RTX 2070 stuck @ 97W with PL set to 125W

[Mozgala]

HI,

I have received RTX 2070 today it is ASUS with Blower style cooler, I instaled it on Test Rig and under windows run for few hours under full load to check for errors and cooling capabilites and then some hours on ETH mining around 40 MHS with core locked at 1000 MHz and Memory at 7500 MHz. Then i switched to Hive OS and hit some strange PL block @ 97W. I tired diferent miners buth all with similar result, card ois stuck at 97W.

Any ideas what’s the problem.

Update1:
I have downgraded Hive few versions back - no fix
I Have tried few older drivers al the way back to 460 - no fix

If i increase Fan speed or try to overlock memory Hash rate decreases as the core clock is getting lower and lower. at power is stuck at 96-97W range

Update 2:
I have switched back to windows and it runs at 170W no problem, run some more test under windows and it hash normaly, then i moved it to diferent hiveOS rig and now it is again stuck at 96 W

[keaton_hiveon]

increase the core clock until you get full hashrate, probably another 100mhz or so. you also have no memory clocks set(leaving hash on the table) power limit only limits the power draw from going above the set amount. doesn’t make it use that amount.

[Mozgala]

It is not case of OC settings, i cen stet core to 1000, 1200, 1500 at card still stays at 96W,

When for example i set core to 1100, leave memory at 7000 MHz and set fan to 70% power stays at 96W byt has rate goes down to like 31 MH/s and core in nvidia info is at 600 MHz

When i set core to 0 , and set memory to 7500 MHz fan to auto, MH/s drops .

If i dont touch OC at all and leave FAN at auto it is making 35,5 - 36 MH/s @ 96W
I have tried card on two rigs under HiveOS and it acts the same

And if i put it into windows system it works perfectly fine.

I have few other RTX’s 2070 at they all run at 41-42 MH/s at 1000 core and +1000 mem in 100-110W range

[keaton_hiveon]

change fan to 100% and see if it changes anything.

[Mozgala]

@keaton_hiveon

power limit only limits the power draw from going above the set amount. doesn’t make it use that amount.

BUt from my experience if you leave card Core, and memory OC untouched and set PL to some value lower then maxiumu card TDP, card will take that amount, it will start decrease only when you start to manualy decrese core clock.

In my case card is taking 96W even if I dont touch core and memory and set PL to 125 in this case, and in nvidia tool it says it is Throthling with PowerLimit as reason even as it set at higher walue then 96W.

Blockquote

[Mozgala]

@keaton_hiveon

change fan to 100% and see if it changes anything.

Here you go Fan at 99%, (forced in autofan to 100%)

> Blockquote

Card stays at 89W with Pl set at 175, and temps around 33 on core (ambient temp in warehouse it is installed is around 15 Celsius atm) Trottle : PowerLimit

And if i go and put in in windows rig it works at 175W no problem

[keaton_hiveon]

set core clock, disable autofan and just use 100% fan in oc config, does the clock stick?

[Mozgala]

As fan go higher up to 100% , core clocks go down, and card power throttle at 89-96W

If i turn fan to auto, it slowly goes down and core clock climbs up but stops at 700-800 MHz and power drawn 96W

[Mozgala]

I found the fix:

This ASUSRTX 2070 TURBO EVO with 90.06.18.40.81 bios need to be set to PL 195 W, anytihing lower and card will power throttle. It is first card im my mining history in wich i must increase PL instead of decreasing it, Anyway real power taken is keept in check witch locked core clock so card takes 110W but PL must be set to 195W

[krispawelek]

well ive been dealing with this card for a long.
Looks like adding some PL to 195 do the job.
thx