[SOLVED] Pulsating FPS in games following upgrade

[Obs1dian]

I've just upgraded my PC but kept my existing PSU and GPU. The GPU had been working well before in my old PC (i5 4th gen) and games ran generally smoothly. Now it is in my new build, when playing games e.g. Doom Eternal, the fans start "pulsing" and the FPS in game goes up and down like a yo-yo, as in 80-40-80-40 except that the counter is more of a garbled mess.

Weirdly, on my monitor when I look at the refresh rate using the built in menu the frame rate fluctuates 75hz down to 35 or so then back up every half a second.

I know the GPU is capable of running the game smoothly as I completed it on my old PC set up.

All my drivers are up to date and I've tested the PSU with a multimeter and the voltages appear to be solid. The PSU is quite old, it's a Thermaltake 750w semi modular. And when I say old, it could be 10 years old.

Temperatures for CPU/GPU/RAM etc. all seem fine.

I nearly bought a new PSU but if it could be something else I would rather not drop the cash on a new one if I don't have to.

Spec:
Ryzen 3600
MSI Mortar MAX motherboard
Crucial Ballistix 16gb DDR RAM
Sapphire Radeon RX480+ OC
Thermaltake 700w PSU
2x SSD (Crucial)
Cooler Master 240l CPU cooler

 

[sizzling]

The RAM, is it 1x16gb or 2x8gb and what speed is it?

Did the upgrade include changing the motherboard? If yes did you reinstall Windows?

 

[Obs1dian]

The RAM, is it 1x16gb or 2x8gb and what speed is it?

Did the upgrade include changing the motherboard? If yes did you reinstall Windows?

RAM is 2x 8Gb. It's 3000Mhz although I've overclocked it to 3200Mhz and I also tried not having the XMP profile on so it was running at 2400Mhz. The problem happens regardless of RAM speed.

The upgrade did include changing the motherboard and I wiped all SSDs and installed Windows from scratch.

I've just been stress testing using Heaven benchmark and the pulsating happens in only certain scenes but on this run through the entire system shut off.

When I turned system back on, fans in the case were going utterly nuts.

Edit: I also get coil whine from my GPU in certain game menus (e.g. loading Doom Eternal) and RMAd the card a couple of years ago with the place I bought it from. They gave it back to me saying there wasn't an issue with it.

At this point I'm feeling like it is the PSU regardless of the fact the voltages look good given the PC is turning off.

 

[sizzling]

What are the cpu and gpu temperatures when you his happens?

 

[Obs1dian]

What are the cpu and gpu temperatures when you his happens?

GPU is rock solid at 75 C. CPU seems to be around 40-50 C according to the sensors.

 

[sizzling]

Try running userbenchmark and she are the link to the results. It’s not the best tool but can sometimes help point us in the direction of the problem.

 

[Obs1dian]

Try running userbenchmark and she are the link to the results. It’s not the best tool but can sometimes help point us in the direction of the problem.

Didn't know about this thanks! Here's my report:
https://www.userbenchmark.com/UserRun/30784733

 

[sizzling]

Didn't know about this thanks! Here's my report:
https://www.userbenchmark.com/UserRun/30784733

Ok nothing stands out. I’m out of ideas at the moment.

 

[Obs1dian]

Ok nothing stands out. I’m out of ideas at the moment.

The fact that the PC is shutting down is making me think it could be the PSU needing replacing. Do you think this is a reasonable assumption?

It was shutting down at the end of DOTA games when the fountain blows, it shut down during Heaven benchmarking this morning and it does the pulsating thing in Doom Eternal, DOTA and sometimes in Heaven benchmark scenes.

 

[sizzling]

The fact that the PC is shutting down is making me think it could be the PSU needing replacing. Do you think this is a reasonable assumption?

It was shutting down at the end of DOTA games when the fountain blows, it shut down during Heaven benchmarking this morning and it does the pulsating thing in Doom Eternal, DOTA and sometimes in Heaven benchmark scenes.

Oh wow random shut downs are a bad sign. I’d definitely change the psu for this problem and I’d say it is risky using the system on a psu that is showing signs of failure.

Some people say a psu can impact gpu performance but I’ve never seen any evidence to support this.

 

[Obs1dian]

Oh wow random shut downs are a bad sign. I’d definitely change the psu for this problem and I’d say it is risky using the system on a psu that is showing signs of failure.

Some people say a psu can impact gpu performance but I’ve never seen any evidence to support this.

Thanks I'll buy one then! Not that there are many in stock...anywhere.

 

[sizzling]

Thanks I'll buy one then! Not that there are many in stock...anywhere.

Make sure to get a high quality one, there is a huge difference between high and low quality psu's of the same wattage rating.

 

[Obs1dian]

Make sure to get a high quality one, there is a huge difference between high and low quality psu's of the same wattage rating.

Thanks yeah I've been doing a lot of reading up. It seems all the decent ones I'm looking at are out of stock, overpriced or long delivery times. Looking in all the usual places.

 

[Obs1dian]

Make sure to get a high quality one, there is a huge difference between high and low quality psu's of the same wattage rating.

Reviving this as I'm still getting issues.

I've bought a Corsair RM750x - it should be all the PSU I'll ever need. Yet I'm still getting the stuttering.

