XMP causing CTDs. Also new microstutters have appeared

[109Sqn 1,580]

I made a 6 or 7 minute flight from Leuchars to Dundee and the PC shut down around 1 minute before touchdown.🙄 I had NZXT CAM running, with just the PU and GPU temps displayed (no stuttering this time!), and they were sitting at 60°C and 80°C respectively right before the crash. Neither temp had gone over those values during the flight (or since starting the sim  5 minutes before the flight).

Could there possibly be a USB port problem (i.e. one of the peripherals plugged into that port 'spiking' or doing something else odd)? I'll try a flight with the Honeycomb Bravo* and Fulcrum yoke unplugged first, and fly with the Thrustmaster Airbus stick and TQ. If it still rashes, I'll try the opposite combination. If it comes down to it, I'll even try with just the Xbox controller.

 

* One thing I have noticed recently, is that the Bravo keeps wanting to light up the HDG button and switching back to it a second after I push the NAV button. I'm not sure if that's only with the A320 but that's all I've noticed it on so far. Possibly the culprit...?

[109Sqn 1,580]

Incidentally...

4 hours ago, 109Sqn said:

... WhoCrashed ... hasn't found a single crash log after any of the tests. I've only run each test once so far.

For some reason, Windows was set not to save crash dumps!🤔 And there page file on my C Drive! I hadn't disabled that manually since installing Win11. So I enabled them, and even increased the page file size to more than equal the 32GB RAM, as recommended by WhoCrashed. Then I ran the Crash Dump Test which worked and led to 2 dumps being created. But when the above flight shut down the PC... no dump files!😒

[109Sqn 1,580]

2 flights so far (Leuchars to Dundee each time). #1 completed just fine, no issues. #2, apart from being unusually directed to land on runway 09, completed as far as joining the taxiway upon landing but switched off the PC as I was about to ask ATC for a parking spot. I thought I was onto something but not yet.🙄

Time to switch to unplugging the Airbus controls and using the Fulcrum/Bravo combo...

 

Edit: Also, no dump files being created.😠 Still just the ones from the Whocrashed Crash Dump Test. I've manually deleted them to see if it makes any difference.

Edited February 26, 2022 by 109Sqn

[109Sqn 1,580]

I notice that Event Viewer is listing the following several times after each forced shutdown:

Quote

 

Event 2, volmgr

Failed to read KSR data - disk 5 sector 73 operation 0 status c0000225.

 

It's always Disk 5, with differing sectors (i.e. the same line as above with only the number in italics changing each time. Disk 5 under Disk Management is my new 2TBm.2 NVMe drive which only contains my add-ons, and since my Community folder is empty, nothing from that should be getting accessed anyway. System File Checker found no issues and Chkdsk found no errors in that drive. Now wondering if I should take that drive out of my mobo and see how things go after I've tested MSFS's behaviour with the other controllers.

[109Sqn 1,580]

The above volmgr errors may be the key to my PC shutdowns! I took the offending drive off the motherboard and I've managed a couple of multi-hour flights without a shutdown (one CTD but that was after leaving the AI flying the default A320 and then trying to switch from external view to internal via the yoke button. I've found that occasionally the sim doesn't like being left alone for hours before a button is pressed).

I'm just updated to SU8 from MS Store, then I'll be installing a couple of new Noctua NF-A12 120mm fans in the case roof before I update MSFS in-sim. I'll re-seat the drive at the same time as fitting the fans and see if that is the issue fixed. Otherwise, it may be that the drive needs RMAed. Hopefully I don't need to do that though...🤞

[109Sqn 1,580]

@NismoRR, @737_800, @rka, @iwebber, @bobcat999, @teletom, @Bob Scott, @Paul J, @Maxis, @martin-w, @Emerson67, 

I don't know who's still reading this saga so I've just tagged everyone who offered advice!😁

I'm not totally out of the woods but I think I can maybe see a field somewhere through the branches.🙂

Problems which appear fixed: CTDs and microstutters as originally posted about. I've had an issue with both but in a different context (see below) but they seem unconnected to the original problem. Volmgr showing a problem with my Add-ons M.2 SSD - no signs of "Failed to read KSR Data" since I reseated the drive on the motherboard. I'm assuming the drive wasn't fully in the slot and that was resulting in data-reading issues.

