Bitcoin Forum
June 15, 2024, 08:07:12 AM *
News: Voting for pizza day contest
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: H81 PRO BTC - unstable with more then 4 GPU - Win 10 [Closed]  (Read 820 times)
HerbPean (OP)
Legendary
*
Offline Offline

Activity: 1638
Merit: 1005



View Profile
November 24, 2016, 06:24:33 PM
Last edit: November 25, 2016, 06:25:48 PM by HerbPean
 #1

Hey guys,

I'm just throwing this here wondering if someone has a special tricks to help get my H81 to handle more then 4 cards. This isn't a detection problem, all 6 cards will be detected and if i'm lucky they will all hash but most of the time, GPU hang or i'm even able to start all 6 at the same time. I have this kind of behavior with 4 cards but it is rare. (I have like 4 days of uptime at the moment, no issues)

My setup is => 6 1x to 16x powered riser (via PCI-E). All 6 card are rx 470.

I tried to reset the CMOS, use different version of the Crimson drivers, reformatting win 10 etc but no luck. Set the PCI to Gen 1 or Gen 2 or Auto. Well i tried pretty much everything. I though it was a card problem but it is not. I have 2 H81 pro BTC and they do the same thing. They were used so i'm starting to think they are faulty for some reason.

Well, i'm just asking.

I will be running 4 cards only for now. (Which piss me off)

Thanks ! Cheesy
Xdragon
Hero Member
*****
Offline Offline

Activity: 543
Merit: 500


View Profile
November 24, 2016, 06:45:48 PM
 #2

Do you use USB extenders?  If not, try it.
joblo
Legendary
*
Offline Offline

Activity: 1470
Merit: 1114


View Profile
November 24, 2016, 06:50:42 PM
 #3

Power.

AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
https://bitcointalk.org/index.php?topic=5226770.msg53865575#msg53865575
BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,
HerbPean (OP)
Legendary
*
Offline Offline

Activity: 1638
Merit: 1005



View Profile
November 24, 2016, 09:18:52 PM
 #4

Do you use USB extenders?  If not, try it.

I tried, not on all 6 cards tho.

Why would it change something ? Just wondering

Thanks
HerbPean (OP)
Legendary
*
Offline Offline

Activity: 1638
Merit: 1005



View Profile
November 24, 2016, 09:19:28 PM
 #5

Power.

tried with 2 RM1000 and 1 Seasonic 1250
toptek
Legendary
*
Offline Offline

Activity: 1274
Merit: 1000


View Profile
November 24, 2016, 09:20:54 PM
Last edit: November 24, 2016, 10:12:08 PM by toptek
 #6

Hey guys,

I'm just throwing this here wondering if someone has a special tricks to help get my H81 to handle more then 4 cards. This isn't a detection problem, all 6 cards will be detected and if i'm lucky they will all hash but most of the time, GPU hang or i'm even able to start all 6 at the same time. I have this kind of behavior with 4 cards but it is rare. (I have like 4 days of uptime at the moment, no issues)

My setup is => 6 1x to 16x powered riser (via PCI-E). All 6 card are rx 470.

I tried to reset the CMOS, use different version of the Crimson drivers, reformatting win 10 etc but no luck. Set the PCI to Gen 1 or Gen 2 or Auto. Well i tried pretty much everything. I though it was a card problem but it is not. I have 2 H81 pro BTC and they do the same thing. They were used so i'm starting to think they are faulty for some reason.

Well, i'm just asking.

I will be running 4 cards only for now. (Which piss me off)

Thanks ! Cheesy

on new egg under question ans answers they say using over three cards, to use USB powered raiser cards the MB can't handle it on it's own . i would use them any way the riser kits don't cost much and come with everything you need and  are safer then using a MBs direct power.

FROM new egg

There is a BIOS setting to turn off the "plug in the 4-pin connectors" boot-up splash screen. If you are connecting more than 3 gpus, I would recommend that you use separately powered risers. Just the two 4-pin connections are not enough to support the total power draw of that many cards. If you are already using powered risers for all the gpus, you do not need to connect anything to the 4-pin cnnectors - just disable the warning screen.


