August 12, 20214 yr Lots of misconceptions here. High temperatures are not due to "bad coding" - quite the opposite. You WANT your card to work as hard as it can - this increases the frames per second your game renders at. Everyone involved (card manufacturer, software devs) is trying to make it go as fast as they can. Complaining about the "GPU working too hard" is like complaining that your car can go too fast when you fully floor the accelerator. But contrary to your car, graphics card have the accelerator nailed to the floor, they will always go as fast as they can. Now just like in your car, going full throttle will also make the engine run hotter. If you play a game that does not make your GPU run hot, it means that it is not getting used to its full potential (aka "GPU bottleneck") and you are missing out on potentially having your game run faster/look better. If you are worried about your card getting "too hot": If you are using an approved cooling system and there is no undue buildup of dust or other obstructions, don´t worry. The manufacture should have designed the card so that even prolonged full bore should not harm it. If you buy a more beefy card, chances are that it will not be fully utilzed and therefore "run cooler" - however this also means that you spent money on something you don´t really need. One more aspect is the volume of cooling fans - it is true that more expensive cards often have a higher quality cooling system that does not make as much noise. Also don´t forget that temperature measuring itself isn´t quite as standardized as one would think. I am not sure that there is a standard as to where and how accurate the temperature of the card is measured, and I wouldn´t be surprised if the temps measured are not very accurate at all. In fact, the manufacturer probably has no interest in showing you how hot your card REALLY gets. Edited August 12, 20214 yr by Janov
August 12, 20214 yr 5 hours ago, Waldo Pepper said: In the nVidia control panel there is a power management option, if set to "prefer maximum performance", the GPU will maintain it's highest clock speeds at all times, and never dynamically down clock in response to gpu utilization percentage. Windows has a similar power option, which can have a similar effect on your cpu when it's set to high performance. War Thunder and FS20 both load my gpu to 100% in the Hangar. My Asus Strix 3090 draws 480 watts max, limited by it's bios. My overclocked 10900k can pull up to 320+ watts. Sitting in fs20's hangar my system pulls 600 watts just doing nothing. Add 150 watts for my oled/cable modem/router/ Sim peripherals. According to the battery backup UPS, I've seen it pull 770 watts while flying over NYC. HWinfo64 can show your power statistics and temps as well. I use the portable version, which doesnt require installation. It just runs wherever you unzip it. https://www.hwinfo.com/ MY gtx1070 runs at 100% almost constantly in msfs.. so adjusting the power managment in the Nvidia control panel would be pointless ?
August 12, 20214 yr 5 hours ago, killthespam said: I don't know what happened. It didn't look like that on my phone, but I see it now on my PC. That iGPU was additional info that I found, you can disregard it as well the whole info if bothers you. Again. I don’t know the source of this. But it’s just wrong. Sure. If you undervolt it too much you sure get instability. But otherwise that’s just nonsense. Edited August 12, 20214 yr by swiesma
August 12, 20214 yr 5 hours ago, neil0311 said: So why don’t the manufacturers “factory under-volt” them if there’s no penalty to pay and you get better performance with lower heat? Sounds like nirvana. Because the OEMs sell 120% stability 😉 Just watch the video above and read some sites. Just google „undervolting 3080“ or for Ryzen CPUs „Ryzen undervolting pbo2 curve optimizier“. let me ask you this: why is there chip tuning for cars? It’s the same thing basically. You use the boudaries more than the OEM. But at undervolting you are not risking to destroy anything. The only thing that happens at GPU undervolting is That your driver crashes. Edited August 12, 20214 yr by swiesma
August 12, 20214 yr As @Janov pointed out. That temp is normal. Don't worry. Running on ultra settings at 4K, yeah. The card is gonna do the max it can. And, depending on your cooling setup, will run pretty warm. 83 degrees sounds about right for a closed case, air cooled setup. You can always run at 2K and that setting will not tax the video card as much and therefore, run cooler. Those who are posting their card's temps should also post the resolution and settings they are running. Otherwise, it's not a valid comparison unless they too are running 4K with ultra settings. Richard Chafey i7-8700K @4.8GHz - 32Gb @3200 - ASUS ROG Maximus X Hero - EVGA RTX3090 - 3840x2160 Res - KBSim Gunfighter - Thrustmaster Warthog dual throttles - Crosswind V3 pedals MSFS 2020, DCS
