VEGA 64 mining FREE WISDOM

[jimmykl]

Why would it pull power and burn itself out without even trying to spin a fan? Ask powercolor or whoever made that shitty bios. It didn't destroy the card completely, but it never performed stably at any mining after that incident.

Er don't blame a 'shit bios'. Every card I've used has pulled power for the fans from the PCIE slot. No riser connected = no power for fans = crispy hardware.

[1715603715]

1715603715

[1715603715]

1715603715

[1715603715]

1715603715

[1715603715]

1715603715

[Permonik]

Why would it pull power and burn itself out without even trying to spin a fan? Ask powercolor or whoever made that shitty bios. It didn't destroy the card completely, but it never performed stably at any mining after that incident.

Er don't blame a 'shit bios'. Every card I've used has pulled power for the fans from the PCIE slot. No riser connected = no power for fans = crispy hardware.

That´s not even true!!!!!! The power for fans is not supply by risers. When the riser is not connected, the gpu card just do not start, so the fans do not spinning. You need to deliver the gpu start signal from the mobo via riser!!!!!!

[jimmykl]

Why would it pull power and burn itself out without even trying to spin a fan? Ask powercolor or whoever made that shitty bios. It didn't destroy the card completely, but it never performed stably at any mining after that incident.

Er don't blame a 'shit bios'. Every card I've used has pulled power for the fans from the PCIE slot. No riser connected = no power for fans = crispy hardware.

That´s not even true!!!!!! The power for fans is not supply by risers. When the riser is not connected, the gpu card just do not start, so the fans do not spinning. You need to deliver the gpu start signal from the mobo via riser!!!!!!

Ok my bad, but even if they do get power they won't spin up = same end result…

[wolverine5pl]

i reinstalled windows from scratch as was playing up.
and still same.
i start cast_xmr and it crashes straight away with window
cast_xmr-amd has stopped working.
any idea why??

[teamjk]

just to add one more experience with vegas if it adds on free wisdom subject

decided to build a 6 vega rig (my first one ever). knew btc from 2013, looked again at 2014-2015 and din't do anything.  i just decided to give it a try just to not regret it again for not doing something, even if i am very late  in this and I know  what i am messing with.
ordered 6 vega 64 (2  liquid from saphire and 4 gigabyte watercooled) late november-early december at amazon at 599 uk Pounds.
the last 2 gigabyte delivered mid and end december. The last 2 are to be return back to Amazon as not working ones. (problematic ones- the one crashes the system as hard as you try and the second is not recognized by windows as a gpu anymore-worked a bit in the beggining and then was lost completely-not recognized as hw anymore. tried also on second motherboard, another rig of a friend with same behaviour.)

So have spent almost a month with setup of the rig with the 4 cards (mobo Asrock btc pro r2.0, 2 *evga 1000 PSU  , win 10 pro etc). Followed steps from various sites (vega.miningguides covers most of it), installed drivers numerous times and windows from scratch 2 times to make it stable-my first ever rig, so newbie completely. had many years to deal with pc equipment). currently run a bat file to disable/enable cards, set settings by overdrivetool and run the miner of  choice.
Have a stable system now , system doesn't need reboot , runs stable for almost a week now  -will let it run to see if it crashes sometime.
(not optimum setup though, 250-255W per card , mem 1100, tp -30% and fan speed at some 2k are all the changes i have done, with overdriventool , no registry files used so far, might try changes when add a  5th nitro saphire i will get the following days)
mainly use xmrstak for etn, sumo, monero with good results. Have also tried nicehash , working  perfect on cryptonight, just can't find a reason to leave the coins to someone else.
pulling some 2010-2040 hash/sec per card  constantly. temps at 45-46 (watman readings, not using this anymore other than that) . Rarely using 2 other utilites ie GPU-z to read gpu settings like temp etc due that it affects hash power everytime i use them, so need to disable/enable the cards again.