Set the GEN to auto plug each one in one a at time, i wouldn't use after burner with that many cards if you are, if heat, is gonna be worry find  some thing like fan speed that only controls the fans . run each card by it self  for 15 mins each then if it keeps crashing or under hashing  after  all six are plugged reinstall the drivers with all six plugged in  . and anything else that was monitoring the cards uninstall  and run with the drivers only for a while . ? .  or do a clean install on the PC/rig doing a clean install does wonders .

other wise if you have done all that then send them back if you can ? . I did read some reviews of some having issue with that board some place it wasn't on new egg .


last thought is IOMMU on or off in the MB bios if it's off try with it on and on turn it off .


Set the VM without or with IOMMU on or off . to at least 16 gb like Claymore read me says to do with Multi cards.

For multi-GPU systems, set Virtual Memory size in Windows at least 16 GB:
"Computer Properties / Advanced System Settings / Performance / Advanced / Virtual Memory".

That's any way not just Using CM .

For security, your account has been locked. Email acctcomp15@theymos.e4ward.com
Xdragon
Hero Member
*****
Offline Offline

Activity: 543
Merit: 500


View Profile
November 24, 2016, 09:57:40 PM
 #7


Just try with  USB extenders on all cards. My  friends also had trouble until they put usb extenders on all cards. This is just case with Polaris. Also on other MB big problems with AMD RX cards, while old cards works great.
HerbPean (OP)
Legendary
*
Offline Offline

Activity: 1638
Merit: 1005



View Profile
November 24, 2016, 10:37:08 PM
 #8

Hey guys,

I'm just throwing this here wondering if someone has a special tricks to help get my H81 to handle more then 4 cards. This isn't a detection problem, all 6 cards will be detected and if i'm lucky they will all hash but most of the time, GPU hang or i'm even able to start all 6 at the same time. I have this kind of behavior with 4 cards but it is rare. (I have like 4 days of uptime at the moment, no issues)

My setup is => 6 1x to 16x powered riser (via PCI-E). All 6 card are rx 470.

I tried to reset the CMOS, use different version of the Crimson drivers, reformatting win 10 etc but no luck. Set the PCI to Gen 1 or Gen 2 or Auto. Well i tried pretty much everything. I though it was a card problem but it is not. I have 2 H81 pro BTC and they do the same thing. They were used so i'm starting to think they are faulty for some reason.

Well, i'm just asking.

I will be running 4 cards only for now. (Which piss me off)

Thanks ! Cheesy

on new egg under question ans answers they say using over three cards, to use USB powered raiser cards the MB can't handle it on it's own . i would use them any way the riser kits don't cost much and come with everything you need and  are safer then using a MBs direct power.

FROM new egg

There is a BIOS setting to turn off the "plug in the 4-pin connectors" boot-up splash screen. If you are connecting more than 3 gpus, I would recommend that you use separately powered risers. Just the two 4-pin connections are not enough to support the total power draw of that many cards. If you are already using powered risers for all the gpus, you do not need to connect anything to the 4-pin cnnectors - just disable the warning screen.


Set the GEN to auto plug each one in one a at time, i wouldn't use after burner with that many cards if you are, if heat, is gonna be worry find  some thing like fan speed that only controls the fans . run each card by it self  for 15 mins each then if it keeps crashing or under hashing  after  all six are plugged reinstall the drivers with all six plugged in  . and anything else that was monitoring the cards uninstall  and run with the drivers only for a while . ? .  or do a clean install on the PC/rig doing a clean install does wonders .

other wise if you have done all that then send them back if you can ? . I did read some reviews of some having issue with that board some place it wasn't on new egg .


last thought is IOMMU on or off in the MB bios if it's off try with it on and on turn it off .


Set the VM without or with IOMMU on or off . to at least 16 gb like Claymore read me says to do with Multi cards.

For multi-GPU systems, set Virtual Memory size in Windows at least 16 GB:
"Computer Properties / Advanced System Settings / Performance / Advanced / Virtual Memory".

That's any way not just Using CM .

Yeah All my ribbon cable riser are powered by a single PCI-E. I would have believe it's better then using the USB one. I Also tried to  plug both molex on the MOBO while all the riser avec also powered with no luck.

I have over 25GB of VM.

I will try => "last thought is IOMMU on or off in the MB bios if it's off try with it on and on turn it off ."

Thanks for the info !


HerbPean (OP)
Legendary
*
Offline Offline

Activity: 1638
Merit: 1005



View Profile
November 24, 2016, 10:37:49 PM
 #9


Just try with  USB extenders on all cards. My  friends also had trouble until they put usb extenders on all cards. This is just case with Polaris. Also on other MB big problems with AMD RX cards, while old cards works great.

Thanks !
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!