Author

Topic: Adding gpus slows my rigs, cant run more than 5 (Read 211 times)

hero member
Activity: 1274
Merit: 556
September 14, 2018, 10:04:20 AM
#13
It's mostly that it takes a very long time to actually install 1803 (or any of those creator updates for that matter). And once you've done that you'll have to (re-)install the video drivers and any soft/reg-mods you had applied prior. The process of Windows going through all devices after a big update like this can take several hours.

My Onda D1800 based rig (6x Vega) took ages to become normally responsive again.

So after any update like this I've basically resorted to leave the rigs alone for several hours before attempting to make them ready to mine again. The more GPUs, the longer it'll take.
newbie
Activity: 1
Merit: 0
Hi
Pls share the bios setup for > 4 GPU. Thank you for help.
newbie
Activity: 6
Merit: 0
IT WORKED.  I felt it had to be a Windows problem but it never occurred to me (obviously) that the newest version of windows won't run multiple GPUs.  I think there must be a setting in v1803 that enables it to work but I am running 9 GPU on B250 Mining expert and 7 on Asus Prime 370 P.3

Seems weird to me that its not a well known thing NOT to run version 1803.

Thank you 2 guys for taking the time to help me out!!!

newbie
Activity: 6
Merit: 0
i have 1803, I'll try 1709, although i'm not sure how to revert
jr. member
Activity: 269
Merit: 4
I just spent the last 8 hrs or so trying to get 6 various RX's back up to speed on a B250 ME.

Started with a driver upgrade (had still been using the blockchain drivers) and went down hill from there.

Anyway the fix for me ended up being to go back to the Win10 fall creators version. Just couldn't get 1803 to work...reinstalled 1709 and everything fell right into place. Using 18.3.4 drivers with Claymore 11.8. Back up to full speed on all 6 cards.



Exactly!  I was going to ask which build of Windows.  I went through the same thinglast week over a 3 day period.  Finally went back to 1709 build and everything else fell into place.
newbie
Activity: 49
Merit: 0
I just spent the last 8 hrs or so trying to get 6 various RX's back up to speed on a B250 ME.

Started with a driver upgrade (had still been using the blockchain drivers) and went down hill from there.

Anyway the fix for me ended up being to go back to the Win10 fall creators version. Just couldn't get 1803 to work...reinstalled 1709 and everything fell right into place. Using 18.3.4 drivers with Claymore 11.8. Back up to full speed on all 6 cards.

newbie
Activity: 6
Merit: 0
It wasn't enabled, tried with it enabled. No difference.  I think the iGPU is for the onboard graphics card which I'm not using (of coarse)
jr. member
Activity: 269
Merit: 4
Are you using intergrated video?
newbie
Activity: 6
Merit: 0
That's not helpful.  Not looking for lectures on the evils of GPU mining!
member
Activity: 924
Merit: 15
considering the situation, why would you be adding more GPU's¿¿¿ havent you seen the actual profit you are getting? or are you betting that it will go up, with no actual reason for it??
newbie
Activity: 6
Merit: 0
Thanks
HCP
legendary
Activity: 2086
Merit: 4363
Just FYI, you might have more luck if you ask this over in the "Mining (Altcoin)" section.

I suggest you use the "move topic" option in the lower left corner and select this board: https://bitcointalk.org/index.php?board=160.0

newbie
Activity: 6
Merit: 0
I am building my first 2 mining rigs and I am having the same problem with both rigs: Hook up one gpu, fine.  Hook up 2, fine but I can notice slightly slower load time. Hook up 3, slightly laggy. Hook up 4, laggy but functional, 5, takes forever to load windows and everthing is slow to react. hook up six, barely functions. hook 7, 8 or 9 and doesnt function at all. (before running miner)

Rig 1: Asus B250 Mining Expert mobo
16 g ram 2666mh (had 8, but now have 16 because i had i so i tried it
windows 10 pro
I3 7100
9 rx580: 2 Asus, 6 MSI, 1 Gigabyte.   All bios modded
2 EVGA 850w titanium psu, evenly balance hook up, no more than 2 risers per sata, mobo and 4 gpu on one, 5 gpu on the other
Virtual memory set to 32000
Kingston 120 ssd
amd driver; Adrenalin 18.5.2 ( tried Crimson 17.10.4, works the same way)

Rig 2: Asus prime Z370 P
8 g ram
windows 10 pro
I3-8100
7 rx570 all bios modded
EVGA 1200w titanium psu, only 2 hooked up per wire
Virtual memory set to 32000
Kingston 120 ssd
amd driver; Adrenalin 18.5.2 ( tried Crimson 17.10.4, works the same way)

Bios 4 gpus enable on both, tried all combos of gen 1,2 or 3 (although i think gen 3 is correct one since gpus are gen 3 and mobo are gen 3 and gen 3 is backward compatible)

High speed internet, wired

So here is what i Do to get both rigs running:

hook up 5 gpu, start up.  In Device Manager disable all gpu (in rig 2 since there are only 7 gpu disable 3 gpu)  shut down and connect remaining gpu.  Restart

Now there are 4 gpu enabled.  Start miner (Claymore 11.6).  once miner is initiated and mining begin enabling 2 or 3 more gpu, slowly so not more than 1 is enabling at a time, ie enable and wait for 30 seconds until the icon turns green in device manager.  once I have 2 or 3 (not 4, 4 over whelms them) enabled  WHILE STILL MINING i start Claymore (a second instance).  When the second claymore shows the gpus (green lines) and then the algorithm lines quickly shut down the first instance of claymore.  The second instance will be running 7 gpu.  If the miner shuts down or needs to restart it won't because there are 7 gpu enable and Windows comes to a virtual standstill. I need to hard shut down, disconnect 3 or 4 gpu and start the whole process again.


If while the second instance of claymore (now the only one running) shows a gpu failed to connect (a red line) and I catch it I can open the 3rd instance, when the algorithm lines show up on it shut down the 2nd instance and I can usually save it from a crash

Now there r seven running, (with the miner using the gpu Windows is now not tied up by them and runs normal) so i enable the last 2 gpu and open another instance of Claymore.  Again when it shows the Algorithm lines shut down the previous instance.  If a gpu failure line shows up during this process another instance of claymore needs to be opened quickly before the miner shuts itself down or tries to restart.  If it tried to restart the computer takes over the gpus and with 9 gpu it can,t run, comes to s virtual standstill.  Disabling a gpu at this point completely freezes the computer.

To get the 9 gpu rig going takes about 20 minutes, 7 rig about 10-15.

The computer runs slower with more gpu enabled, as soon as the miner takes over the gpus the computer gets its normal function (speed) back

To put it a different way: the more gpu hooked up the slower windows starts after 6 or 7 its virtually not functioning.  With 5 the miner may start, open claymore and after 30-40 minutes it ,catches and may open up.  with six the miner will not get to the point of recognizing the gpu, with seven i cant even disable a gpu the computer is so overwhelmed, the miner will open the command line will scroll through but the computer is so overwhelmed it doesn't appear to communicate with the pool.

I have Googled and I have Binged there seems to be no one in this world having this problem, TWICE.

I feel its a Windows setting, I have gone through every bios option and Google what each means and none seem to me to be the solution,

please help
Jump to: