Microfreezes in games

Hi all!

I have started encountering this weird problem.. Games would be running at their usual FPS and all and suddenly the screen freezes and the audio turns into this static/crash sound that you hear when the BSOD shows up but after 0.5-2 seconds (depends) the game just continues running.

I don't know if there is any rule to it it seems to be irregular although new areas provoke it.

So far I noticed it in: Metro Last Light Redux, Heroes of the Storm and PubG (I know it's only early access but it wasn't happening before..) as those are my most played games at the moment but the same thing is happening in all of them so I assume it's a general issue.

So far I tried: rolling back nVidia drivers, uninstalling Geforce experience, reverting the CPU undervolt (usualy set at -140V core/cache), setting bios to default settings (was running 3000mhz memory via profile), tried with and without g-sync, forcing the SSD to never go into sleep mode, switching battery settings to performance also I stress tested the GPU using 3dmark (funny enough didn't happen on the one run I did in 3dmark), CPU using intel XTU for 30min, did the RAM memory windows test and installed crystal disk to see the integrity of the SSDs. Everything seems to be doing fine. The temperatures seem normal too as GPU peeks at 71-73 and CPU is similar

Therefore I ask what might be the problem? It's not something completely dealbreaking but it just feels like the laptop is going to crash and especially in online games is just horrible sometimes.

Will be very grateful for any ideas on how to fix it!

All the best to all you wizards!

Cheers

 

Hi @0Style! Can you post your model and build for us, including the OS you're running?

Has this been going on with a clean install of the OS, stock drivers and all?

 

Hi! Sorry forgot the most important thing haha

It's a P751-DM2:

1070GTX @ stock clocks and voltage
6700K @ 4.0ghz with -140V undervolt
G-Skill 16gb 3000mhz (although now running 2400 as mentioned)
1 512gb Samsung SM961 + 1 500GB samsung 960EVO
Win 10
It has Prema Bios installed but I've had it on since beginnning 2017 and it wasn't an issue.

Everything was running fine and then I repasted the laptop with IC diamond and added the 960EVO as well as did a clean Win10 instal to refresh the system.

After that I had absolutely no problems for like 2-3 weeks and it started happening 4-5 days ago.

Much appreciate the help!

 

Seems like it might be Windows, or a hardware issue

Sent from my OnePlus 1 using a coconut

 

Check your windows update log and see if there were any updates around the time you first noticed it. More than likely it's a Windows driver conflict issue, but with the fact you just re-pasted something could be wrong with your hardware.

Have you moved the laptop around a lot after the re-paste? Something may have shaken loose.

Did you use enough thermal paste that it may have squeezed out onto the socket or components surrounding the chips? Usually that's not an issue, but IC Diamond has had issues in the past with bad batches. It's not likely, but it's a possibility.

 

Probably software related, then. How far back did you roll your NVidia driver to? You might want to try another version via the included "Clean installation" option the installer has (and opt out of GeForce Experience).

Since you mentioned a repaste, what are your temps looking like now?

 

Does that P751-DM2 also have the MSHybrid / Discrete switch option?

I've a P670HS-G and experienced basically the same thing randomly in games only when in Discrete mode, doesn't happen in MSHybrid.
I'm always running on an external monitor so it's always running on the GTX 1070 either way.

 

No, everything is routed to the NVidia GPU (Discrete). There's no Optimus/switchable graphics option.

 

Check the event log. When you get those spikes they are likely critical errors. The explanation might point to the cause.

 

So, you're running W10... Open Settings, go into the Gaming section, then disable all of the options for everything under Game bar, Game DVR and Game Mode menus, reboot and see if anything improves. Also, what option to you have set in NVIDIA Control Panel for the Power management mode? If it is set to Optimal or Adaptive, try changing it to Maximum performance.

Checking that for possible clues is also a good idea.

If you want to make it easier to look for errors in the event viewer, download the zip file. Extract and run both batch file as Administrator (right-click for that option). That will wipe out past logged events and let you start with a clean slate. Run your game(s) and see what new errors show up in the event viewer.

 

I had all that on..I keep finding out about new settings hidden away by MS that chips a little away at performance for things that only cater to the "average web user" vs. gamer/power user. Wish they had a setting you could click that would immediately go to just the bare OS. Thanks... really need to just sit down and tweak this system.

 

I would not mind if there really was a full game mode that re-started the system in a bare form

 

MSCONFIG gives you some degree of control over trimming the fat. I guess it depends on how bare you want it.

 

Hey everyone thank you for suggestions! So far I decided to reinstal the Windows from scratch and wipe the hard drives clean but as I mentioned I added a new one with the repaste (Samsung 960 EVO 500GB) and as the old, system disk took like 15 minutes to wipe completely the fresh one started fast but around 1/3 just slowed down dramaticaly and showed an hour more to go.

