When is too much wattage too much ? (RTX 3070)

[Mark Kaine]

Honestly i wasn't expecting OW of all things to use that much power... but what do you guys think, is this still okay for a supposed 220w card? (although my Vbios says 270w, tbf)

 

 

 

also those temps... time to clean my PC? (i would say yes)

 

82C for CPU (5800x3D) seems actually high also, though ?

 

But also note, no throttling of any kind, perf limit is still "idle".

 

Should i be concerned? I guess I'm just not used to these kind of temps outside of benchmarking,  maybe its time to update chipset drivers too? 

 

 

ps: nothing OC except the RAM either,  usually i UV the GPU and CPU, but Overwatch really seems to hate that (MSI Afterburner specifically...) maybe that's why all those numbers seem so high to me.

[OhYou_]

lgtm, gpus these days have turboboost that is usually on by default and pushes power limits higher than normal if it can keep cool.
the temps are going to be normal because the gpu is targeting that temp and keeping the fans quiet to do it

[Mark Kaine]

8 minutes ago, OhYou_ said:

lgtm, gpus these days have turboboost that is usually on by default and pushes power limits higher than normal if it can keep cool.
the temps are going to be normal because the gpu is targeting that temp and keeping the fans quiet to do it

oh right... haven't considered the terrible default fancurve either... but since OW hates Afterburner im SOL on that front lol...  

 

its just crazy to me a "220w" card pushes 275w at "stock" ...

 

im really curious if cleaning the pc and the fans especially does anything  - its long overdue tbh (and yes there will be a lot of dust, sadly...) 

[OhYou_]

it might run a bit cooler

[TrigrH]

It's all normal. TDP is a bullshit lie anyway.

 

My 5800x3d has similar temps.

[Zando_]

Nvidia won't let you pull unsafe wattage/voltage on modern cards. So unless you've done some shenanigans to get around that, yes it's safe.

[Godlygamer23]

What specific RTX 3070 are you running? I don't pay much attention to those figures, but perhaps I could go through a session of OW and see what figures I have(I also have an RTX 3070).

[SkilledRebuilds]

1.08v on GPU is at the top end of the whole curve, you got plenty of under-volting headroom available... can still increased clocks if you want with reduced voltage as well, just not as much. Or leave clocks alone and reduce voltage or power targets.

Voltage down to say 1.0v and watch the power drop.... then add clockspeed (if you want) and test stability.

 

My 2080Ti will default around 1.043-1.05v 1980Mhz-2025Mhz, depending on temps summer or winter.

MSI Afterburner's Voltage Slider all maxed out, it'll see 1.08v as well.

But I set it for 2100Mhz 1.018v instead.

I targeted 1.018v with 2100-2115Mhz as my Undervolt/Overclock (+135)

Could do better still leaving it at 1980Mhz and setting between 0.875-0.950v (give or take some voltage to ensure stability)

 

[Mark Kaine]

2 hours ago, Godlygamer23 said:

What specific RTX 3070 are you running? I don't pay much attention to those figures, but perhaps I could go through a session of OW and see what figures I have(I also have an RTX 3070).

its a Gigabyte vision OC. As i said the card is rated at 270w, which is so far ok, i was just honestly a bit confused,  since i updated windows the whole pc seems to run quite a bit hotter, and outside of benchmarking i haven't seen these temps, and neither these power figures,  certainly not in OW, that typically uses more like 210w on the gpu for example (even without undervolt) 

 

And the thing is, as said i cant run afterburner at all in OW, i keep getting random crashes, even if its just running with the overlay no overclocks or undervolts... 

 

Whats crazy i played a bit again (like 30 minutes)  and it used even more power, 276w..  and ran even hotter,  but apparently no throttling whatsoever (not noticeable at least) 

 

but its a bit scary tbh as im really not used to such temps... and power usage.

 

yeah, if you can check for comparison that would be appreciated,  kinda a small sanity check.

 

im running 1440p max settings (except volumetric fog, i think, on low) FSR1. 165fps (it never drops) no Vsync,  or Gsync.

 

edit: my Vbios, stock:

 

 

 

[Mark Kaine]

48 minutes ago, SkilledRebuilds said:

1.08v on GPU is at the top end of the whole curve, you got plenty of under-volting headroom available... can still increased clocks if you want with reduced voltage as well, just not as much. Or leave clocks alone and reduce voltage or power targets.

Voltage down to say 1.0v and watch the power drop.... then add clockspeed (if you want) and test stability.

 

My 2080Ti will default around 1.043-1.05v 1980Mhz-2025Mhz, depending on temps summer or winter.

