Topic: ASrock H110 Pro BTC+ 13 GPU Board - page 3. (Read 54645 times)

Hi got some problems too with this board, 

I got one rig with 10 gpus (rx580, 560,  2 vega 56) - no problems at all.  all gen2, onboard graphic etc..

I mount another rig with 10 vega 64 but only can connect 9, if i installed the 10 gpu, i have no post on poweron..

Any ideia too what could i check?

I changed to 16 gb ram its the same, have a lot of power too..

No problems in risers too,

my solution was to put one vega in my desktop pc.. but its frustating

Regards,

Anyone had problems upgrading the BIOS to 1.60? I just did it with a Celeron G4900 and the system will not power on now. Intel ME and BIOS flash all went to 100%, the system powered itself off and now the power supply clicks on, fans spin for a second or two and then shut off. I'm going to try and pull the battery with the CMOS reset jumper set for a few hours. The board was fine before the BIOS update albeit with a DPC watchdog crash during Windows 10 1709 install hence attempting the BIOS update.

So I set up page file and all is golden. I thought that was needed only for mining, but lesson lernt I guess.

I was having issue with this board and windows, that i had black screen on rdp, black screen on teamviewer and same on vnc some time after login.

I use iGPU as primary, no gfx in 16x slot, 12 gpus in other slots.

I reinstalled latest windows clean, did offline install, so it could not get drivers from windows update and I was getting black screen and all the hell broke lose, weird errors and stuff.
I reinstalled latest windows clean, did online install, and all good, until I installed latest nvidia drivers, now it does boot into windows, gets ip, and then only black screen.

So disconnected all 12 nvidia cards and it booted into windows no issue, ofter i connect even just one card, it does not boot into windows.

Also I updated bios.

All my other rigs run on biostars with no issues, all 12 gpus amd or nvidia rig.


but at least on previous windows 10 it was mining, but only in black screen so I could not see temperatures and stuff and generally it sometimes went don't, so something is off. But in bios or in windows installer i do get graphics output, no issues, no artifacts.

I don't want to  go linux route as this would be my first linux rig, I don't want to bother with all the fake X server displayes to get nvidia overclocking and stuff.


Update: so another offline install a and now I get image after reboot, but after pressing start or starting edge, i get code 43 on all cards, including iGPU, also wallpaper flashes for blue color on logonscreen if i don't log in - i suppose this is while driver crashes

I have several 10/12 GPU-rigs running on this mobo with no issues. Default BIOS-setting except boot on power loss.

Can someone who has this mother board Bios settings working for more than 8 GPUs share the settings?

If you have like 10 nVidia (my case) or similar (> same cards.

Thx

just an update.
after the 100th reboot, I managed to run 10 nVidia cards without crashing when starting mining.
Now testing how long it will last, my guess is no too much.

I did not do anything to the system so not sure what changed and I do not like it since I have no idea what is going on.

Anyway, will try to play with the PCIe slot GEN 1,2 settings in the BIOS to see it makes a difference as soon as the WIN 10 system crash again !

I use Kolink Platinum 1200 Watts.
The cards are distributed equally power wise on the 2 PSU around 600 watts each.
So there is no power issue.

As I mentioned before everything was working fine before I add the 1070 and 1080.
If I try to scan for new Adapters in the Display Manager it crash the system, no load applied to cards i.e. no mining, which indicates something is in conflict at the bus I/O level, maybe same IRQ are assigned??

At the bios I have PCIe in auto, I can try to specify Gen1 or Gen2, not sure.

Any other suggestions is appreciated, before I switch to Linux !
 

which PSU you are using,  a single rail or multi-rail ?? If you use multi-rail, make sure the split rail is used ideally with each rail not exceeding the maximum power.
And You should try this. Add another PSU that only use for a motherboard, and I think 350w is more than enough.

