GeForce RTX 3060 Ti - high load

[Gecon]

Hi,
Im getting really high CPU load with n3060ti LHR cards.

There is IRQ from Nvidia process that slowly poison entire rig.
Check this:

root@rig-stable-2:/# ps aux | grep nvidia
root      1102  0.0  0.0      0     0 ?        S    17:46   0:00 [nvidia-modeset/]
root      1103  0.0  0.0      0     0 ?        S    17:46   0:00 [nvidia-modeset/]
root      1105 94.4  0.0      0     0 ?        R    17:46   5:55 [irq/69-nvidia] <<===


LA 31.09 20.53 9.43 BOOTED 8 minutes ago     <<=== (L 31+)
0.6-209@210826  5.4.0-hiveos #140  N460.84 RTX 3060 Ti × 5

GPU 0    04:00.0    GeForce RTX 3060 Ti 7982 MB · NVIDIA
Hynix GDDR6 · 94.04.46.80.81 · PL 100 W, 200 W, 210 W


# top
op - 17:54:19 up 8 min,  0 users,  load average: 33.48, 23.28, 11.13
Tasks: 273 total,   7 running, 180 sleeping,   0 stopped,   1 zombie
%Cpu(s):  0.6 us,  8.6 sy,  0.0 ni, 90.6 id,  0.1 wa,  0.0 hi,  0.0 si,  0.0 st
KiB Mem :  8066256 total,  5851664 free,  1809916 used,   404676 buff/cache
KiB Swap:        0 total,        0 free,        0 used.  5890584 avail Mem

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
 1105 root     -51   0       0      0      0 R  93.8  0.0   7:51.58 irq/69-nvidia <<===

Im on N460.84 driver. Ive tried also next version. Same load.
Tried ETH and RVN coin mining.

But the load is not from miner program, it goes from cards. Any tips there?
Or having someone same performance issue.

(who doesnt know what load is … well, by the time, this rig get stucked due cpu overload and become unresponsible).

[Gecon]

This happens if you set MEM to cards.
If you set CLOCK, FAN, PL … all works fine.

But since you touch MEM, it goes to hell.
nvidia-info would NOT return anything. Any by quering frequently, load would go up.
No matter what you put to MEM (2000, 2400 …). Ive tried several AMD drivers, same result

So… any tips hot to set MEM and NOT to get high_load loops for 3060ti LHR v2?

[caramelzane]

I am facing this exact issue too! Gosh. My CPU LA goes up and the CPU temp also goes crazy high. Once I set Mem OC as 0, LA goes back to normal.

I am currently using G4560. I was told, may need to use i5/i7 to avoid this issue.

What CPU are you currently using?

[caramelzane]

I think this OC fixes it. Got this from a fellow miner in PoW@SG Telegram Group

[Lazym]

This is on ETH, seams stable:

[harman90]

I am facing same issue!
My CPU is G4400T

i am mining ergo!! anyone resolved the issue?

[Akky]

May I ask your card is LHR version?

[harman90]

@Akky

card is LHR version?

yes
i somehow found the stable setting from comments above …
1330 2100 130, I am getting around 133 mh/s with it

but I am not able to go beyond 26mh/s in raven… any OC setting ?

[allaboutpcs]

For me with 4 GPUs Hynix mem I have 1200 Core/1810 Mem/165 PL, works stable with NBMiner. When I added 2 more GPUs LA started shooting up. I will try to keep new cards OC low and increase gradually . lets see if that helps. I tried increasing LA for CPU in Monitoring to 2 but no luck.

[Lazym]

Hi,

Try this it’s been running for a month now.

[allaboutpcs]

Found this process is killing the system. This was happening with 1100 core/ 1770 mem / 165 pl. Let me try reducing mem and increasing core with lower pl

[detr0yt94]

Any updates on this 100% cpu usage? I feel like this may be happening to me as well…. I’m not sure…. I’m running 8 cards off a

asus h570,
16 gb of Corsair ram
Latest hive os
“8” Rtx 3060ti msi gaming x duo
“1” x1 to x4 pcie splitter

What program do you run to look at your system specs ?

[Lazym]

Whats your OC settings ?

[detr0yt94]

[Lazym]

Have you tried my OC?

[detr0yt94]

I did just now, nothing changed except more wattage being used…… any recommendations on driver I should be using ?

[Lazym]

I’m currently running them on N 465.31