[HELP] Issue with some sapphire tech rx 580 SE 8gb gpus(good challenge)

[CryptoWatcher420]

rig specs:
h81 pro btc r2.0
8gb ram
80gb ssd
6x rx 580 SE 8gb gpus (stock settings)(3 on turbo,3 on silent)
pci express powered risers
1500w hp server psu & breakout board
180w pico psu
windows 10 pro build 1709
just over 17gb pagefile


okay so the problem I am having is windows detects all 6 gpus just fine, but when I goto run the claymore zcash version 12.6 it detects all 6 but does not hash with 1 of them
some of the more simple things ive tried are restarting the mining program which causes it to most of the time change which 1 gpu it does not want to hash, but 5 others hash just fine
this has persisted over 2 different motherboards by asrock, the h81 and a z97, ive changed risers, still persists, ive even gone as far as to change the OS from windows pro to the enterprise version and still this perplexing issue shows

I need some fresh ideas maybe ive missed something
please note Ive tried all the basic things from pagefile changes, mobo settings for pcie and other various basic things to be checked, so please do not bother mentioning things ive Already tried, fresh ideas ONLY

if anyone who has these specific gpus and has come across this issue any help would be greatly appreciated!

[1714151531]

1714151531

[1714151531]

1714151531

[CryptoWatcher420]

could really use some help, ive tried all that I can think of to do

[CryptoWatcher420]

geeze nobody?

[avalon44]

Hi, did you try mining a different coin? What happens if you mine ETH with claymore or any other mining tool?

How many memory modules do you have for the 8GB? Only 1x 8gb or 2x 4gb?

[Bigpiggy01]

Hi, did you try mining a different coin? What happens if you mine ETH with claymore or any other mining tool?

How many memory modules do you have for the 8GB? Only 1x 8gb or 2x 4gb?

Seconded, try running another miner and see if the bug persists.

Also what drivers, exact OS etc are you running? There have been a couple of instances of issues with 1709 win builds that have been sorted by switching to the overnight builds.

[CryptoWatcher420]

Hi, did you try mining a different coin? What happens if you mine ETH with claymore or any other mining tool?

How many memory modules do you have for the 8GB? Only 1x 8gb or 2x 4gb?

Seconded, try running another miner and see if the bug persists.

Also what drivers, exact OS etc are you running? There have been a couple of instances of issues with 1709 win builds that have been sorted by switching to the overnight builds.

tried a few different builds, I have builds 1709, 1703 and 1607

currently the 1709 is installed, I tried the 1703 build but same problem persisted

its using latest driver version, ive switched to an earlier one and still same thing

windows that's on it right now is windows 10 pro v1709, I have to grab the rest of the build version numbers

as for the ram its got a dual channel kit so 2x 4gb sticks

[CryptoWatcher420]

Hi, did you try mining a different coin? What happens if you mine ETH with claymore or any other mining tool?

How many memory modules do you have for the 8GB? Only 1x 8gb or 2x 4gb?

doesn't make much sense to try a different pool or coin, others have rx 580's just gotta find someone with somewhat the same issues as I have
rig runs fantastic with just 5 gpus, just not with that 6th one. windows picks it up and installs it just fine, even claymores zcash miner picks up the 6th gpu but just doesn't HASH with it which leads me to believe its a not enough power issue same with not being able to run all the rx 580 8gb SE's in turbo. either gpus don't like it or theres not enough juice, I have 3 in turbo and 2 that are not its the strangest thing, I ran into the same kinda thing with the rx 480's when trying to do 6 on a 1200w psu, un-modded and stock I had to up the psu to something bit higher wattage to run all 6 on stock settings

[No.15]

I had this problem once in boinc where 1 card would not be detected by the program.  I had to go into the bios and set "decode above 4g" to on (not auto) and they all worked after that.

It is a shot in the dark but it may be worth trying.

[CryptoWatcher420]

I had this problem once in boinc where 1 card would not be detected by the program.  I had to go into the bios and set "decode above 4g" to on (not auto) and they all worked after that.

It is a shot in the dark but it may be worth trying.

you do know that the asrock h81 pro btc r2.0 mobo doesn't have that option in the bios, please people do not mention things like this when ive already stated what mobo I'm using, if you don't know about the h81 pro btc r2.0 you have no reason to post here, I'm looking for certain people that have these gpus and a people having simular issue to what I'm dealing with

thank you:D

[CryptoWatcher420]

bump

[cvsea]

crossfire disabled?

[avalon44]

Try removing one of the memory modules. Could be an interrupt problem.

[xxcsu]

I think you tried already a lot of things ...
Waht i would do .... is to find a non hashing card and place in another computer...
Start the miner program let in run for a while , and check which gpu fans are not spinning , or just check the temp with yuor hand or any infrared thermometer . Remove that card ...
I just built 2 rig  each rig have 8 of those cards what you have . I had a problem with one of the card. It was hashing  way under 20MH/s on ETH out from box.
Took the card of from the bulit , placed in my "test bench" pc , found out this is the card what have Samsung and Hynix memory , it was the hardest for me to find a right timing , overclocking/undervolting settings for the gpu ...

To me its looks like you have a power issue with your rig , i might wrong ..

What memory your problematic card have ?
Samsung / Hynix / Micron / or mixed memory on the card ?

Its always the same card what hashing with 0MH/s ?
( try switch around pcie slots , and see if the same card is the non hashing card )

Do you have a monitor or dummy plug connected to that non hashing GPU ?

Did you get the cards from a trustable retailer ?
(Soemone might aleady screwed up the bios and sent that card back , reflahsed the card with the wrong bios before he sent it back )

Did you ever tried the card in another computer ?

Did you tried your rig with another psu ?
1500 watts PSU should be ok with 6 RX 580 but you are pretty much pulling the max from that psu , if you dont undervolt your gpus and your mem's .

Did you tried to place the card on antother rail on the psu ?

Did you set the virtual memory to 16GB minimum ?

Do you know what bios versions yor cards have ?
( i got my cards with a few different bios version numbers : 113-BE366EU-Z48 , 113-1E366CU-S5S , 113-1E366CU-S52 adn a few more )

[fanatic26]

You should be running linux to mine, mining on windows is asking for headaches.

Have you tried taking off all cards but the troubled one to make sure it can still hash on its own?

Have you tested all the individual components? (riser, cables, PCIE slot, etc)

[CryptoWatcher420]

I think you tried already a lot of things ...
Waht i would do .... is to find a non hashing card and place in another computer...
Start the miner program let in run for a while , and check which gpu fans are not spinning , or just check the temp with yuor hand or any infrared thermometer . Remove that card ...
I just built 2 rig  each rig have 8 of those cards what you have . I had a problem with one of the card. It was hashing  way under 20MH/s on ETH out from box.
Took the card of from the bulit , placed in my "test bench" pc , found out this is the card what have Samsung and Hynix memory , it was the hardest for me to find a right timing , overclocking/undervolting settings for the gpu ...

To me its looks like you have a power issue with your rig , i might wrong ..

What memory your problematic card have ?
Samsung / Hynix / Micron / or mixed memory on the card ?

not sure what mem it has

Its always the same card what hashing with 0MH/s ?
( try switch around pcie slots , and see if the same card is the non hashing card )

not always the same gpu, if I close the claymore zcash miner and open it back up usually its a different gpu that's not wanting to hash

Do you have a monitor or dummy plug connected to that non hashing GPU ?
nope, its connected via riser to one of the x1 slots

Did you get the cards from a trustable retailer ?
(Soemone might aleady screwed up the bios and sent that card back , reflahsed the card with the wrong bios before he sent it back )
newegg

Did you ever tried the card in another computer ?
yes, actually I had a different mobo that the gpus were on before this one and same problem

Did you tried your rig with another psu ?
1500 watts PSU should be ok with 6 RX 580 but you are pretty much pulling the max from that psu , if you dont undervolt your gpus and your mem's .
yes tried with 1200w atx psu, 1400w server psu and with it is sitting with now is the 1500w server psu

