19 minutes ago, Nyxx said:

If thats your hot spot I fear your mem temp could be 105-110.

This happened on my first 3080, it went down hill from there. I got it replaced and the new 3080 same make never goes over 82 and thats the mem temp and that is always hotter than the hot spot.

The mem temp is 5 move than the hot spot even at rest.

I fear you will need to replace it like I had to do. The first 3080-3090 seem to have a lot of probelms with heat. 

I made my "new" 3080 run as hard as it could on a very hot day and it hit mem temp of 86 thats the highest its ever gone. It seems they have changed somes things! 82 is the max 99% of the time.

My memory junction temps never exceed around 85°C.

54 minutes ago, Nyxx said:

Like I just said, glad your sorted it out before melt down starts. I think the first 3080-3090 thermal pads were just bad...and fan curves.

Not exactly meltdown, but thottling and performance lost. But yeah, default ones were total rubbish.

25 minutes ago, xenonsk said:

Not exactly meltdown, but thottling and performance lost. But yeah, default ones were total rubbish.

Yes i was getting thottling and then CTD started....

I use GPU-Z to watch everything.

Edited September 12, 20214 yr by Nyxx

1 hour ago, ThomseN_inc said:

My memory junction temps never exceed around 85°C.

On two 3080 and my son's 3070 always mem temp is higher than Hot spots etc.

GPU-Z

 

2 minutes ago, Nyxx said:

On two 3080 and my son's 3070 always mem temp is higher than Hot spots etc.

GPU-Z

 

Might be, but not with mine. OCCT.

In amongst all this talk of gpu temps (which is important) are we missing the main point that Thomsen thinks that flying through clouds at high level is more demanding than flying through the same clouds at a lower level?  I.e. a bug? Or some other effect?

9 minutes ago, bobcat999 said:

In amongst all this talk of gpu temps (which is important) are we missing the main point that Thomsen thinks that flying through clouds at high level is more demanding than flying through the same clouds at a lower level?  I.e. a bug? Or some other effect?

Yeah that was my actual interest. To see if someone has an explanation for what might be causing this. Thanks for bringing it up again. 🙂

40 minutes ago, ThomseN_inc said:

Yeah that was my actual interest. To see if someone has an explanation for what might be causing this. Thanks for bringing it up again. 🙂

No problem! 

I am sure I have seen this mentioned before, and maybe even experienced it myself with my old graphics card, but I cannot for the life of me think of a logical reason for it, so sorry I cannot help you on this.

I did face a massive issue with frame drop on the ground a few months ago, and that seems to have stopped now.  Maybe they fixed it or the recent optimisations solved it, but it just shows that you can get these random issues like your clouds issue. 

I spent many nights trying to solve mine, but had no real success other than it was maybe down to shadowing of the aircraft on the ground.  That would not be related to your issue of course.

Edited September 12, 20214 yr by bobcat999
Swapped 'would not' for 'would' in last sentence.

Just now, bobcat999 said:

No problem! 

I am sure I have seen this mentioned before, and maybe even experienced it myself with my old graphics card, but I cannot for the life of me think of a logical reason for it, so sorry I cannot help you on this.

I did face a massive issue with frame drop on the ground a few months ago, and that seems to have stopped now.  Maybe they fixed it or the recent optimisations solved it, but it just shows that you can get these random issues like your clouds issue. 

I spent many nights trying to solve mine, but had no real success other than it was maybe down to shadowing of the aircraft on the ground.  That would be related to your issue of course.

No worries! Its not a big deal for me since i have plenty of headroom performance wise. I am just being curious.

 

I was just thinking of airspeed. Since we (usually) fly faster through those higher level clouds than we do at lets say 3000ft. Maybe it has to do with how fast the clouds move along the plane and the lighting situation (brighter, darker, shadows, reflections etc) that has to be computed in a faster manner at 300kts....just wondering. i really have no clue. 🙂

31 minutes ago, ThomseN_inc said:

No worries! Its not a big deal for me since i have plenty of headroom performance wise. I am just being curious.

I was just thinking of airspeed. Since we (usually) fly faster through those higher level clouds than we do at lets say 3000ft. Maybe it has to do with how fast the clouds move along the plane and the lighting situation (brighter, darker, shadows, reflections etc) that has to be computed in a faster manner at 300kts....just wondering. i really have no clue. 🙂

Well that is not a bad guess!  I have heard volumetic clouds are quite a performance hog.  If they are coming at you faster, the redraw would be more taxing on the GPU, so maybe you are on to something.

Do you fancy 120 knots on the Airbus at 30,000 feet to try it?  

1 hour ago, bobcat999 said:

Well that is not a bad guess!  I have heard volumetic clouds are quite a performance hog.  If they are coming at you faster, the redraw would be more taxing on the GPU, so maybe you are on to something.

Do you fancy 120 knots on the Airbus at 30,000 feet to try it?  

I'd rather take the XCub to see how it goes 300 kts at 2000 ft 😄

I loose 20-30 FPS when going through those high level clouds using a 3090. Luckily I am pulling mostly 60fps, so it’s not as noticeable but still annoying. 

I know the temps look scary on the 3000 series but I think it’s within their specs. I often run HW Monitor which shows numbers in red if they get within range of throttling. So far that hasn’t happened for my 3090 (despite some memory junction temps of 98) but it did happen on one core of my 10900 when I had hyper threading enabled.