due to the failures of the last 2 cards , i don't really want to push the remaining ones hard or to do many changes to the setup.
looking to add 2 more vega (if found anyway- i stopped asking for a replacement from amazon due to last 2 failures-not sure if i should question gigabyte reliability, or vega in general-since the last 2 cards were failures the only thing i can assume is that the cards came from returns from other customers and they were problematic from start-shame that they send them to customers)

initial target of the mining effort was  to see if a system with 6 vegas 64   can give 20 euros per day recurring revenue (on average-low set target and a ROI of some 6-8+ months).  
after less than a month ,monthly target has met . I meet the target with just 4 cards and not 6. With some lack on sumo, have some 0,1 btc (0,75 btc , along with some monero and   etn  )
For as long as it lasts, and if the whole industry doesn't collapse overnight, i am pretty sure that mining can work.
Also as a thought, i am leaning towards the high end (either vega or 1080ti) with a reasoning that even in the worst case scenario of downward trend, if there are any cards left producing even minimal revenue, these will be the top ones currently (ROI will be in red on this scenario but anyway).

Thinking seriously to build a second rig but on 1070 or 1080ti, seem to have better availability and also to diversify the mining  to other coins also.

[wolverine5pl]

almost two days stress. system reinstalled
still system is playing up
So managed to set everything up. Vegas 64 and 56 are ashing but.... after a while hashrate is dropping from 1.9-2.0 to 1.7ish
any idea why? in past i had to restart gpu few times and would be rock solid but now its problem
its not psu as i have tried different one.
using windows 10. maybye setting im missing?


Looks like it was something to do with power settings

[teamjk]

almost two days stress. system reinstalled
still system is playing up
So managed to set everything up. Vegas 64 and 56 are ashing but.... after a while hashrate is dropping from 1.9-2.0 to 1.7ish
any idea why? in past i had to restart gpu few times and would be rock solid but now its problem
its not psu as i have tried different one.
using windows 10. maybye setting im missing?

Have noticed this behaviour on mine one.s  only when I tried to run gpu-z utility while mining
But have read many users have this problem for uncertain  reason

One possible workaround is the one provided on vega. Miningguides .com site
At the section with title "vega hash drop",where there is a script that can monitor hash rate and if the drop exceeds a certain level defined by you it automatically restarts everything. Not solving the root cause, but is better than nothing

[wolverine5pl]

almost two days stress. system reinstalled
still system is playing up
So managed to set everything up. Vegas 64 and 56 are ashing but.... after a while hashrate is dropping from 1.9-2.0 to 1.7ish
any idea why? in past i had to restart gpu few times and would be rock solid but now its problem
its not psu as i have tried different one.
using windows 10. maybye setting im missing?

Have noticed this behaviour on mine one.s  only when I tried to run gpu-z utility while mining
But have read many users have this problem for uncertain  reason

One possible workaround is the one provided on vega. Miningguides .com site
At the section with title "vega hash drop",where there is a script that can monitor hash rate and if the drop exceeds a certain level defined by you it automatically restarts everything. Not solving the root cause, but is better than nothing

i went to check power savings and dissabled all and so far is working

[martyroz]

PLEASE HELP!!!

My windows 10 - 4 * Vega64 rig has always behaved in this manner;

On any given boot/reboot, it will detect a random number of cards (1-4) in roughly this proportion;
* 80% of the time it will detect 3 cards
* 10% of the time it will detect 2 cards
* 5% of the time it will detect 1 card
* 5% of the time it will detect 4 cards

This is incredibly frustrating and sometimes I have to remote in and reboot 15-25 times in a row just to get all 4 cards showing up! If I'm lucky, it will work first time.

Is this a known problem? are there any available solutions?

I have a 4 * Vega56 rig that never behaves like this, 100% of the time 4 cards are seen in Windows.

[Bakhtra]

PLEASE HELP!!!