Did you tried to place the card on antother rail on the psu ?
its a server psu

Did you set the virtual memory to 16GB minimum ?
yup its actually at a little over 17gb

Do you know what bios versions yor cards have ?
( i got my cards with a few different bios version numbers : 113-BE366EU-Z48 , 113-1E366CU-S5S , 113-1E366CU-S52 adn a few more )

not without checking, I assume id have to load up a program to find that out or is it on a sticker on the gpu?

how many watts total would you say one of the rx 580 8gb SE's from sapphire tech use? in turbo and in silent, seems like its close to around 250w per gpu with NO undervolt


btw I too see this as theres not enouh juice for all the gpus because there not undervolted, having someone else say the same thing does confirm this might be the culprit, I do have an extra psu atm, so I think this next step is to separate couple of the gpus from one psu and put it on a second one,
and see where that gets me

[xxcsu]

May i ask why you dont want to undervolt your VGA cards ? Why are you mining with stck settings ?
Its very easy and simple , well.... a little bit hard to find a stable numbers , but once you have it , you going to have a stable rig.
Try to mine ETN or Monero without undervolting your cards , mining Monero or ETN using much less power .
My RX 580 Nitro+ ( custom bios , undervolting , overclocking applied ) using around 90-100 Watts at the wall mining Monero or ETN (830H/s) , 120-130 watts mining ETH (30,5-31.2 MH/s), and around 150 watts during dual mining ETH/DCR ( 29.2MH/s / 925MH/s )

[CryptoWatcher420]

May i ask why you dont want to undervolt your VGA cards ? Why are you mining with stck settings ?
Its very easy and simple , well.... a little bit hard to find a stable numbers , but once you have it , you going to have a stable rig.
Try to mine ETN or Monero without undervolting your cards , mining Monero or ETN using much less power .
My RX 580 Nitro+ ( custom bios , undervolting , overclocking applied ) using around 90-100 Watts at the wall mining Monero or ETN (830H/s) , 120-130 watts mining ETH (30,5-31.2 MH/s), and around 150 watts during dual mining ETH/DCR ( 29.2MH/s / 925MH/s )

tbh rigs have been pretty problem free using stock, that and if im gunna do a UV, might as well do it via bios mod but thats just that much more to mess with, and everyones opinion is different and what information there is, is scattered

honestly i just dont know a good way to go about it. what would you say are some good settings for the blue shrouded rx 580's?
i know of alot of the programs, just not how to use them efficently

[CryptoWatcher420]

bump

[xxcsu]

tbh rigs have been pretty problem free using stock, that and if im gunna do a UV, might as well do it via bios mod but thats just that much more to mess with, and everyones opinion is different and what information there is, is scattered

honestly i just dont know a good way to go about it. what would you say are some good settings for the blue shrouded rx 580's?
i know of alot of the programs, just not how to use them efficently

the only thing i would do with bios mod is change the mem straps , and leave the undervolting/overclocking to a software , preferably OverDrivenTool 2.4 . If you "burn in" the voltage / frequencies to the bios you might going to get into trouble later , troubleshooting maybe going to be much harder.

I might can help you modding your bios , and sending you the settings what my cards using. But you need to learn how to use OverDrivenTool . Its very simple , its can start with windows and automatically reset your cards after each boot , and set back the tuned settings for each cards .
I need to know your bios version number.

You always have to make sure , you are flashing the same bios version back to your cards (after mem straps changes) , what they have from factory !!!

There is a example for a bad flash ...

[CryptoWatcher420]

id have to check each gpu for that since so far ive come across Hynix,samsung and micron memory with these sapphire tech rx 580 8gb SE's.

could i getcha to join my discord channel i run for crypto? would make getting instructions and questions easier and faster. i have a invite link in my SIG area

[CryptoWatcher420]

bump

[blackpirl8]

Resolve this yet?

all pcie slots set to "Gen1"?  (including the 16x)?

