Im actually thinking it isnt the PBO now. I've been running my 3080 at 900 1905 and it was fine in a few games but god of war and cyberpunk it just doesnt like so im testing it at the moment. PBO back on and the 3080 at 925 1800 and its much better.
Games randomly shutting down
- Thread starter tally2425
- Start date
Im actually thinking it isnt the PBO now. I've been running my 3080 at 900 1905 and it was fine in a few games but god of war and cyberpunk it just doesnt like so im testing it at the moment. PBO back on and the 3080 at 925 1800 and its much better.
I'm having the same issue with my recent upgrade.
Now running 12900kf, 4x16gb 5600 vengeance ram, 2x PCIe 4 m.2 wd black a 1tb and a 2tb. 3090. Strix LC ii 360 rad.
During testing I found my pc is failing the CPU test on time spy. I removed 2 sticks of ram earlier tonight and it passed time spy fully. I didn't get the chance to fully investigate but it was a start for me. I know DDR 5 has issues with ram speed. But as you have DDR 4 may be another issue.
That said you did say you just added more ram so start there. I upgraded to Windows 11 at same time thought it was that but since returned to 10 and issue persists
Now running 12900kf, 4x16gb 5600 vengeance ram, 2x PCIe 4 m.2 wd black a 1tb and a 2tb. 3090. Strix LC ii 360 rad.
During testing I found my pc is failing the CPU test on time spy. I removed 2 sticks of ram earlier tonight and it passed time spy fully. I didn't get the chance to fully investigate but it was a start for me. I know DDR 5 has issues with ram speed. But as you have DDR 4 may be another issue.
That said you did say you just added more ram so start there. I upgraded to Windows 11 at same time thought it was that but since returned to 10 and issue persists
Last edited:
Soldato
Having this same issue with a new set of DIMMs I got.
If I install my original ram, PUBG runs fine.
if I install the upgrade at XMP settings, it crashes after 8 or so mins of gaming.
Original ram: 32 gigs Vengeance RGB pro 3200 @ CL16 in Dual rank.
the upgrade: TeamGroup 3600 CL14 T2 in single rank dual channel.
Both sets of ram pass multiple RAM stability tests and I get no BSOD.
If I install my original ram, PUBG runs fine.
if I install the upgrade at XMP settings, it crashes after 8 or so mins of gaming.
Original ram: 32 gigs Vengeance RGB pro 3200 @ CL16 in Dual rank.
the upgrade: TeamGroup 3600 CL14 T2 in single rank dual channel.
Both sets of ram pass multiple RAM stability tests and I get no BSOD.
What ram stability tests?
If it's not tm5 anta777 or karhu it doesn't count as being stable in my book. If you did run one of these and it passed the minimum 3 cycles of anta then the only other explanation is heat. The extra voltage on the DIMMS/different dies is going above normal temp tolerances. Usually this isn't an issue but since your sig says you have a 3080 FE in such a small case it's likely burning the DIMMS. Ofcourse I could be totally wrong but based on what you said it's worth a mention.
If it's not tm5 anta777 or karhu it doesn't count as being stable in my book. If you did run one of these and it passed the minimum 3 cycles of anta then the only other explanation is heat. The extra voltage on the DIMMS/different dies is going above normal temp tolerances. Usually this isn't an issue but since your sig says you have a 3080 FE in such a small case it's likely burning the DIMMS. Ofcourse I could be totally wrong but based on what you said it's worth a mention.
Soldato
It is TM5. I know that certain tests dont work entirely to ensure stability.
Im between chassis and my current system is installed in a P400a with the side cover off and an extra fan extracting air out the top (I lock my BIOS so often with RAM overclocking that its easier to just leave the sider cover off, I should really have a test bed but hey)
It could be heat in all of our situations. It most likely is since the time it takes to crash is slightly longer than the TM5 takes to run.
Going to test the temps and see what happens.
You need to select the anta777 profile via the "Load config and exit" button then run the program again. I use this as it has all mainstream profiles: https://www.overclock.net/threads/memory-testing-with-testmem5-tm5-with-custom-configs.1751608/
I bring it up cause you said it takes slightly longer than TM5 takes to run but the proper stability test of anta777 is just under 3 hours. You said 8 minutes before a crash previously. Just making sure no break down in communication before you go for more elaborate measures. If you don't have 3 hours there's also another profile called 1usmus_v3 that takes about ~30 minutes. Not as good as anta but usually finds stuff fast.
I bring it up cause you said it takes slightly longer than TM5 takes to run but the proper stability test of anta777 is just under 3 hours. You said 8 minutes before a crash previously. Just making sure no break down in communication before you go for more elaborate measures. If you don't have 3 hours there's also another profile called 1usmus_v3 that takes about ~30 minutes. Not as good as anta but usually finds stuff fast.
Last edited:
Soldato
Thank you!!
Soldato
Started Tm5 and ran the anta777 script. Seemed happy after 35 mins so I opened up PUBG and started a game.
I'd already loosened out the timings from CL14-15-15-15-35 at 1.45v to CL15-16-16-16-35 at 1.5v and it's gone from an drop to desktop to a BSOD of constant crash to pubg crash reporter.
Will try looser timings and running TM5 for longer to test absolute stability.
I'm not sure what other people have tried.
35 minutes is generally a semi decent indication of stability so it does seem like temperature is your problem. If you can make it to the second cycle starting (~1 hour) that will be okay too. Last month I was stress testing a CPU for 90 straight hours so 3 hours is nothing to me.
What you should be doing is the following (if it is temp related): Run the standard XMP 14-15-15-35 but at a lower voltage like 1.40v. Kits of this caliber have ridiculous headroom to lower voltage. I've seen them run xmp profile at 1.37v before.
Lowering voltage might take like 3-4c off your peak temps that it will then be stable. If you still crash then yeah try something ridiculously lower like 16-16-16-36 at 1.35v. If you still have crashes at 1.35v something is horribly wrong with cooling or the RAM is just bad.
Soldato
I’ll see how long I can let it run before my GF kills me.35 minutes is generally a semi decent indication of stability so it does seem like temperature is your problem. If you can make it to the second cycle starting (~1 hour) that will be okay too. Last month I was stress testing a CPU for 90 straight hours so 3 hours is nothing to me.
What you should be doing is the following (if it is temp related): Run the standard XMP 14-15-15-35 but at a lower voltage like 1.40v. Kits of this caliber have ridiculous headroom to lower voltage. I've seen them run xmp profile at 1.37v before.
Lowering voltage might take like 3-4c off your peak temps that it will then be stable. If you still crash then yeah try something ridiculously lower like 16-16-16-36 at 1.35v. If you still have crashes at 1.35v something is horribly wrong with cooling or the RAM is just bad.
Going to do the thing that you suggest and go from there.
I suspect that my motherboard and cpu might not be up for the task. Been meaning to get a 5700X/5800C for ages.
Soldato
Had a spare SSD and did a quick windows 11 install to see if that was the problem.
PUBG is a lot more stable now. RAM can run 3600CL14-14-14-34 easily.
I think that there is a memory leak on the previous installation and that was causing a lot of the problems.
PUBG is a lot more stable now. RAM can run 3600CL14-14-14-34 easily.
I think that there is a memory leak on the previous installation and that was causing a lot of the problems.