monitor connected to it right now but that will be replaced with an HDMI dongle shortly. Thoughts anyone?
First I would try applying a separate ONT profile of maybe 890 / 870 to the GPU0 PCIe card. Or try lowering values in P7 & P3 for GPU0.
If that doesn't work, stick a different video card in the motherboard PCIe slot, maybe an old Nvidia card or something. Does the problem go away?
Or does a different GPU start hanging with the 841/805 mv settings?
Interesting predicament. I am running all 6 vega 56 at GPU P7 1408 MHz 1000 mV and memory P3 1100 MHz 950mV. Where it seems to be fairly stable. I did get 1 hang in GPU0 which caused a restart. However, when I undervolt the cards down to GPU 841 mV and memory 805 mV GPU0 that hangs in openCL or something like that. It is always GPU0 and that confuses the hell out of me as these are all Sapphire RX Vega 56 and all have samsung memory. The only difference is that GPU0 is in the actual PCIE slot on the motherboard and the others are all in risers. It does have a monitor connected to it right now but that will be replaced with an HDMI dongle shortly. Thoughts anyone?
I am running on Stak-xmr right now. And getting about 9900 h/s so I believe it was the miner that was holding me back. However, my energy went up to 1300 from 960. Any thoughts on how to optimize? I am using suggested settings from oldcomers video.
All I can say - check the following post here https://www.reddit.com/r/MoneroMining/comments/74hjqn/monero_and_vega_the_definitive_guide/
Usually vega 56 shoud pull about 150W when minig monero
I am running on Stak-xmr right now. And getting about 9900 h/s so I believe it was the miner that was holding me back. However, my energy went up to 1300 from 960. Any thoughts on how to optimize? I am using suggested settings from oldcomers video.
It has started to hang after mining for a few minutes. Different gpu each time.
Jump to: