1.9.5.0 is cooking my GPU again

[LRBS 948]

@Drumcode is not the only person complaining about this issue.

I noticed good advice that might help regarding a "overheat" situation and some condescending remarks.

Some people here cannot accept that it might be something wrong with the software as to not being properly optimized.

I would like to remind to the "nonbelievers" that P3D was doing the same and XP11 for a long time suffered from the same issue, running video cards hot and fans continuously.

This MSFS2020 is not properly optimized (CPU and GPU) and will raise temperatures, you can run P3D for hours and it will not have any issues with the temps, as he mentioned.

Let's concentrate  and ask ASOBO to optimize the software properly and stop being condescending to each other, it doesn't fix this broken simulator fighting between ourselves.

[Glenn Fitzpatrick 3,798]

4 minutes ago, z06z33 said:

In 20+ years of gaming I've never had a game over heat my PC. Either somthing is messed up with your GPU cooling, dead fan,bubble in thermal paste,or just dusty...  If you've OC'd your card that may be pushing it past thermal limit,or your card is dying. Its not the game its either cooling or a fault in the GPU the 3080 have a capacitor issue causing black screens currently.

Be aware that badly written game code CAN cause hardware issues. EVE online was infamous for it about 10 years back, here is just one thread about EVE destroying graphics ram this issue was real and acknowledged by CCP games:

https://oldforums.eveonline.com/?a=topic&threadID=1029933

I do DOUBT there are any issue like that here but it is worth bearing in mind it can happen.

Aside from that, if someone is running a stock PC with a gaming card rather than an custom built gaming rig (never lone an otherwise off the shelf laptop with an nvidia card in it)  you can often get overheating. Even in a custom built gaming rig you can still get issues with the more budget ranged GPUs that tend to have small heatsinks and very few fans.  As a case in point, the Inno3D RTX 2060 (which has a rep for running hot)  in my own system was always throttling until I undervolted it. it performs substantially better undervolted as it does not hit throttle temp near as often.

[wthomas33065 994]

29 minutes ago, killthespam said:

I would like to remind to the "nonbelievers" that P3D was doing the same and XP11 for a long time suffered from the same issue, running video cards hot and fans continuously...

Let's concentrate  and ask ASOBO to optimize the software properly and stop being condescending to each other, it doesn't fix this broken simulator fighting between ourselves.

And I would like to remind people that the program automatically sets a suggested value for your graphics settings based on the detected hardware in your system.

If he is like most of us, he is choosing graphics settings that exceed those that are set by the program when it analyzes his hardware.  We all do this.  But some of us make sure we are using temperature monitoring software and carefully monitoring temperatures as we tweak our settings to see how the increased load affects our system temperature.

Pushing a card beyond it's suggested settings with a GPU that might have borderline heat management to begin with is a recipe for thermal throttling.

I have an i7-7700 with a GTX 1660 Super at 1080p.  I let my system pick High End and leave well enough alone.  I also locked my frames at 30 FPS which lowers the load on my card even further, and offers a much quieter experience as my fan speeds are lower as a result.

I can almost guarantee that if the OP sets his graphics settings to default using F12 he should not have a problem with thermal overload.  If he IS running the program at system defaults and STILL has issues, then I might say he has a valid complaint, but if he is like 99% of us here, I assure you that is not the case.

So let's concentrate and run the software with the settings suggested by the publisher and stop blaming the publisher for our own attempts push our own hardware beyond the limits suggested by them.

 

[overt 55]

1 hour ago, killthespam said:

Some people here cannot accept that it might be something wrong with the software as to not being properly optimized.

I don't understand how you can say this unless you aren't thinking it through.

Optimized doesn't equate to universally lower hardware usage, it just means that there is little to no waste in that utilization. But that's beside the point; his card should be designed to run at 100% load and dissipate the heat without issue, optimized or not.

If his card is overheating, it means he needs to focus on reducing temps, not reducing load. If your car is overheating while going the speed limit, the solution is not to reduce the speed limit.

 

[Drumcode 1,893]

40 minutes ago, wthomas33065 said:

And I would like to remind people that the program automatically sets a suggested value for your graphics settings based on the detected hardware in your system.