Problems which have appeared recently: One CTD, but that was on a flight of several hours when I'd left the AI in charge and tried to switch from external to cockpit view with the yoke button. The sim sometimes doesn't like being left unattended for several hours and to then have an action taken via a controller. Chalking that down to one-off grumpiness from the sim. I am still experiencing microstutters or very low FPS when I open Game Bar (whether via Win+G or the Gamepad. It means I can't record the flight with Game Bar as it will be a stuttering mess. The same stutter/fps drop occurs when there is any other type of overlay on the sim, such as a resource monitor (e.g. NZXT CAM) or even the windows volume display when I alter volume from the keyboard. I've seen nobody else complain of this, so it's a really odd one.🤔

Problems as yet 'unresolved': I am still to determined once and for all whether I can use XMP at all, and if it's worth putting the 3600MHz RAM back in. If I really can't use XMP with that RAM, I'm just as well to stick with the 3200. I won't even go back to that issue until I'm convinced everything else is resolved and SU8 has had time to 'bed in'.

Thanks to everyone who has offered assistance so far. This one's not yet done, but I think we're getting there...🤞👍

 

Edited March 1, 2022 by 109Sqn
typo

[bobcat999 8,489]

8 hours ago, 109Sqn said:

...

Problems as yet 'unresolved': I am still to determined once and for all whether I can use XMP at all, and if it's worth putting the 3600MHz RAM back in. If I really can't use XMP with that RAM, I'm just as well to stick with the 3200. I won't even go back to that issue until I'm convinced everything else is resolved and SU8 has had time to 'bed in'.

Thanks to everyone who has offered assistance so far. This one's not yet done, but I think we're getting there...🤞👍

Sorry to hear of the problems, but like you say, you are getting there!  Perseverance is the key like with most things.  Time consuming I know! 

I had RAM problems with my recent upgrade.  I was getting corruption, random blue screens, and windows violation errors.
After eliminating a few other items, I tested my RAM with the Memtest program built into the bios, and it showed failures within seconds.  (you should also have this in your Z690 bios).

I bought a good brand of 4400 DDR4, but it wouldn't run at that XMP speed. 
I later was told they only guarantee XMP at those speeds for 16GB matched kits! I had two of those kits to make 32GB, but they could quite do it, so I have had to turn my RAM down manually.  The non-XMP setting was about 2400 - a bit low.

I will say one thing though, even with my 12900k, RAM speeds over 3200 don't seem to make much difference.
While you are solving problems, I would leave your RAM dead stable, or it can throw you a few red herrings and you can end up chasing your tail for a bit, like I was initially. 
I have now settled for lower speeds but with slightly better timings on CAS etc.  Mine is now tested dead stable in Memtest for a few hours while I went out.

This same problem happened to me before many years ago, but I seemed to have forgotten my very expensive lesson at the time.  I was kicking myself when it was RAM speeds again!

My advice to anyone with a new build - don't believe the RAM speeds at 32 GB unless the manufacturer explicitly guarantees them.  Most are for 16GB matched kits only.

Edited March 1, 2022 by bobcat999

[109Sqn 1,580]

Latest progress report. Having tried the 3200MHz RAM at the mobo's stock speed (i.e. not XMP) I decided to try the 3600 at stock. At first, although the monitor got a signal, it went from the usual no display message to nothing - the PC wouldn't complete POST.🤔 I tried different combos of only 2 sticks in the different pairings of slots and it was only when I realised I must have set them aside matched (sticks AABB instead of installation ABAB) that I solved the problem! I also hadn't noticed till checking CPU-Z that the two pairs of sticks aren't entirely identical after all, despite being the same model of RAM. Although they have the same code number, one pair is single rank, the other pair is dual rank! Now, despite reading up on rank, I don't really understand it too well, but I know enough to make sure pairs are matched ABAB in the mobo slots, so that was for me the likely cause of the failed POST. Sure enough, after putting all four sticks in ABAB order, I got into Windows just fine (the stuttering in MSFS with any screen overlay such as the volume display or Game Bar still exists - I think that may be a GPU driver issue as I've yet to try the latest one, so I'm putting solving that on the backburner for now). 

Next, I tried XMP 2 but I got a failed POST message and had to hit F1 to reset the speed. Same with XMP 2. I tried XMP Auto (where the BIOS sets the XMP values) but when I exited to the BIOS main page, the XMP window showed "disabled", so it looks like it does a test before you even try rebooting (either that or XMP Auto doesn't work properly!). So, I decided to manually set the RAM to 3400MHz (no other changes) to see what would happen. So far, so good - I'm in Windows, typing this on my MSFS PC - and the next thing will be to see if MSFS behaves. If it does, I'll set 3600MHz manually just to see if that works better than XMP doing it, though I won't be surprised if that fails.