I made a video using FurMark - computer is at completely stock settings with updated drivers. In the video you can see it starts smooth (not perfect due to YouTube) but then as the fans spin up the jerkiness starts and by half way through is really noticeable if you look at the edges of the picture.

View: https://youtu.be/3us6ZiEQJu8


Is the GPU dying?

Edit:
Another video I did showing the frames dipping up and down in Doom Eternal including the monitor refresh rate:

View: https://youtu.be/pnJA1pLhTuM

 

[Obs1dian]

Sorry about replying to myself here but here's a graph of FurMark performance on the GPU clock. Something doesn't feel right and this seems to coincide with the problem I'm getting.

GPU Clock

 

[Flayed]

Sorry about replying to myself here but here's a graph of FurMark performance on the GPU clock. Something doesn't feel right and this seems to coincide with the problem I'm getting.

GPU Clock

It looks like the GPU is downclocking due to thermals to me.

 

[Obs1dian]

It looks like the GPU is downclocking due to thermals to me.

Thanks, any idea why when it's not reporting above 80c? Anything can be done about this?

 

[tennis2]

These are the settings I use on my MSI RX480 GamingX 8GB (for now, I know the Nitro+ is at 1350MHz stock OC)

• Go into AMD Settings and click on the Performance tab at the top.
• One row below that, click on the Tuning tab.
• Set tuning control to manual
• Enable GPU tuning
• Enable advanced control
• My frequencies and voltages are as follows:
  • 300MHz / 800mV
  • 600MHz / 820mV
  • 910MHz / 870mV
  • 1075MHz / 910mV
  • 1145MHz / 930mV
  • 1190MHz / 945mV
  • 1235MHz / 970mV
  • 1305MHz / 1040mV
• Enable VRAM tuning
• Memory timing - automatic
• Advanced Control - enabled
• Voltage - Manual
  • Memory voltage sets the lowest GPU core voltage also. The memory and core get the same voltage, but the core has more frequency steps so you need a baseline voltage to keep the memory stable (the VRAM wont be stable at 2000MHz/870mV for example).
  • Going with a fairly safe 950mV on this.
• Fan tuning - disabled
• Power tuning - disabled

See how that works for you. I'm able to run 1425MHz rock solid in FurMark indefinitely. My card runs quite cool due to the fact that I replaced the thermal paste a year or two ago. At 1300MHz in FurMark I get about 73C @ 1000rpm fans with the frequency/voltage settings listed above.

 

[Flayed]

Thanks, any idea why when it's not reporting above 80c? Anything can be done about this?

I think it is usually around 83c when GPUs start to throttle. I'm not 100% sure though. How many case fans do you have? Maybe you could add one?

 

[Flayed]

If you have a large house fan handy then taking the side panel off and pointing a 2-foot fan at the inside is a good way to see if your issue is temperature related.

 

[Obs1dian]

These are the settings I use on my MSI RX480 GamingX 8GB (for now, I know the Nitro+ is at 1350MHz stock OC)

Thanks for that, interesting to try different settings. I mirrored those settings and immediately the computer crashed with a weird green/grey screen and a restart required. On the restart I had red streaky lines down the screen. Weird.

 

[Obs1dian]

I think it is usually around 83c when GPUs start to throttle. I'm not 100% sure though. How many case fans do you have? Maybe you could add one?

I have 2 fans at fron with radiator for CPU. A case fan at the back for exhaust and one at top for exhaust. 2 fans on the GPU itself. Can't see it being heat related as barely gets above 80cm - if ever.

 

[tennis2]

Thanks for that, interesting to try different settings. I mirrored those settings and immediately the computer crashed with a weird green/grey screen and a restart required. On the restart I had red streaky lines down the screen. Weird.

Like CPUs, each chip is different.

Try setting the VRAM voltage to 970mV first (which is the stock value for my card IIRC, or just leave it on auto). See if that helps.

Next you could try adding 10mV to each core frequency until you get stability. But if the card crashed before you applied a load, my guess is the VRAM voltage isn't enough to keep it stable.

 

[Obs1dian]

Like CPUs, each chip is different.

Try setting the VRAM voltage to 970mV first (which is the stock value for my card IIRC, or just leave it on auto). See if that helps.

Next you could try adding 10mV to each core frequency until you get stability. But if the card crashed before you applied a load, my guess is the VRAM voltage isn't enough to keep it stable.

Thanks for the advice but I've tried multiple combinations of voltages now and it either crashes immediately or stays stable but still get the juddering problem as per the FurMark video. Could the GPU be duff? Could it possibly be anything else in my system? If not I may take the plunge on a 5700xt...

Edit: The problem ALWAYS kicks in as soon as the temps get to 77 and then it starts dipping 76, 77, 76, 77, fan load is 98 - 99%. Then the core starts drtopping to 800-900 MHz. Seems like throttling to me?

 

[tennis2]

Was this card purchased new or used?

If the temp throttle point is set at 77C, that's pretty dang low. Usually it's around 90C. Might want to try flashing the VBIOS.

Not encouraging you to give up, but as an owner of an RX480 driving a 2560x1440 144Hz monitor, I can say that using an RX480 at 3440x1440 isn't going to give you good FPS in demanding games.