If he is like most of us, he is choosing graphics settings that exceed those that are set by the program when it analyzes his hardware.  We all do this.  But some of us make sure we are using temperature monitoring software and carefully monitoring temperatures as we tweak our settings to see how the increased load affects our system temperature.

Pushing a card beyond it's suggested settings with a GPU that might have borderline heat management to begin with is a recipe for thermal throttling.

I have an i7-7700 with a GTX 1660 Super at 1080p.  I let my system pick High End and leave well enough alone.  I also locked my frames at 30 FPS which lowers the load on my card even further, and offers a much quieter experience as my fan speeds are lower as a result.

I can almost guarantee that if the OP sets his graphics settings to default using F12 he should not have a problem with thermal overload.  If he IS running the program at system defaults and STILL has issues, then I might say he has a valid complaint, but if he is like 99% of us here, I assure you that is not the case.

So let's concentrate and run the software with the settings suggested by the publisher and stop blaming the publisher for our own attempts push our own hardware beyond the limits suggested by them.

 

I am running on high end, nothing is overclocked on my box, I never do this to my system unless the simulator version is stable and since this one isn't, the CPU is at factory clocks and so is the RTX. 

[wthomas33065 994]

11 minutes ago, Drumcode said:

I am running on high end, nothing is overclocked on my box, I never do this to my system unless the simulator version is stable and since this one isn't, the CPU is at factory clocks and so is the RTX. 

Your system should be able to run what settings the program picks based on your configuration.   If you suspect thermal overload, then you should download CPUID HW Monitor.  It's free and it will monitor a whole lot of stuff and will keep tabs on max values.  Run that in the background and see if your max temp on your GPU ever gets to the thermal design limit.  If your GPU is spiking beyond 89C (TMax for the RTX 2070), then you have a thermal issue.

If you don't hit a thermal limit, then your issue is being caused by something other than heat.

https://www.cpuid.com/softwares/hwmonitor.html

[Scottoest 1,561]

I'd be curious to see what your benchmarking info in the developer mode says about GPU utilization?  I'd also run a utility that monitors temperatures on the card and in the case while you're using MSFS, to see if it's getting up to sustained bad levels in the minutes before a failure.  Do you see visual glitches or artifacts in the minutes before it fails? Does the sim feel like it's struggling to run in a way that would suggest something is over-taxing your GPU?

It's entirely possible that there could be a weird bug in the sim causing some sort of bad loop that is spiking your GPU utilization until it fails from thermal problems. But if it was a bug like that, there'd be other signs beyond simply losing screen output for a few minutes.

It's definitely not an "optimization" issue.  Optimization just means how efficiently the sim uses the power available to it.  If the sim was unoptimized to the point of melting your card, it would also be running in an unplayable state in the time leading up to that hardware failure.

This sounds more like either a driver conflict leading to eventual crash, or a flawed card.  The fact that you might be able to run P3D without the same problem doesn't necessarily mean anything (and when is the last time you tested that?).

I remember I had an overclocked CPU years ago that would cause me problems - it would run fine, until I was doing something that taxed it enough, and then it would fail.

 

[Glenn Fitzpatrick 3,798]

48 minutes ago, wthomas33065 said:

  If your GPU is spiking beyond 89C (TMax for the RTX 2070), then you have a thermal issue.
 

https://www.cpuid.com/softwares/hwmonitor.html

TMax is the point after which damage might occur. It is like Vne in an aircraft.

Graphics card are designed to thermal throttle well before Tmax, in the case of the 2070 that will likely start to occur from around 82 or 83 degrees.

Regardless, the OP potentially has a heat issue with his setup in the game.  Assuming it is actually heat (it could be bad VRAM in the upper registers for example)  either:

1. There is a software issue with the game and his drivers etc that needs sorting out

OR

2. The game simply stresses his card more than other games and it is not coping with the current cooling setup, in which case the solution is either cool the card/system better or undervolt the GPU to try and get the same frequencies at a lower voltage. Most GPUs by default have a higher than needed voltage just to make 100% sure all cards shipped are stable. 

