Crash in rdr2, advice please.

Holst1981

Holst1981

Holst1981

Associate
Joined
1 Dec 2015
Posts
1,194
Hi all.

Pc has been running fine for months but I've not been gaming much or if I have it's been wreckfest

I recently reinstalled rdr2 as I've not tried it on the 3080ti but was disappointed with my PC completely crashing once the game has run for a few minutes.
I just get a black screen with no light on the GPU, pressing reset restarts the PC, power button does nothing....

Spec is

3080ti FE no overclock
5950x no overclock
MSI tomahawk max (latest bios)
32gb 8pack B die (4x8gb)
Seasonic Focus Plus 750W 80 Plus Gold

All air cooled in a pc011xl

Running win 10 with latest drivers.

First thing I tried was dropping the memory speed to 2666, when that didn't help I removed half the ram and then swapped the sticks. Crashed exactly the same regardless of memory settings.

I can run cinebench and 3d mark without issue.
I can also play wreckfest with no crashes.

Temperatures are ok given the high ambient and everything is clean.

I did get the rdr2 benchmark to complete with the game on lowest graphics settings and resolution 1024x768 60hz
Once I changed back to 1440p 165hz the crash was back.

So what do you reckon the problem is?
GPU
PSU
Motherboard?

Any other tests or ways to narrow it down to one component?

I don't have any spare parts.

Thanks for any help.
 
My immediate thoughts would be that power usage is spiking/peaking and the 750W PSU is unable to provide enough power.

I believe 750W is listed as recommended/minimum for the 3080, not sure if it's still the case for the 3080 ti or if you need to consider switching out to a 850W.
 
nade said:
My immediate thoughts would be that power usage is spiking/peaking and the 750W PSU is unable to provide enough power.

I believe 750W is listed as recommended/minimum for the 3080, not sure if it's still the case for the 3080 ti or if you need to consider switching out to a 850W.
Click to expand...

Yeah that's my thought too. 750w is the minimum recommended so should be okay but I have read that some others had problems, changed to higher wattage and problems went..
 
VortexA1 said:
Yeah that's my thought too. 750w is the minimum recommended so should be okay but I have read that some others had problems, changed to higher wattage and problems went..
Click to expand...

Thanks for the suggestions.

The PSU should be ok, and has been ok previously. Maybe the hot weather pushed it over the edge.
I'm going to try a lower power limit on the GPU and see if that sorts it out.

I'm reluctant to buy a new PSU especially as this one has a 10 year warranty but I don't want any down time.
 
I tried 90% gpu power and it still crashed.

Tried 50% and it didn't crash.

Now trying 75%.

I'm going to remove and check all the GPU cables next.

I don't have another PC to test the GPU, how can I tell if it's GPU or PSU?
 
Holst1981 said:
Thanks for the suggestions.

The PSU should be ok, and has been ok previously. Maybe the hot weather pushed it over the edge.
I'm going to try a lower power limit on the GPU and see if that sorts it out.

I'm reluctant to buy a new PSU especially as this one has a 10 year warranty but I don't want any down time.
Click to expand...
Yep no worries. Testing older GPU drivers will be easier to test.

You could also try downloading a program like OCCT (https://www.ocbase.com/download). I used this last year when i sold a GTX1060 card and wanted to run some stress testing. Could be worth running some stress testing against specific components using it and see if you can re-create the issue.
 
Holst1981 said:
I tried 90% gpu power and it still crashed.

Tried 50% and it didn't crash.

Now trying 75%.

I'm going to remove and check all the GPU cables next.

I don't have another PC to test the GPU, how can I tell if it's GPU or PSU?
Click to expand...
Tbh I think that if the PSU is in effect functioning 'normally' but can't handle everything running with transient spikes while gaming now the only way to find out would be to try a higher wattage excellent quality psu and see if issue disappears unfortunately. The usual voltage tests would come through fine on the psu if its hitting its transient limits now .

Best way is swap them out.hassle I know. If it's indeed the PSU you can RMA it saying it used to work and basically not now and get replacement then sell/send back ..something like that anyway.

That's my thoughts anyway.
 
Last edited:
Testing with games is not really the best methodology. It is actually good that you have repeatable results.

If you want to stress test the psu with your components then run blender and furmark/Heaven at the same time. I know you said you have run benchmarks but not sure if you ran them simultaneously to stress the psu. If your psu is faulty or not upto the task it will shut down and should exhibit the same symptons as the RDR2 shutdowns.

Have HWmonitor open and see how much power the 5950x and 3080ti are pulling. 5950x should be in the region of 240w and the 3080ti 300w. Your psu should be upto the task of powering your system but it may be faulty or the power spikes may be triggering the over power protections.
 
Im running OCCT power test

GPU is using 350watts temperature is 78 after 5min

CPU is only using 132watts and temp is 63.... seems low.
This may be due to my setting the ram so slow, I might put it back to something faster and try again.
 
Its just finished another 15min run on OCCT power test with the memory at 3600.

Benchmarks were better with the faster memory.

So maybe I have no hardware issue but a RDR2 issue.....

Temperatures were all fine.
 
Nope..

Also crashed in Blender CPU test, exactly the same black screen GPU powered off.

I'm now thinking motherboard.
It was great with the 3600 but the 5950x was never great especially with 4 sticks of ram.
Maybe its time to swap the b450 for something else, preferably something that can run 4x16gb b-die at a decent speed.
 
This is progress. Now that you have a benchmark where you can test you can alter the memory timings and speeds till you find the stable point. I know it is not fun to painstakingly test configurations to find the stable timings but it is cheaper than buying a new mobo.

I found Prime95 blend test to be good to test for memory testing, if it was unstable it would drop worker threads and not crash the entire PC
 
Holst1981 said:
Nope..

Also crashed in Blender CPU test, exactly the same black screen GPU powered off.

I'm now thinking motherboard.
It was great with the 3600 but the 5950x was never great especially with 4 sticks of ram.
Maybe its time to swap the b450 for something else, preferably something that can run 4x16gb b-die at a decent speed.
Click to expand...
Have you tried running the memory at default settings (xmp disabled)
 
You could under volt the GPU. Miners do that all the time. Ampere GPU's get hot video memory, and sometimes when the GPU's are put together the thermal pads are not all that great.

I suggest you try MSI Afterburner to under volt the graphics card, just check for recommended under volt for gaming?

That way you can rule out if CPU or GPU perhaps.

I am suspicious that it crashes with RDR2, that game puts a GPU through the grinder.
 
Sorry for not replying sooner.

So I tested a load more times and decided that this was most likely motherboard related.

So I ordered a MSI 570 tomahawk and just installed it. (Very fast delivery from OCUK, thanks)

So far... completely fixed :D
No more crash, just installed the wifi drivers and ran the rdr2 bench a few times over.

As a bonus XMP also works now although my benches are a bit low.

Going to actually play some rdr2 rather than mess with the PC now, hope it wont crash :)

I will do a chipset driver reinstall.

I wont bother to overclock the CPU or GPU, they are faster than I need for the moment.
 
You must log in or register to reply here.
Back
Top Bottom