I to run an Asrock board with Sapphire 8GB SE Micron and have odd issues.  (even stock)
With the asrock h110 pro btc+ and Windows 10 (1703 or 1709), I can only mine if I have the Integrated GPU on.  When I set to disabled, my RX580's lose "OpenCL" which is required to mine.
Currently have 8 of these identical GPU in the rig on Silent mode, and can only mine for approx 6 hours before the AMD driver crashes.

This might help your situation?  See latest Beta Bios revision as of Feb 27
http://www.asrock.com/mb/Intel/H81%20Pro%20BTC%20R2.0/#BIOS

[CryptoWatcher420]

Resolve this yet?

all pcie slots set to "Gen1"?  (including the 16x)?

I to run an Asrock board with Sapphire 8GB SE Micron and have odd issues.  (even stock)
With the asrock h110 pro btc+ and Windows 10 (1703 or 1709), I can only mine if I have the Integrated GPU on.  When I set to disabled, my RX580's lose "OpenCL" which is required to mine.
Currently have 8 of these identical GPU in the rig on Silent mode, and can only mine for approx 6 hours before the AMD driver crashes.

This might help your situation?  See latest Beta Bios revision as of Feb 27
http://www.asrock.com/mb/Intel/H81%20Pro%20BTC%20R2.0/#BIOS

no its not resolved, ive tried setting the slots to gen1 and I tried gen 2 which made no difference on the asrock h81 pro btc R2.0 mobo, also tried with and without the intergrated gpu still no difference, pretty sure why I cant run 6 is because of not enough juice for the 6th gpu. 5 run just fine no problems what so ever, even runs for months no issues

I believe I have the feb bios from asrock on the mobo already, about 95% sure lol and didn't make any difference lol

[blackpirl8]

Do you have a power meter connected to see how much power you're draining while mining?
I think you may be correct about the power supply issue.

I believe that on Boost, cards will drain 250W per card
Silent when I tried, GPU-Z gave 150-160W to the GPU, but that would be higher to the wall outlet.
Just there, that would be 1200W approx.
That leaves 300W for risers, SSD, motherboard, fans, etc.  (Anything electrical should only run 75-80% max power)

I just googled "180w Pico PSU" as I have never heard of that.  What exactly is this connected to on the rig?

Things I would try:
1. Run all of the cards on Silent instead of (3) on Boost and (3) on Silent?  That would lower the watts.
2. Ensure that your Windows 10 Power Management is set to "High Performance" and ensure the Hard Drives Never shutdown, screen never shuts down, PCIe slots can't be managed, etc (the regular stuff)
3. Review PCIE/SSD/HDD in "Device Manager" and ensure that the Power Management is disabled on all of these drivers

[vuhoang1giang]

I was have same Issue like you. And my choice that time are sell all @@

[Midcrypto]

You should be running linux to mine, mining on windows is asking for headaches.

Have you tried taking off all cards but the troubled one to make sure it can still hash on its own?

Have you tested all the individual components? (riser, cables, PCIE slot, etc)

I agree with this. Put SMOS (Simplemining OS) on a flash drive and boot from that. It if detects and runs with all 6 cards right away you know it's a Windows issue.

[CryptoWatcher420]

Do you have a power meter connected to see how much power you're draining while mining?
I think you may be correct about the power supply issue.

I believe that on Boost, cards will drain 250W per card
Silent when I tried, GPU-Z gave 150-160W to the GPU, but that would be higher to the wall outlet.
Just there, that would be 1200W approx.
That leaves 300W for risers, SSD, motherboard, fans, etc.  (Anything electrical should only run 75-80% max power)

I just googled "180w Pico PSU" as I have never heard of that.  What exactly is this connected to on the rig?

