2.8 seems to get the GPU hot with RTX 4090 at 4K at least 20 degrees more than 2.7!

[itye1970]

I tried 2.7 and was getting 75 to 85 degrees on 4K mode. Once it hit 90 degrees. I went back to 2.7 and cpu temperarure at 4K high only 62 degrees with hotspot about 68. All settings were in 4k on High or Ultra

Switched back to 2.7 for now, are the devs going to optimise this?

Edited November 6, 2022 by itye1970
spelling

[Vakarian]

Your GPU temps going higher means your GPU is getting used more (which is good), nothing wrong with that. Play around with your GPU settings to get better cooling under high load conditions

[TZeer]

It's not good if it's hotter and more used without getting more performance.

 

[Vakarian]

4 hours ago, TZeer said:

It's not good if it's hotter and more used without getting more performance.

 

It's not that simple. If the GPU is getting told to work harder, it's going to work harder. That doesn't directly translate to how many FPS you get. If there are more stuff to render than before, that's additional tasking on GPU it must complete in same time it used to perform less tasks (that's just to keep your FPS the same as before) so it's going to work harder and as result get a bit hotter.

Regardless of that, if your GPU is getting too hot while gaming, then you need to check your cooling. If it's simple air cooler, bump those fan speeds higher

[TZeer]

I know. But when you have one patch 2.7 --> 2.8 creating this kinda behaviour, something is off.

I would put it in context of the issues many are seeing with loss of FPS and higher loading times with 2.8 vs 2.7

I have not yet updated to 2.8 due to all issues with performance, so I don't have first hand experience with it.

But if he is going from 60-70 degrees up to 90 degrees on the GPU.. there is indiaction that the bottleneck was previously on his CPU, and going to 2.8 has moved it over to his GPU, or at the very least taxing his GPU significantly higher than before.

People loosing 10-15% performance, going from 120 FPS to roughly 105-110 FPS, wont notice. But VR and 4K users running 50-60 FPS will notice theese drops very easily.

Then again, "smooth frames" is a personal taste. There are people running 30-50 frames claiming it to be smooth

[itye1970]

went back to 2.7 and avg temps are now 52.5 and GPU hotspot is 59.9 average. If I switch on Vysnc its even lower!

Edited November 7, 2022 by itye1970

[Vakarian]

What a surprise. You tell GPU to do less, and it gets colder... This is completely normal. 4090 is no slouch, that beast is going to get hot when run at it's max so get ready to cool it down.

[Dogmanbird]

I'd imagine the new lighting effects in 2.8 alone, will hit the GPU more than 2.7 and now making use of some of the surplus gpu performance previously sitting idle due to an already cpu bound system. Therefore the game might now look prettier but fps will remain about the same as before

 

Edited November 8, 2022 by Dogmanbird

[Hiob]

Guy, honestly, without some real numbers (afterburner stats? benchmarks?) this is nothing but hearsay (and pretty unlikely).

I don't have a 4090, only a 3080 and there is no differences in temps whatsoever.

DCS doesn't use any fancy GPU properties, just plain simple DX11 rasterization. I strongly doubt that there is a 4090 specific issue with 2.8. That is not how interfaces like DX11 work.

Most likely the 4090 was bottlenecked by some kind of V-Sync or capped fps and the settings changed with the update.

Without proper numbers and screens of the exact settings, this whole topic is pointless! (and misleading!)

In other words: This topic cries "individual/ system specific problem", even more so because it is the only single report on such an issue afaik.

@itye1970 Don't get me wrong. I'm absolutely willing to help you. I'm just not a fan of making every personal system hickup a general DCS-issue right upfront! So please provide the data, and we will work it out!

Edited November 8, 2022 by Hiob

[Mars Exulte]

#1 As long as your numbers are within spec, it's irrelevant. Your GPU is working more, it will get hotter. That simple. This does not equate MOAR EF PEE ES

#2 If it's getting too hot, your cooling solution is subpar. Period.

#3 There is no such thing as a game melting/damaging a GPU. There are only poorly designed cases and ''diyers'' that don't know what they're doing.

Edited November 9, 2022 by Mars Exulte

[Hiob]

24 minutes ago, Mars Exulte said:

 

#3 There is no such thing as a game melting/damaging a GPU. 

 

Well…. 

certainly not DCS, but there has been cases where poor GPU design met poor game design and poof.

[SkateZilla]

Normal Software does not Kill Hardware, Users Kill Hardware, or Malicious Software Kill Hardware, DCS is Neither.

Software is processed by the Hardware, GPUs melt because of Abuse, Misuse, or Disabling of Fail safes

User Error and User not Understanding what Power / Fan Curves are Leads to Melting GPUs, not GPUs Processing the Commands it was given.

DCS 2.7 Graphically did less than DCS 2.8.

DCS2.8 now renders more environmental items (Lights on PG Map, as well as weather Elements.)

You getting the same performance (Fr/PS) while doing more work, so yes, the GPU / CPU will be hotter and draw more power..

However, The System itself that governs temps is the Driver and Fan Control, not DCS.

DCS2.7 is doing less render work, and less environmental work, so it will use less GPU to the same Fr/PS.

If you don't want the GPU to Get hot, change the fan curve and power limit.

Edited November 9, 2022 by SkateZilla

[Mars Exulte]

5 hours ago, Hiob said:

Well…. 

certainly not DCS, but there has been cases where poor GPU design met poor game design and poof.

Yeah, I'm still gonna say that's probably from the user fiddling with overclocking/fan curves on top of having insufficient airflow.

Short of having improperly installed cooler/thermal paste from the factory, I seriously doubt you can have a temperature related failure unless the end user did a lot to help it happen. Even in the above case, I'm inclined to think the end user was probably having issues and simply chose to ignore them until failure. They are literally designed to save themselves in the event of routine problems like that, so barring a malfunction in THAT process... yeah, the user ignored the warning signs until disaster struck.