I am trying this:
https://lh6.googleusercontent.com/Spn0i0HaZoMeRFcIP6P5Gp9wpwqRpqsX53X1sFYGPx3RF5InOuqKAFSNor4XXoLMd63CP_79SV-wxXh-1CxSa9pWkIx_TdUa8vcc9dlz-Ylcf8e863XJr-nEdXoWaY-3ID2QoGDw=s800

no luck!

I noticed some cards have the same IRQ which should not be the case, anyone check this?

I have tried all your suggestions.
About the video cards connection are ok as it was working fine with 8 cards. All the slots are populated one after the other.

Do you know software that gives me more details about the bus/cards information?

have you tried to uninstall your display drivers with DDU?

what's with additional power plugs (2 molex and 1 sata), are they connected to MB? did you insert your x1 riser boards to MB slots one after other, beginning from CPU side and to the bottom, how it's stated in the manual?
I've seen some issues when the risers were connected not one after other but with spaces between them, this MB really doesn't really likes the cases when white and black PCIe slots connections are not "balanced".

yes! I am pretty careful at distributing power on all cables and connection.

My guess at this stage is that there is a conflict on the bus/drivers attributed to cards but win 10 does not show it.
Maybe using different slots but ASRock does not mention this.

this is why " if I try to scan for device changes on the display adapters in the device manager of win 10, it crashes the system !?"

I mean I can crash the system without mining, just scan for new video adapters

Update: tried a new miner with CUDA 9.2 support, no luck!

Did you follow this rules?
- max 1 sata riser per single sata cable
- max 2 molex risers per single molex cable

risers and usb are ok as everything was working fine before and new hardware works fine, tape is there.
Power supply is more than enough as I have less than 50% load on each one.
 

first of all - check and maybe replace your risers and riser USB cables. I bet that the problem lies there. bad cables and bad risers can cause all kinds of instabilities.

you can also wrap the PCIe x1 riser boards in electrical tape - otherwise they can touch and short each other when they're inside the slots, I've encountered issues caused by these shorts.

and secondly - consider upgrading your power supply. 2400W PSU is a little bit weak to run all these cards. lowering the PL can help, but it will also hurt your hashrate. you can check the voltage under load on the 12V PSU output, it shouldn't drop lower that 11.9-12V.

Sorry I forgot, no power issues:
I have 2 1200Watts supply, power load equally distributed pulling 1100W for all cards on the 2 ps.

Beside the OC which I tested various configuration, I set 80% PL on 1060/1070 and 75% PL on the 1080.
I tried setting core and memory OC to 0 or even negative but the random card crash occurs.

Currently one card (1070 crashed) and others are running, not sure until when, will update.

If you have any ideas, like settings in Win 10 or cards settings let me know.
I am waiting for z-enemy to release 1.12 with Cuda 9.2 supports to see if it helps.

One more thing, when a card crash it is no more accessible via OC software even though is visible in the devices list of Win.
So to gain it back I need to restart the system. Is there a way to get it back with the need to restart?


ok just a strange thing: if I try to scan for device changes on the display adapters in the device manager of win 10, it crashes the system !?

It's a power problem for sure. If you could tell us which psu you are using and how you connected your gpus, we can help you better ;-)

What's your power situation look like? Sounds like power if the only thing you've done is add GPUs to an existing, working system.

Hello,

I have this MB with the following configuration:
Asrock H110  pro BTC+
Celeron 2.9 Ghz
8 GB Ram DDR4
120 Gb SSD
Virtual Mem 60 Gb

I installed 5 Zotac 1060 6Gb, 4 Zotac 1070 Ti and 1 Zotac 1080 Ti. I recently added the 1080 and 1 1070 Ti and getting an unstable system with cards crashing.
Until I had 5 1060 and 3 1070 ti everything worked fine and as soon as I added the 1070 and 1080 things got flaky.

I am on the latest nVidia driver and the latest Windows 10 update.

Mining using z-enemy 1.11 (cuda 9.1) miner on X16r.

I tried to underclock the cards to see if it helped but no luck as cards crash anyway.

Any suggestions beside moving to a Linux OS ?

Thx