Things I would try:
1. Run all of the cards on Silent instead of (3) on Boost and (3) on Silent?  That would lower the watts.
2. Ensure that your Windows 10 Power Management is set to "High Performance" and ensure the Hard Drives Never shutdown, screen never shuts down, PCIe slots can't be managed, etc (the regular stuff)
3. Review PCIE/SSD/HDD in "Device Manager" and ensure that the Power Management is disabled on all of these drivers

the pico psu powers the motherboard and a ssd, its just a super mini version of a normal psu with the exception that the pico psu's input volts is 12v

[Mohondoa]

    Generaly what i know about your problem came from GPU memory error or too much over clock setting, so at the console cmd miner software the gpu hashrate show is zero (0.000 M h/s).
    Try to use stock bios and don't use any OC tools, if there's show hashrate, probably your problem is OC setting. This is from my experience

[blackpirl8]

    Generaly what i know about your problem came from GPU memory error or too much over clock setting, so at the console cmd miner software the gpu hashrate show is zero (0.000 M h/s).
    Try to use stock bios and don't use any OC tools, if there's show hashrate, probably your problem is OC setting. This is from my experience

Cryptowatcher already said he is running stock.  These video cards have 2 settings (A dip switch).  3 on Boost (default) and 3 on Silent

[skablast]

My 2 cents:

"just over 17gb pagefile"

6*8 = 48

can you make one pagefile of at least 32 gb ?

I would say that if 5 work well with 17, 6 will work with 20 if you can't make it 32.

[CryptoWatcher420]

My 2 cents:

"just over 17gb pagefile"

6*8 = 48

can you make one pagefile of at least 32 gb ?

I would say that if 5 work well with 17, 6 will work with 20 if you can't make it 32.

pagefile means dick, come back with something better, my rigs all run 8gb ram anything over 17gb pagefile is pointless, if that were needed I would have seen the same issue with the rx 480's I have from the same company. now if I were running rigs with 4gb ram sure I could see that, but I'm not. fact is more RAM you have the less you need a pagefile. pc's use mobo ram before pagefile. as it stands 8gb ram + 17gb pagefile is 25gb of memory the gpus can allocate, so please why don't you explain where a 32gb pagefile is going go have ANY affect in this very specific situation? look folks I'm not looking for your danm 2 cents, most/all of the common things ive tried some many times over just to check. if you don't have information that is of some rarity then it needs not to be mentioned, not to sound like an ass here but I'm looking for specific information and not just the everyday stuff.

thanks crypto

[skablast]

I don't know what you did and haven't written here.
It was just something if it was mine I 'll try and I'm lazy to double check if it work as I have exactly( besides the PSU) the same configuration, same card, same windows you have and it work fine with all.
that was my 2 cents


there is only one thing I can bet 1000 $ on:

[/quote], not to sound like an ass
thanks crypto
[/quote]

Yes, that's exactly how it sound

have a nice day

[CryptoWatcher420]

I don't know what you did and haven't written here.
It was just something if it was mine I 'll try and I'm lazy to double check if it work as I have exactly( besides the PSU) the same configuration, same card, same windows you have and it work fine with all.
that was my 2 cents


there is only one thing I can bet 1000 $ on:

, not to sound like an ass
thanks crypto
[/quote]

Yes, that's exactly how it sound

have a nice day

[/quote]

tbh if you had really read my original post, you should have gotten the impression that im not looking for some noobs 2 cents, im looking for a specific information on a issue that its clear some noob cannot answer as you have clearly demonstrated, sorry if you thought i was an ass but im not here to please you in the form of words, im super blunt so if people repeatedly give me useless information or tell me to do something IVE ALREADY DONE, youd eventually get irritated too its as simple as that. in my original post i state ive already done all the normal stuff, and some of it twice over for good measure but i guess you missed that part?

[skablast]

As I stated, It happen that I have the very same configuration.Hw & sw beside Card are mod SRBpolaris- pimp straps and I have different PSU.
I was unable to replicate your error resetting the card to stock(but straps still modded). That lead me to think your problem is there or in the PSUs.

- Run more instance of Claymore starting them in different orders is the only out of the box  idea I can give you in order to identify if its a claymore/driver issue or not

tests and comments finished