 

 

 

 

Edited October 15, 2020 by Glenn Fitzpatrick

[Elvensmith 165]

6 hours ago, Glenn Fitzpatrick said:

Be aware that badly written game code CAN cause hardware issues. EVE online was infamous for it about 10 years back, here is just one thread about EVE destroying graphics ram this issue was real and acknowledged by CCP games:

https://oldforums.eveonline.com/?a=topic&threadID=1029933

 

As did the first version of N3V's Trainz A New Era (TANE). One of the few times I've experienced thermal shutdowns of the PC while trying to build a route in Surveyor. At first the developer, encouraged by their fan boys and faithful acolytes, tried to blame the user even down to asking if we owned a pet then suggesting our PC's and laptops were clogged with cat hairs. However as if by magic the first service pack came out and the CPU and GPU temperatures dropped from 99C to around 60C where they ought to be...

Edited October 15, 2020 by Elvensmith

[MarcG 2,980]

2070Super user here, stock settings the GPU usage is high 90s and often at 100% (Vsync turned on). Runs at a cool 65c with no problems whatsoever (Air cooled system).

 

Of course the first port of call is to always blame the software straight away (...if I had £1...), chances are it's something on the users end, so I would always check your own equipment first before laying blame at others.

Edited October 15, 2020 by MarcG

[Noel 4,502]

14 hours ago, killthespam said:

This MSFS2020 is not properly optimized (CPU and GPU) and will raise temperatures, you can run P3D for hours and it will not have any issues with the temps, as he mentioned.

It's the other way around here.  P3D, w/ HT enabled which is my preferred approach runs a full 10-13C hotter than MSFS does w/ HT enabled, no matter how many hours either is running.  The GPU runs exactly as it would dependent on load, and for that matter so does the CPU in either sim.  MSFS' main thread in very busy areas ranges from 40-70% tops, but the average of all cores comes in there at about 28%.  GPU is always <70C.

[honanhal 1,420]

1 hour ago, Noel said:

It's the other way around here.  P3D, w/ HT enabled which is my preferred approach runs a full 10-13C hotter than MSFS does w/ HT enabled, no matter how many hours either is running.  The GPU runs exactly as it would dependent on load, and for that matter so does the CPU in either sim.  MSFS' main thread in very busy areas ranges from 40-70% tops, but the average of all cores comes in there at about 28%.  GPU is always <70C.

I’m also having the opposite experience. P3D has always run very hot. While the MSFS original release was roughly similar, after the first patch it has runs about as cool as most other games and about 10 degrees cooler than P3D.

[LRBS 948]

So I have a question please.

I see  lots of people with GPU knowledge here, is any reason why it says Limited by GPU.

My specs

I9-10900 K @ 4.9 GHZ

64 GB MEM @ 3400

CPU @ 40%

GPU Titan RTX (24 GB) @ 90%

I had it OC @ 5.4, 5.3, 5,2 and so on with the same results so I decided to stay at 4.9.

https://drive.google.com/file/d/15sWXhi92oWdy3pkiE64nFBaWEXO_Tw_K/view?usp=sharing

Thanks.

 

[wthomas33065 994]

1 hour ago, killthespam said:

So I have a question please.

I see  lots of people with GPU knowledge here, is any reason why it says Limited by GPU.

My specs

I9-10900 K @ 4.9 GHZ

64 GB MEM @ 3400

CPU @ 40%

GPU Titan RTX (24 GB) @ 90%

I had it OC @ 5.4, 5.3, 5,2 and so on with the same results so I decided to stay at 4.9.

https://drive.google.com/file/d/15sWXhi92oWdy3pkiE64nFBaWEXO_Tw_K/view?usp=sharing

Thanks.

 

There is always a "bottleneck" even when there is no actual bottleneck.  According to the snapshot, the GPU is the one that takes the longest to complete (13.2ms).  Thus it is the limiter, If you hit an area where main thread momentarily had the highest number than it would say "limited by CPU". Limited by GPU is a good thing especially if your GPU is greenbar.    If you are limited by CPU, it means your GPU is sitting around waiting for content.  

[whitav8 98]

I believe it was mentioned before about overheating that using RTSS (Riva Tuner Statistics Server ) can be setup to limit frames at 100FPS or so and therefore it will make MSFS2000 pause and run cooler