[rayrayblues]

On 11/8/2022 at 7:38 AM, Hiob said:

Guy, honestly, without some real numbers (afterburner stats? benchmarks?) this is nothing but hearsay (and pretty unlikely).

Yeah Dude, I have to concur. Let's start with the obvious s**t first.

COOLING!!. The 4090 runs up to 450 watts then it throttles down.

Are you air cooled or water cooled? how much case ventilation do you have? Is it positive or negative pressure?

The 4090 is the biggest and baddest card out there. I would expect it to run hotter.

Check your power settings and fan curves and don't worry, it can handle temps in the low to mid 90's.

I wouldn't be concerned until it reaches 100^o, at which point the card will throttle back to protect itself.

1 hour ago, Mars Exulte said:

I'm inclined to think the end user was probably having issues and simply chose to ignore them until failure. They are literally designed to save themselves in the event of routine problems like that, so barring a malfunction in THAT process... yeah, the user ignored the warning signs until disaster struck.

There are various ways that you can adjust power settings and fan curves. MoBo BIOS, MSI Afterburner, etc.

How about the card's own BIOS? Even NVCP settings can make a difference.

Start with the obvious. 

Bigger card = more power = higher temps = more cooling/ventilation.

Make sure your case is not in a place that will block any airflow. Blow out the dust bunnies.

Do whatever you can to reduce temps, but don't be surprised if you can't get things to where they used to be. As long as you are not exceeding

100^o constantly, don't worry. It will shut itself down before it burns up. 

You might have to live with a bit more fan noise than you're used to. It comes with the territory.

 

While you're doing all of that, don't forget to help us help you. 

On 11/8/2022 at 7:38 AM, Hiob said:

Guy, honestly, without some real numbers (afterburner stats? benchmarks?) this is nothing but hearsay (and pretty unlikely).

Yeah, exactly. 

Give us as much info as you can. We're here to help, but you need to help us too.

 

There are known issues with the PCI e power adaptors melting as well. Make sure that this is not your problem.

 

Edited November 10, 2022 by rayrayblues

[Hiob]

7 hours ago, Mars Exulte said:

Yeah, I'm still gonna say that's probably from the user fiddling with overclocking/fan curves on top of having insufficient airflow.

Short of having improperly installed cooler/thermal paste from the factory, I seriously doubt you can have a temperature related failure unless the end user did a lot to help it happen. Even in the above case, I'm inclined to think the end user was probably having issues and simply chose to ignore them until failure. They are literally designed to save themselves in the event of routine problems like that, so barring a malfunction in THAT process... yeah, the user ignored the warning signs until disaster struck.

I was referring to this incident. I thought everybody knew about it by now:

https://www.techspot.com/news/90503-amazon-new-world-reportedly-destroying-rtx-3090-cards.html

[Mars Exulte]

Yeah, I heard about it. And in the very article it describes baaically 100% usage and some guys having abnormally high temps in the 80s but not cooking, implying

A. They're monitoring such things and actually noted the change (which you should notice the fans spooling up if nothing else)

B. Their system controlled the cooling situation well enough to prevent runaway overheating. A loooot of people run thermally borderline rigs, which you can get away with during day to day usage 99% of the time. Then something like that happens and they have no headroom to absorb it.

Barring a glitch interfering with fan operation or thermal throttling (possible, but not what that article implies happened) you're not going to fry your system JUST with 100% usage. If it does, you've done something to help it along substantially.

 

''I clicked apply and bricked it'' is bs.

Edited November 10, 2022 by Mars Exulte

[5ephir0th]

I dont see where is the problem as long as temps are inside the normal working temp of your GPU on similar workload, they added new weather effects so its absolutely normal the temps go up as the GPU are working harder, i havent see any anormal on my 4090 with 2.8

[Hiob]

1 minute ago, 5ephir0th said:

I dont see where is the problem as long as temps are inside the normal working temp of your GPU on similar workload, they added new weather effects so its absolutely normal the temps go up as the GPU are working harder, i havent see any anormal on my 4090 with 2.8

Of course not. I think it's pretty clear by now, that there are other things at play here. Most likely settings like v-sync that changed between the versions.

[draconus]

Mind that nvidia 4090 reference card is specd at max 90°C. I'd take a moment to look for better cooling by either better fan profile or hardware changes if needed.

[skywalker22]

On 11/6/2022 at 12:28 PM, itye1970 said:

I tried 2.7 and was getting 75 to 85 degrees on 4K mode. Once it hit 90 degrees. I went back to 2.7 and cpu temperarure at 4K high only 62 degrees with hotspot about 68. All settings were in 4k on High or Ultra

Switched back to 2.7 for now, are the devs going to optimise this?

 

You PAID for it, make it pay back

[5ephir0th]

21 hours ago, draconus said:

Mind that nvidia 4090 reference card is specd at max 90°C. I'd take a moment to look for better cooling by either better fan profile or hardware changes if needed.

This.

OP has a problem with his case config, my 4090 FE doesnt go above 65º running Port Royale test that is WAY more complex that anything DCS can show, on my (not finished) undervolt curve doesnt gets above 60º (again, on Port Royale), if you are getting 85º running DCS on a 4090 definitely you have a problem and it´s not a DCS one

[petsild]

On 11/6/2022 at 12:28 PM, itye1970 said:

I tried 2.7 and was getting 75 to 85 degrees on 4K mode. Once it hit 90 degrees. I went back to 2.7 and cpu temperarure at 4K high only 62 degrees with hotspot about 68. All settings were in 4k on High or Ultra

Switched back to 2.7 for now, are the devs going to optimise this?

 

Which RTX4090 model do you have?