My windows 10 - 4 * Vega64 rig has always behaved in this manner;

On any given boot/reboot, it will detect a random number of cards (1-4) in roughly this proportion;
* 80% of the time it will detect 3 cards
* 10% of the time it will detect 2 cards
* 5% of the time it will detect 1 card
* 5% of the time it will detect 4 cards

This is incredibly frustrating and sometimes I have to remote in and reboot 15-25 times in a row just to get all 4 cards showing up! If I'm lucky, it will work first time.

Is this a known problem? are there any available solutions?

I have a 4 * Vega56 rig that never behaves like this, 100% of the time 4 cards are seen in Windows.

I don't know about others. But I always use CLONED harddisk from whichever rig that run good using macrium. With that I save some time from installing each rig from scrap. Try clone your good rig.

[ss10miner]

Great! why not make it into marketable product? I am looking forward to it.

[jimmykl]

PLEASE HELP!!!

My windows 10 - 4 * Vega64 rig has always behaved in this manner;

On any given boot/reboot, it will detect a random number of cards (1-4) in roughly this proportion;
* 80% of the time it will detect 3 cards
* 10% of the time it will detect 2 cards
* 5% of the time it will detect 1 card
* 5% of the time it will detect 4 cards

This is incredibly frustrating and sometimes I have to remote in and reboot 15-25 times in a row just to get all 4 cards showing up! If I'm lucky, it will work first time.

Is this a known problem? are there any available solutions?

I have a 4 * Vega56 rig that never behaves like this, 100% of the time 4 cards are seen in Windows.

I don't know about others. But I always use CLONED harddisk from whichever rig that run good using macrium. With that I save some time from installing each rig from scrap. Try clone your good rig.

Yep that's what I do too. I keep incremental backups too before any config changes – great to be able to roll back easily.

[RIGED]

Nice thread, I have just 5 VEGA 56 in an ASUS Mining Expert rig for now. I appreciate the info on whatever that overheating component is on the motherboard. Stuff like that is a major source of hair loss! I'm sure some of the stuff you've mentioned is just 'best practice' or 'common sense' and even 'paranoia', but we need to keep all these things in mind when messing with VEGAs and 20 slot motherboards and crappy blockchain drivers...
I've noticed a lot of incorrect assumptions and entire guides/walkthroughs with incorrect information because miners are not understanding really anything about this hobby from the operating system, hardware, right through to the mining software and coins. Please just take everything said with a grain of salt -- do your own tests and verify things are behaving as expected.
I've burnt out a VEGA before just by leaving it plugged into the riser without the pci express / usb thing connected.  I heard it make a cracking sound and realized it was cooking itself a bit too late. Why would it pull power and burn itself out without even trying to spin a fan? Ask powercolor or whoever made that shitty bios. It didn't destroy the card completely, but it never performed stably at any mining after that incident.
AMD tip: make yourself a .reg file to disable the ULPS and Crossfire settings for like 10 cards at one click (21 if you have the mining expert). It won't hurt anything to run it on non-existent cards or schedule it to run (regedit -s file.reg) although I haven't gone this far myself. But it's true sometimes you will need to do this on cards you haven't touched because windows decides to mess with the drivers.
VEGA tip: Don't plug your screen into a VEGA card -- it will probably lose about 200 h/s on cryptonight and unplugging your screen again won't necessarily fix that. It often makes it much much worse. You might have to reboot. Either use another card for display or just use remote desktop or something to manage it. I ASSUME using a dummy plug will cause this problem too.
AMD tip: The order of your cards in the registry control set folders is not necessarily the same as the order of the cards in your favorite overclocking tool or the device manager. If you have all the same cards, or you use DDU methodically, or if you've never have a loose cable before, sure, it might work like this. Don't rely on it. Any time windows wants to reinstall a driver it leaves the old folder and just makes a new one. Try to verify what you're doing by listening for the fan speed change and keep your cards numbered with physical labels so you don't mix them up.
AMD tip: atikmdag-patcher.exe should be renamed to atikmdag-patcher-bios.exe first so it only fucks with ONE dll file. What it's doing is hazardous and not easily reversible should it ever fail. Windows can decide that these patched dll files, lacking a valid digital signature, will not be loaded. You can end up with a bunch of 0-byte AMD dll files in your system folder and/or the inability to run any programs including the amd driver installer or possibly even ddu. If this ever happens, you need to delete the symbolic links to the patched files from your system32 folder and then copy virgin dll files from the extracted amd driver package to system32. Don't reboot until reinstalling the drivers again, or corrupted files might be restored again by Windows.
VEGA tip: Cast XMR does some weird shit with hash rates dropping for no apparent reason after a while that can be really really hard to fix. The only thing that worked for me is to take all the non-vega cards out of the system. Having other cards in there mining other stuff or even trying to help the vegas is probably a bad idea. I don't know why, but they seem to mess up the thread synchronization in cast. It's running 2 threads per VEGA, offset by a perfectly optimized amount of time to result in the best utilization of the card's resources. Anything fighting cast for resources throws a wrench into this and will result in losing 10% hash rate until you restart cast. Don't ask me why the dev can't fix it -- I've mentioned it twice and I'm sure countless others have too.