You'll note that all the way through this thread I have made no mention of having changed any other RAM settings other than a brief attempt with altering the voltage. That's because CAS etc is one area where I really don't know what I'm doing and I'm not even confident on creating a sensible workflow to do so, let alone knowing what to change and when. As things stand, if I can get MSFS to behave with manual 3400 (or even 3600 if I'm lucky), I'll probably leave it like that rather than strive for gains I may not even notice.

 

Finally for this post, I said I was putting the overlay stutter issue to one side, but I notice nobody has had anything to say about that. I take it no-one has experienced it or has any possible solutions? it never used to happen so it has me baffled. I did have my second display enabled so wondered if the move to Win11 had affected that, but disabling second display didn't stop the stutters. Again, this only affect MSFS - neither of my Forza games have the problem and they are GPU-intense.

Edited March 4, 2022 by 109Sqn
typo

[109Sqn 1,580]

Suffered a BSOD at 3400.😑KERNEL_THREAD_PRIORITY_FLOOR_VIOLATION whilst posting on the forum and before having a chance to try MSFS. I believe the RAM chips are seated properly but I ran a full Windows Memory diagnostic. It seemed to be at 21% forever though (after 2 hours) so I quit and will see if things are stable or not before I consider lowering the RAM below 3400MHz.

 

[G-RFRY 1,712]

1 hour ago, 109Sqn said:

Suffered a BSOD at 3400.😑KERNEL_THREAD_PRIORITY_FLOOR_VIOLATION whilst posting on the forum and before having a chance to try MSFS. I believe the RAM chips are seated properly but I ran a full Windows Memory diagnostic. It seemed to be at 21% forever though (after 2 hours) so I quit and will see if things are stable or not before I consider lowering the RAM below 3400MHz.

 

If I were you I would remove two ram sticks run dual channel on 2 then test, all memory and XMP are handled by the CPU and more sticks gives it more work as it has to control four memory slots.

[109Sqn 1,580]

On 3/1/2022 at 9:06 AM, bobcat999 said:

Memtest program built into the bios, and it showed failures within seconds.  (you should also have this in your Z690 bios).

 I think this might be another thing (in addition to the Q-Connector) which is missing from TUF Gaming mobos unless I've simply failed to find it. I may be stuck trying the standalone MemTest but I'd rather test within the BIOS if possible. Can you tell me where you find it on your ROG mobo?

8 minutes ago, G-RFRY said:

I would remove two ram sticks run dual channel on 2 then test

I did all of this previously but may end up having to go back and do it again. I also just had a second BSoD, this time the old "irql_not_less_or_equal".🙄 Starting to look like I'm pushing my luck with 3400MHz after all.

[109Sqn 1,580]

 

And another BSoD ("system thread exception not handled") as I tried to shut down the PC just now! This is happening shortly after I installed some DTS Audio optional updates. Wonder if it's connected?

[G-RFRY 1,712]

42 minutes ago, 109Sqn said:

I did all of this previously but may end up having to go back and do it again. I also just had a second BSoD, this time the old "irql_not_less_or_equal".🙄 Starting to look like I'm pushing my luck with 3400MHz

What are the ram sticks rated at are you changing the ram speeds are the ram stick matching pairs, I don`t recommend messing with ram let it ran at default speed, and your manual should show the dual two slots to use for just two stacks.

[109Sqn 1,580]

As in my original post, and 6 posts above this one, they are 3600MHz, 2 matching pairs (as opposed  to a set of 4). Problem as stated is that the BIOS wants to limit it to 2133MHz. Again as above,

8 hours ago, 109Sqn said:

I know enough to make sure pairs are matched ABAB in the mobo slots

 

[Maxis 1,878]

7 hours ago, 109Sqn said:

 I think this might be another thing (in addition to the Q-Connector) which is missing from TUF Gaming mobos unless I've simply failed to find it. I may be stuck trying the standalone MemTest but I'd rather test within the BIOS if possible. Can you tell me where you find it on your ROG mobo?

I did all of this previously but may end up having to go back and do it again. I also just had a second BSoD, this time the old "irql_not_less_or_equal".🙄 Starting to look like I'm pushing my luck with 3400MHz after all.

IRQ not less than equal ..

yep .. pushing it

3200 sounds like what you are stuck at.. but i wouldnt despair because once you get it stable for a few weeks then if you want to we can then move to timings