August 12, 20214 yr As I did a lot of testing I will just leave you this for a 3080ti (Power Target set to 110% to not limit the card, EVGA 3080ti FTW3 Ultra): Clock/voltage - memory freq - temps from HWInfo - Power consumption - 3d Mark Score 1890/900 - 0 Memory - 85/92/91 - ??? Watts - 18519 1906/900 - 750 Mem - 87/94/94 - 407W - 18833 1815/863 - 710 Mem - 80/88/87 - 345W - 18136 1740/837 - 600 Mem - 78/88/84 - 310W - 17692 1770/831 - 700 Mem - 78/88/85 - 317W - 17931 1725/800 - 700 Mem - 74/86/81 - 289W - 17735 1800/831 - 700 Mem - 69/80/76 - 313W - 18050 1860/862 - 700 Mem - 73/84/80 - 350W - 18595 1890/869 - 710 Mem - 75/84/82 - 360W - 18778 Now tell me undervolting is not worth it. Edit: I don’t know why I feel I need to defend something here. Read it, tune your system or leave it. It’s up to you. But spreading wrong information is not good. With these guides you get a good 10% more performance from your components coming with less power draw and lower temps. And it’s for FREE. You just need to invest some time and find your settings. That’s not some witchcraft that will destroy you PC. Edited August 12, 20214 yr by swiesma
August 12, 20214 yr 2 hours ago, VegasPilot said: I removed the case door from my computer and my gpu ran on average 7c cooler or 44 F cooler.... Its a no brainer,., I dunno why pc makers put doors on these things.. must be to sell fancy cooling systems... ^This While I'm also in the '82 degrees for long flights isn't going to cook your GPU' camp, I also think that cooler is better. So as well as the improvement you got from cleaning the fins, etc, @Darren_Beattie , it's worth looking at your cabinet and where the cooling air is coming from and going to. It's worth remembering that the past few years the cabinet makers have let their marketing departments swing towards 'form over function'. I was intrigued with my new desktop that the three multi-colour-flashing front fans were displayed nicely immediately behind a smoked glass screen. So where was the fresh air coming from?. And it took some searching until I found a line of teeny slots in the front casing either side of the glass. I took off the front panel to experiment (DON'T do this with kids and pets around) and immediately and sustainably the temp of my 2070 Super dropped 12 degrees c (53F) from previous levels. So I took the glass out, replaced it with a finger-safe but relatively wide plastic mesh (a few £'s only from an auto spares outlet) and the GPU has been running 10 degrees c (50F) cooler ever since. And I note that the cabinet makers are now advertising 'new airflow cabinets', where they seem to have replaced the glass fronts with an open mesh - at a modest premium, of course 😉 Edited August 12, 20214 yr by AJZip Added Fahrenheit conversion Ryzen 7 9800x3D @5.2GHz; ASUS X670-P Motherboard; nVidia 4080 (factory o/c); 32G 5600MHz DDR5 SDRAM; Pimax Crystal Light VR Headset; Quest 3 VR Headset
August 12, 20214 yr If i use afterburner my temp (2080ti) stays below 80 but my fans are running at almost 90%. Using the NVidia OC feature, the temp rises up to 85 but the fans are a lot quieter. The question is which of the fans or the chip will last longer ? I bet 85 deg for a chip is ok…
August 12, 20214 yr Not sure why anyone would want less performance? As others have said here, MAX IT out! that is what it is designed for......
August 12, 20214 yr 2 hours ago, Ianrivaldosmith said: Not sure why anyone would want less performance? As others have said here, MAX IT out! that is what it is designed for...... Yeah, if your fans sound like an Apache Helicopter you would maybe think twice....
August 12, 20214 yr 6 minutes ago, swiesma said: Yeah, if your fans sound like an Apache Helicopter you would maybe think twice.... Unless you are flying an apache helicopter in which case it would only add to the immersion.
August 12, 20214 yr Can't argue against this..... 🙂 Will install this on Monday. https://photos.app.goo.gl/asV1fF8pfs2bvvu8A Edited August 12, 20214 yr by swiesma
August 12, 20214 yr 14 hours ago, killthespam said: IMHO, this is not the correct fix to downgrade a good video card just to accommodate faulty coding and potentially create more problems. Again, it seems that some prefer to do it versus reporting the problem as an issue and have it fixed by the developer. Hardware should be able to cope with sustained 100% load and if it can't, it is the problem, not any code. Whilst code could be wasteful, bloaty and inefficient making things hotter than they need to be, good code for demanding tasks will fully utilise the available performance to give the best result (be that visuals or framerate) and will generate lots of heat anyway. When a GPU or CPU is overclocked, any sensible person stress tests them to check they remain within safe thermal limits and are stable. If you haven't overclocked or modified a system, provided the coolers are clean, it should be fine to run 100% all day. It's very hard to kill hardware with temperature since most have thermal throttling. If you are hitting the thermal throttle on your graphics card, you need to be taking that up with whoever manufactured the card as the cooling design is obviously inadequate. Moaning to the people who wrote the game code is not going to solve anything. If your card cannot cope with being pushed 100% all the time without throttling, reducing graphics quality so it no longer needs throttle is the easiest cure, beyond that you are into the world of undervolting or cooling modifications to cover for the original graphics card's cooling inadequacies. Edited August 12, 20214 yr by ckyliu ckyliu, proud supporter of ViaIntercity.com. i5 12400F, 32GB, RTX4070, more in "About me" on my profile.
August 12, 20214 yr I had to enable the fps cap to cool down within the menus, and this is where perhaps they could do something in the code ?
Archived
This topic is now archived and is closed to further replies.