[SOLVED] AMD Threadripper 1950X system freezing while gaming.

[LightningsalesUK]

My system keeps freezing while playing games, when playing games such as Assetto Corsa, Overwatch and Minion Masters, Games such as CS:GO and BeamNG.Drivce have never had this freezing issue. At a random point while playing (shorter or longer depending on the game) it will freeze, the screen will be stuck on the frame it froze at. The PC will need to be turned off and on again to get it functional.
I put this thread with the CPU posts as think it's the CPU causing the issue. I have noticed the CPU temperature fluctuating randomly and my CPU idle temps seems to be a lot higher than when I first built my system (6 months ago). I have noted my CPU sitting at around 50-55c when idle compared to around 40-42 from when the build was first built, could this just be from the summer heat? I have also noted my CPU spiking randomly from its idle temperature to 65-68c causing the fans to power up before it quickly reaches the idle temperature again. Once this starts happening it will continue till the machine is turned off.
Since upgrading the GPU with the Kraken G12 the temps of my GPU had dropped significantly and doesn't reach over 50c under full load.

My specs are as follows.
AMD Threadripper 1950x (watercooled via Enermax Liquidtech 320mm)
32GB DDR4 3200MHz Corsair Memory
GTX 1080ti 11GB (watercooled via Kraken G12 + 140MM AIO, RAM heatsinks)
256GB Samsung M.2
120GB SSD
3TB HDD
1TB HDD
500GB SSD
Evga SuperNova 850W (with Both CPU power plugs in)

 

[buttabean2]

That seems really high for idle temp. Mines around 30c in bios

 

[buttabean2]

actually, do you have your fans set to turn off on idle? That could be an issue. There's a setting in the fan options to adjust how fast the fans update. I set mine to 0.1 second to ramp up. but 1 second to ramp down. That makes a big difference in cooling and noise factor.

 

[LightningsalesUK]

That seems really high for idle temp. Mines around 30c in bios

I just had another look at the BIOS, I activated Profile one, which enables D.O.C.P and it booted into Windows all well.

I opened up CPU-Z under the memory tab it shows the DRAM frequency of 1501.5MHz however if I go to SPD and check a RAM stick it shows the Max Bandwith as DDR4-2132 (1066MHz).
So it seems to want it to run at full speed but it just isn't doing so. The idle speed I am getting is around 45-50c minimum. The fans do not turn off at idle and they seemingly just running all fine, I could try unplugging the other 6pin CPU power if needs be.

 

[buttabean2]

No don't unplug the other plug. I don't think that's the issue.

Are you able to get to your fan settings. Make sure the cpu pump is set to pwm so its 100% all the time. it should be just running whether you're idling or not.

The Ram is actually being reported correctly. if you have 3000mhz ram. it's usually 1500x2= 3000. I have 3200mhz and run mine with tighter latency at 3000. I just checked what cpuz read and it's showing 1499.99mhz

 

[LightningsalesUK]

No don't unplug the other plug. I don't think that's the issue.

Are you able to get to your fan settings. Make sure the cpu pump is set to pwm so its 100% all the time. it should be just running whether you're idling or not.

The Ram is actually being reported correctly. if you have 3000mhz ram. it's usually 1500x2= 3000. I have 3200mhz and run mine with tighter latency at 3000. I just checked what cpuz read and it's showing 1499.99mhz

It seems to be showing two different speeds though. One is the 1500 x 2, and the other one is showing 1066mhz
I could only really see one pump show up in the BIOS, that seems to be the pump for the GPU
Last time I checked both pumps are plugged into pump connectors.

 

[buttabean2]

hmm, try putting the gpu on a separate connector and put that fan connector on pwm so it runs 100%. I've had the pump connector actually fry and have to use a fan plug to run the pump instead. maybe its getting overloaded running two pumps

 

[jnsupes]

I agree on not unplugging the power connector and on checking the fan curves. You definitely want the water cooler to be pumping consistently.

I guess this is the first time you noticed the CPU pump not showing in BIOS, did you try unplug and replug and double check correct connections? Or even using an alternate (or switching the GPU and CPU pump connections to see if it is the pump header or maybe the pump itself)?

 

[buttabean2]

the frequency and bandwidth are correct on the memory. bandwidth in the spd tab of cpuz is the chip info, not your settings. you see the timing below states jedec timing tables.