great post thnx for this, any little tidbit of info helps myself and the community and thats what its all about in my book or should i say ledger,
yea i know my paranoia shines on a few points here but i learnt to be cautious in avoidance of being sorry

i almost fried 3 Vegas via the same method as you but was quick enough doing the reinstallation (one by one) not to affect them adversely

weird thing is i notice no hash rate degradation with Cast, but maybe that has to do with the fact that im only running Vega64s as you suggest

im a bit compulsive with uniformity when it comes to electronics and would rather buy a new mobo and parts to run any other type of GPU either standalone or with identical peers

which brings me to my next quest, i decided to dive deeper and purchased a Vega Frontier, i will be rigging it alone on fresh mobo and parts as the commencement of an all Frontier rig. i have a funny feeling again hahah, i have convinced myself that one can run 4 threads on these, or maybe two cast miners running simultaneously on one card, it arrives in a week (from Hawaii) will let you all know of the results if i dont fry it first

thanx again for your valuable post

[STT]

Whats the highest hash rates people are getting.    I did find out my 49 MH on Skunk was not 78 watts or so as reported but that was actually using 260 watts so no wonder it showing such a high hash rate.       Still confusing as the temperates were higher after reboot as well, I dont think  0.62 MH per watt is actually possible on skunk but thats why Im asking.   Watts being the card reading only, I also hash on the cpu

[RIGED]

i reinstalled windows from scratch as was playing up.
and still same.
i start cast_xmr and it crashes straight away with window
cast_xmr-amd has stopped working.
any idea why??

more info would narrow the possibilities
could be:
cast xmr dowloaded file is corrupted
a riser and or cabling is not working well
forgot to set virtual memory
using conflicting PCIE ports
your cast run.bat file has a typo
one of the GPUs used is not AMD
try running Cast as Admininstrator
check windows is not blocking it (Defender)
none of the above

[RIGED]

Whats the highest hash rates people are getting.    I did find out my 49 MH on Skunk was not 78 watts or so as reported but that was actually using 260 watts so no wonder it showing such a high hash rate.       Still confusing as the temperates were higher after reboot as well, I dont think  0.62 MH per watt is actually possible on skunk but thats why Im asking.   Watts being the card reading only, I also hash on the cpu

i managed 2260H/s on Cryptonight with 200W but it was not stable past the hour
also stable 44.5MH/s on ETH 210W (which is looking attractive again at almost $1000ETH)

havent tried anything else as ETN has been working well for me so far (mined a little SUMO too)

i really feel for all you guys having elec cost in mind when mining,
i dont as its stupid cheap here 2640W 24/7 costs no more than $45/month for a return of around $3000/month
everything else is horribly expensive save electricity ($0.04kWh) and electronics

[RIGED]

PLEASE HELP!!!

