cancel
Showing results for 
Search instead for 
Did you mean: 

Graphics Cards

ThunderBeaver
Miniboss

I used to have problems with my RX 6900 XT crashing till I did this.

I never use the fan curve setup option to react to GPU temps and adjust fan speed.

I always have my GPU fans at a constant speed 60% to 70% for standard applications and 75% to 90% for gaming just by creating a flat line out of the fan curve in Adrenaline.

For the last 3 weeks I've kept my GPU fan speed at 75% on all applications. No crashes have occurred sense I've done this.

I suspect either vibration from the RX 6900 XT triple fan cooling unit EMI (Electro Magnetic Interference) or both is the cause of these crashes sense they only happen when I run the fans over 80%.

I know that PC's can be sensitive to vibration and EMI.

Throwing this out there so others who have crashing 6000 series GPU's can try these settings and see if it reduces or stops crashing problems that seem to have no logical reason for occurring.

Hope information provides help to my fellow gamers and PC builders.

6 Replies

I had a simular issue that a cpu fan bearing failed and caused a lot of vibrations, the celeron never overheated, but when the fan would speed up at the level it causes vibration, the computer would crash.

What I ment to say is GPU fans can reach quite high speeds and at 70% they are most likely spinning at about 3000rpm and they produce quite a bit of vibrations witch could cause a crash. The reason I think it is not emf is because, computer fans are made not to produce it and because there is a huge chunk of metal between the fan and the card witch absorbs emf radiations.

You make a valid point about EMI.

Aluminum and Copper are primary material used in faraday cages which shield from EMI or EMP depending on the design specs and applications.

I just find it odd that not running my GPU fans over 75% and I have no display driver crashes.

Ramp those fans to 80% or 90% I experience infrequent crashes about 1 a week.

Put the fans at 100% and I experience 2 to 3 display driver crashes per week.

If you want to get tehnical, by increesing the fan speed, you increesed the oscilation frequency of the motor, increesing the frequency, at higher frequencies, you are closer to the resonant frequency  of the card (that is hanging on 2 screws and a connector) Closer the frequencies are one to another, more  vibrations will be produced in the card, at some point vibrations will drasticly increese, in your case 70%.

At least I assume it has to do with resonance since you said there is a cirtuain treshold at witch it becomes a problem.

I dont think resonance had anything with the celeron exapmle I gave because I am preety sure it was just a bad LGA775 contact under a huge chunk of mettal beeing shaked by a fan with a bad bearing.

clerian
Adept I

It's possible there's a weird performance threshold where your fans aren't providing enough cooling at 70% for certain standard applications that might leverage GPU acceleration.

hrpuffnstuff
Miniboss

Are you using the reference card?  I have a XFX Speedster MERC 319 RX6900XT black with a triple fan that is set to dynamic based upon the temperature and each PCIE plug is on its own rail so I've never had an issue with it in this configuration.   It also will run over 2.3ghz maximizing the load on the PCIE rails like when running Time Spy.

Back when I first installed it I was using an enermax platimax 1350 PS and when stressing the card it would crash until I separated the rails far enough apart.  For some reason if I used numerically adjacent rails it would crash even though they were all rated for the same amp draw.  My current EVGA supernova G2 1300w also has separate and equal 12v rails that never sag under load so the card never acts up when pushed to a 300w draw load.

I would run Time Spy and see how it behaves under load and see what power draw AMD wattman reports.

I'm running the Power Color Demon RX 6900 XT.

Even with my GPU fans at 75% my idle temps are low 20's and load temps only hit mid 30's.

Absolutely no thermal issues.

Been running my setup like this as a test for a month now.

I haven't even encountered the annoying "Global Wattman has reset Adrenaline settings to default for unknown reason" crash since setting my GPU up like this.

I have my VPU at 2235MHz and VRAM at 2170MHz fast timing.

I would have my VPU at 2.4GHz but my X370 MOBO and Ryzen 7 2700x at 4.2 GHz starts to bottleneck or choke out when I try to push the VPU past 2.75GHz. 

So I ran several stress tests and tested on multiple games. Found that these are the optimum settings on my GPU on my current build.

0 Likes