 

[LightningsalesUK]

hmm, try putting the gpu on a separate connector and put that fan connector on pwm so it runs 100%. I've had the pump connector actually fry and have to use a fan plug to run the pump instead. maybe its getting overloaded running two pumps

I just took the system apart, reset the RAM and added new thermal paste to the CPU.
I checked the conectors as well.
The CPU water cooler is plugged into an 'AIO_PUMP' and the GPU water cooler is plugged into 'W_PUMP'
The fans on the CPU are all plugged into the fan power marked with the CPU and have a curve which tries to keep the system under 55c at all times.
Since adding the new thermal paste my idle is now 42-45c and I haven't spiked to over 55c at all, the fans have not spiked in speed to compensate.

 

[jnsupes]

Now that you updated the BIOS and fixed all your settings, are you still getting the issues?

 

[LightningsalesUK]

Now that you updated the BIOS and fixed all your settings, are you still getting the issues?

I will try now.

 

[buttabean2]

What does hwmonitor say your cpu core voltage is? Is that set to auto in your bios? 42-45c just seems high to me, especially without a oc. I'll test mine with defaults and report back

 

[buttabean2]

 

[buttabean2]

My room temp is 74f/23.3C

What's your idle temp if you set your fans to all 100% I guess that would have been a good thing to try first haha.

 

[buttabean2]

My idle temps didn't really change much between fan defaults and max fan 100% . They also didn't change much with or without a oc as well.

 

[LightningsalesUK]

What does hwmonitor say your cpu core voltage is? Is that set to auto in your bios? 42-45c just seems high to me, especially without a oc. I'll test mine with defaults and report back

I tested with the most stressful test I can do on Assetto Corsa, rendering 55 AI cars at once at max settings which would normally cause a crash within 2 laps, I managed to do 20 laps without fail however I was reaching the 70c mark on the CPU, I will tinker with the fan settings again and see if I can bring that temperature down.

 

[buttabean2]

70c is high for a game. ill try that. what map were you using in assetto corsa?

 

[LightningsalesUK]

I was using Le Mans 24hr Day map.

I went back and cranked all my fans to turbo and now my CPU seems to be idling even higher. sits at 52-58c on idle??? I think this CPU has a mind of its own at this point.

 

[LightningsalesUK]

 

[LightningsalesUK]

 

[LightningsalesUK]

It now unfortunately sounds like an air tunnel as the fans are going crazy trying to keep the temps under 50c.

 

[buttabean2]

hmm that's really weird. Can you manually just set everything to 100% or you can't create your own curves?

Try setting your cpu vcore to my voltage. 1.295v(i've run mine at 1.187v at default speeds) the default was 1.4v for mine. when you have it set to auto voltage, it'll over compensate and will generate extra unnecessary heat. This is the main setting overclockers change to maximize efficiency and ocing. But you have to run prime95 to stress the cpu to see if its stable. In this case with default speeds, your cpu should be stable at default speeds and 1.295v. if it's not then i would think there's an issue.

I'll report back in a few with assetta temps

 

[jnsupes]

I tested with the most stressful test I can do on Assetto Corsa, rendering 55 AI cars at once at max settings which would normally cause a crash within 2 laps, I managed to do 20 laps without fail however I was reaching the 70c mark on the CPU, I will tinker with the fan settings again and see if I can bring that temperature down.

Being under that sort of load for that much time and hitting 70c is not that bad considering. I saw the update, yes the fans being cranked up will cause that. The only thing that you can do is find a place between the noise level and temperature in the fan curve that you can live with. For my CPU fans (I have air cooled Noctua ones) I typically go with a 40 percent at 40 degrees so it is silent and then slightly up until hitting higher temps. Since you are using liquid it would be good to keep a steady speed so the liquid keeps moving even at the the low temps and then increase it as needed.

Typically when gaming is when the CPU and GPU will be working harder and making more noise so if you play games at higher sounds you can tolerate the noise from the fans better. You will just have to aim for that sweet spot.

Since you were able to get the stuff working on your test, I would advise you to tweak the fans more towards what you might find as a sweet spot fan curve and try the games that you were having freezing with to see if the issue has been resolved and the temps are better too.

 

[buttabean2]

ok but do your testing and see how your temps go. The arctic fans are really cheap. I'll provide a link later with the ones i bought

 

[LightningsalesUK]

ok but do your testing and see how your temps go. The arctic fans are really cheap. I'll provide a link later with the ones i bought

I am currently using Corsair SP120 fans for the radiator.