Topic: Asrock H110 PRO BTC+ 13 RTX cards cuda errors (Read 267 times)

m2 to pcie adapter  I have some amd boards with more than 6 GPU´s.

Thanks all. I have tried Hiveos in the past and one of the main issues was the overclocking and power usage settings. I will figure out how to best configure those. I just need to make the time since I'm busy.

If you read this forum, then you will find all the tinctures and recommendations you need.
https://forum.hiveos.farm/
The main problem with Windows is in the inconvenient administration of many mining farms, and in Hive OS and Rave OS all the necessary software is installed, and you do not need to spend a lot of time setting up and searching for the necessary software (for example, for BIOS firmware, etc.)

If you really want to move to a stable mining rig than you can migrate to HiveOS.It is not difficult and there are a lot of Youtube videos regarding such setup,a lot of them are over 20 minutes long and very detailed.Migrating by following these HOWTO-s will be as easy as 1,2,3 for everyone.

For sure HiveOS is great but for me I guess I feel I have more flexibility with windows although not as stable. I also haven't migrated to Hiveos because I don't know enough about setting it up correctly and getting the most out of my rig.

I don't understand why people stuck with windows OS for mining most especially those that have over 8x graphic cards ready to mine, hiveOS neutralise any error with driver on windows OS and also works more accurately than windows 10, download hiveOS

if you install Hive Os, then an 8GB flash drive will be enough for you

Just to update that the system has been stable for almost 80 hours, without any crashes or restarts whatsoever.

-200 MHz core/+1300 MHz memory @129watts for each of the cards. I can probably shave a few watts off however I’m happy with current profitability and stability of the 800 MH/s.

All my rigs run on only 4gb ram, including 12 units that are 13 gpus running on H110s.  I run linux mining OSes and they work fine on 4gb.  Windows probably requires more as everything windows requires more.

For fuck sake it wasn't intended to to solve that problem. are you being intentionally obtuse?

Anyway the problem was solved, it wasn't the SSD as I suspected. I'm out of here.

Yeah, I suspected it needed more Ram. 4gb isn't enough these days. In my Biostar TB360 2.0 I needed 16gb minimum to handle max gpus.Good job, let us know how things are going.

congrats.. good to know info

Guys, problem solved!

An extra stick of 8gb ram solved the issue.

Currently running stabily on 12gb ram + 180gb virtual.

Took off the extra SSD drive.

Will update in 48 hours.

Thanks for all the suggestions.

still doesn't solve his problem of getting 13 gpu running on a single system... unless he does pci-e x16 to a four pci-e x1 splitter then he'd be able to possibly run 13 gpu on current amd ryzen boards.  I don't know though, i only run max 6 gpu on my ryzen boards.. and i've only read of people doing splitters on intel boards and i've only seen intel boards with 7 or more pcie slots.

Wake up! It's an H110, you can't put a Ryzen on that regardless of the number of PCIe slots. Did you miss "future reference"?

umm.. that won't solve his problem cause they don't make a board that does more than 6 gpu for the ryzen.. unless i'm missing something here.

Add more Ram 16gb minimum. Not sure what the bios settings are for that mobo but make sure bios is UEFI mode. I agree too that you should upgrade your cpu. Also, did you try Gen 2 in the bios for all PCIE slots?

I don't hink it's the SSD unless you don't have enough for a pagefile the size of the total of all the VRAM on all your GPUs.
Just make sure your pagefile is big enough. It would also help to add more RAM so you don't have to rely on the pagefile
as much.

The recomendation for a bigger CPU is also a good one for future reference. Pentiums & Celerons have fewer PCIe
lanes than Intel Core. A small Ryzen is a good choice.

Maybe I didn’t explain it well but I actually see all 13 GPUs in device manager up and running. Doesn’t that eliminate the pci lanes issue? My previous system wouldn’t boot at all once I surpassed that limitation.

I’ll definitely try suggestion No 2 tomorrow because I’m secretly hoping that it has something to do with virtual memory and paging size.

If that doesn’t work however, I’d go for No 3.

Thanks

You can do that but the PCIe lanes that G3930 supports are less than the H110 PRO BTC+ 13 GPU-s needs.I had a Windows machine of a friend with this same celeron processor (I assume you have one since you mention virtual memory and MSI Afterburner) that could not do more than 10-11 cards in an ASUS B250 Mining Expert and I moved it to HiveOS where it recognized all 12 cards,cards were Rx 580 8GB each at that time a couple of years ago.

You can try a few things
1.Change to at least a core i3 processor,the cost is really cheap for that generation (doubt will solve anything but if you want to continue with windows)
2.Install the OS in the 480 GB SSD and see what happens.
3.Change to a Linux based OS like HiveOS. (should give you the fix)

Hello guys.

Relatively new miner here.

My setup consists of:

Asrock H110 Pro BTC+
G3930
4gb Ram
240GB SSD
1600W + 1000W EVGA T2 PSUs
12 RTX 3060Ti FE cards
1 RTX 3070 FE card
Windows 10

I moved the 12 cards from my previous x299 platform to the Asrock one just for the sake of the 13th card however I haven’t been able to mine with all 13 GPUs in the same rig. The systems boots and recognises all 13 cards, I can tune all of them using Afterburner, however after having installed few of the cards I start getting “cuda_error_out_of_memory” message both in Trex and Phoenix miner. Therefore I start gradually increasing my virtual memory to the point when I run out of SSD space but it’s just enough for both miners to start successfully and mine for 48hrs+ without any crashes.

However as mentioned earlier I run out of space. Which is way I installed a second 480gb SSD drive but after multiple attempts taking up to 100% of its capacity for virtual memory takes me to the same cuda error message.

Any ideas would be greatly appreciated.

Regards

PS the OS is in the primary SSD from which I’m taking up 200000mb and another 300000+ from the secondary empty one. Small increments in the primary one made huge difference however large ones from the new drive show no results. Should I install the OS in the 480GB drive?