MSI Afterburner's Voltage Slider all maxed out, it'll see 1.08v as well.

But I set it for 2100Mhz 1.018v instead.

I targeted 1.018v with 2100-2115Mhz as my Undervolt/Overclock (+135)

Could do better still leaving it at 1980Mhz and setting 0.950v (give or take some voltage)

 

I usually run a very different setting to stock (0.9v @ 2010mhz) but as said i cant run afterburner at all while playing OW, so i gotta run it at stock sadly (the stock fancurve sucks too, tbh)

 

but THE THING IS ive been playing OW like this for a year now and haven't seen such temps at all, 60-70c, 200w usually... so something is up, could just be my recent windows update (speculation) 

[porina]

You don't have a 220W card. You have a 270W one because Gigabyte made it that way. And numbers peaking slightly above power limit are not significant.

 

If you really wanted a 220W GPU you could have got another one, or you can turn yours into one by setting a lower power limit.

 

My MSI Gaming Trio X 3070 

[Mark Kaine]

27 minutes ago, porina said:

You don't have a 220W card. You have a 270W one because Gigabyte made it that way. And numbers peaking slightly above power limit are not significant.

 

If you really wanted a 220W GPU you could have got another one, or you can turn yours into one by setting a lower power limit.

 

My MSI Gaming Trio X 3070 

that's true as said im more concerned about the temps... as ive said since that last windows update my whole pc seems to run hotter 10-20c, which is pretty significant  - especially for no apparent reason, i also checked my bios settings and they seemed unchanged from before the update. and now OW suddenly uses over the TDP, which it never did, not even close... so the more i think about it this is really weird, and while the numbers may be not concerning (yet?) this sudden change certainly is in my eyes...

 

maybe something with the power plans is wrong, even though i didn't change them? 

[porina]

Think about what could cause the GPU to run hotter:

1, the GPU is doing more work. Any patches to the game recently? Did you update drivers? Any other graphic/game setting changes?

2, the GPU cooler is blocked

3, the case airflow is blocked

4, something else in the case is producing more heat now

 

If you're not even reaching the (soft) thermal limit I wouldn't be overly concerned at this point. Even if you did hit the soft thermal limit, it is more a performance drop than risk to hardware.

[Mark Kaine]

14 hours ago, porina said:

Think about what could cause the GPU to run hotter:

1, the GPU is doing more work. Any patches to the game recently? Did you update drivers? Any other graphic/game setting changes?

2, the GPU cooler is blocked

3, the case airflow is blocked

4, something else in the case is producing more heat now

 

If you're not even reaching the (soft) thermal limit I wouldn't be overly concerned at this point. Even if you did hit the soft thermal limit, it is more a performance drop than risk to hardware.

only thing that chaned is windows update. 

 

going from 1809 to 22H2... im honestly surprise how well most things work,  but those temps, nah man, prob going back into safety (of my mind lol) theres also some strange sleep mode problems (the screen "turns off" at the specified time,  but it doesn't really turn off, it just gets no signal,  super strange that hasn't ever happened on 1809) HDR is somehow even more buggy now, thumbnail creation takes forever* (seems to be smaller cache too, super annoying) etc.

 

but yeah, i figured out if i run afterburner just for the fancurves and turn the overlay off, OW might not crash, and so far it didn't, not a real improvement though,  still ca. 20C higher,  and far more power usage than on 1809.

 

 

 

 

 

*just let me choose my thumbnail cache size,  ms, pretty please? about 30gigs would probably do the trick instead of the meager 1gig it seems to be now...!

[Godlygamer23]

As promised, I did do some runs with my RTX 3070, and have some interesting observations(IMO). Overwatch for me is not a good test, at least based on the current configuration I'm using - I do play at 240Hz, but keep my settings down to maintain as high of a frame rate as possible.

 

For reference, I use an EVGA RTX 3070 FTW, and have the core voltage, power limit, and temperature limit all applied to the max allowed(+100, 111, and 90, respectively). Core clock offset is +125MHz, and my fans run at a set percentage of 65%. 

 

During gameplay of Overwatch, which lasted a full match, I observed a maximum board power value of 186.7W, and a max GPU power value of 151.3W. I switched off to Superposition, and ran through the benchmark at 1080p Extreme, and observed a maximum board power value of 300.4W, and a maximum GPU power value of 256.8W.

 

I logged everything, and I have them attached to this post. The one that just says "GPU-Z Sensor Log" was the one used for Overwatch.

GPU-Z Sensor Log(Superposition).csv GPU-Z Sensor Log.csv