After the clean instal we ran the HD tune software to check the disks for errors and this is what happened:

With both disks in the we tested the old one first

Then the new Evo 960

Now after the new disk was working in a test we ran the old one once more

And seeing as there is obviously something wrong with the readings we removed the new SSD and ran the system drive without the other one in the laptop.

As you can see the old drive works very consistent without the new one.

We've done it only yesterday so I haven't been able to play too much since but the problems stopped. Seems the new drive could have been the issue.

As there was a question about the temps they hang around 60's-70's.. It sometimes spikes to like 77 or so on the CPU but the average temps show more around 69-71 and I pay attention to reset the clocks just as I enter the game to have accurate result and the gpu peaks at 71-72. Nothing ever breaks 80.

What do you guys think?

 

Looks like the new drive is messed up. Or, maybe it is overheating severely and cannot function correctly. I see the temperature reading is not working with HDTune. What does HWiNFO64 show for the drive temperature?

 

Try another app to double check the results too.

 

The temperatures are really weird and so are the readings right now too. For somre reason HWinfo shows 2 temps on the drive and HWmonitor only one...

Take a look:

 

Yep, different software may or may not be able to pick up all the sensors and it may give ghost results. That's the down side of third party software. If you have the paid version of either you could speak to the developer to confirm.

 

Yes, the second temperature is more than likely the controller temperature; most laptops don't have good enough airflow to handle that. Even the airflow king P870xMx line still gets my PM961 to 88c if I run a CrystalDiskMark test on the controller. The actual drive temp doesn't pass 56c though.

As to the performance issues, I wonder if it's something to do with the dual-NVMe setup when running both?

 

Yes I am wondering that as well but why then would the laptop have 2 dedicated nvme slots? Are there any ways to help with heat of the PCIE drives on a laptop?

Actually I see the problem now with the temps... because the auto fans on the CCC respond only to the CPU/GPU heating up when I perform the SSD tests the fans are almost or completely idle and it heats up.

 

You misunderstand. NVMe uses a PCI/e x4 connection, but there is only a maximum of 4 lanes from the chipset to the CPU. Two drives will conflict with total bandwidth from the chipset to the CPU, no matter what you so. This is by design and why I find NVMe so pointless.

I am not saying that this is 100% your problem when benching the drive, mind. It is just a theory assuming the 960 EVO is working fine. Try testing it in a system with no other NVMe drives, and if it works fine, try swapping your "working" NVMe drive to the other slot to see if that slot is particularly bad on airflow or something

Sent from my OnePlus 1 using a coconut

 

Ok... I get it now. Man I wish I researched the case deeper than I did. I read some things but this is new info for me. Should have just gone for a 850 EVO in the SATA slot -.-

 

Yup!

Sent from my OnePlus 1 using a coconut

 

If the controller or the NAND chips are hitting 103°C as shown by HWiNFO64 that would explain your system becoming sluggish and the erratic read/write performance after adding an NVMe SSD to the mix. If I am remembering correctly, @Georgel has similar problems with his 960 Pro NVMe temps being crazy high in his P775DM3. Some Alienware systems have similar issues. If you have enough space for it, add a M.2 cooler to the NVMe drive and that might take care of your problem. If there is no space for a cooler and air flow is poor, there may not be a solution other than using an M.2 SATA SSD instead of NVMe.

If you would like to know what the two temperature readings on the same drive correspond with, perhap the author (Martin) can answer. You can post a thread to inquire here: https://www.hwinfo.com/forum/Forum-General-Discussion

 

Now while this makes sense, the issue he also showed was that having both the 960 EVO and the un-named SM or PM samsung drive in the system together hinders the performance of the un-named Samsung drive. This shouldn't theoretically happen, and if the 960 EVO is not being used, it should also be cool and not affecting the other SSD.

Hence why I suggested testing it in another system and that it might be a dual-NVMe shot doing the issue, but I can't test this personally.

 

Quick update on the issue. I mean yes the drive controller temp seems crazy and unacceptable but the laptop was running flawlessly since December 2016 so I don't really understand.

It is worth adding that now as I have removed the 960EVO from the laptop and reinstaled a clean Windows the issues stopped.. for a day and I just now encountered them again.

Seems it's not the SSD in the end. I might have to send it back to the shop as it still has valid warranty.

Thank you all for the help and great insight. I will look into the SSD temps and all but in the end the system was fine and stable until now and now even using just the usual old drive it happened like twice during a short game of pubg and I know this one is pretty buggy but I also thought that it can be just the buggy title the first time before it started happening on all games.