My windows 10 - 4 * Vega64 rig has always behaved in this manner;

On any given boot/reboot, it will detect a random number of cards (1-4) in roughly this proportion;
* 80% of the time it will detect 3 cards
* 10% of the time it will detect 2 cards
* 5% of the time it will detect 1 card
* 5% of the time it will detect 4 cards

This is incredibly frustrating and sometimes I have to remote in and reboot 15-25 times in a row just to get all 4 cards showing up! If I'm lucky, it will work first time.

Is this a known problem? are there any available solutions?

I have a 4 * Vega56 rig that never behaves like this, 100% of the time 4 cards are seen in Windows.

i did mention this in the earlier pages,
had the exact same problem and the mobo would recognise different GPUs each time, seldom all of them

im willing to bet this is because you have Electro Magnetic interference from a nearby source/device, a bad contact that is creating a spark in the wall socket or elsewhere for example (which is also a fire hazard), or a faulty screen or a strong transformer  anything that you can imagine that might be creating a strong enough EM field and is close enough to affect mobo and/or GPUs, heck it could be the smart meter on the other side of the wall, or your neighbour plays with stun guns or a cell antenna array pointing at your location i dunno but it sounds identical to my debacle

are you using a wifi adapter? that might be the cause also

try moving the rig (if possible) to another locale, change contacts/wiring/psu/ remove anything electric and electronic from the rig vicinity

electronics (especially precision ones like what we build)  are very sensitive to EM, keep this in mind when placing your rig

in my case it was an Ozone generator (which uses a spark gap) more than a meter away

good luck

[martyroz]

im willing to bet this is because you have Electro Magnetic interference from a nearby source/device, a bad contact that is creating a spark in the wall socket or elsewhere for example (which is also a fire hazard), or a faulty screen or a strong transformer  anything that you can imagine that might be creating a strong enough EM field and is close enough to affect mobo and/or GPUs, heck it could be the smart meter on the other side of the wall, or your neighbour plays with stun guns or a cell antenna array pointing at your location i dunno but it sounds identical to my debacle

in my case it was an Ozone generator (which uses a spark gap) more than a meter away

The thing is, it sits alongside 4 other rigs that are 100% problem free. And I observed this problem in different rooms of the house (I build them in office then move to garage).

But I will try to swap out the PSU, etc, and test that...

[STT]

Another fire hazard is dust, dont let any dust build up near to a bunch of hot cards or it will possibly provide fuel for some damage to occur.   Get out the hoover if need be.    I dont know about interference especially but some kind of cheap faraday cage around the setup might be an easy solution. 

everything else is horribly expensive save electricity ($0.04kWh) and electronics

I missed where is the electricity this cheap, you pay bulk rates I take it ?   solar or hydro power would be ideal but theres setup costs to it

[CreamyG31337]

I think he said Norway. I'm a bit jealous. Not just for cheap power, but they did a lot better job managing their resources there than where I live.
http://www.huffingtonpost.ca/2017/09/19/norway-s-oil-fund-hits-1-trillion-meanwhile-in-alberta_a_23215451/
(I live in Alberta)
My Vega's are stable at about 1950 H/s each on cryptonight using cast xmr and stock bios, basically set up as per http://vega.miningguides.com/
I'm not sure what else to throw in that rig -- maybe will just hold out for when the custom vega cards actually show up to purchase in Canada.

https://hothardware.com/news/xfx-officially-unveils-custom-radeon-rx-vega-56-and-vega-64-double-edition
https://hothardware.com/news/gigabyte-custom-radeon-rx-vega-64-gaming-oc-8g-windforce-2x-cooling
https://hothardware.com/news/powercolors-cutstom-radeon-rx-vega-64-red-devil-overclocked-preorder
https://hothardware.com/news/sapphire-announces-nitro-radeon-rx-vega-64-and-vega-56  (this one I don't like -- 3 power connectors is unnecessary according to a review that checked the max current draw)
https://hothardware.com/news/asus-radeon-rx-vega-64-four-air-water-cooled-models