Getting CTD recently more often

[andy1252 3,851]

1 hour ago, abrams_tank said:

In software development, the first thing they teach you when you have a bug, is to try to isolate the bug. To isolate a bug, you need to simplify the issue as much as possible and eliminate all other possible factors, while trying to reproduce the bug.

Once you isolate the issue, you can rectify it and then gradually add back your add-ons.

All very well and good, but all my issues started around the time of SU9. I would dearly love to be able to remove that to test that particular factor. But we don't have that option.

I haven't bothered to even turn the simpc on for a while now. I'll just wait and put my hopes in SU 10.

[Reader 2,021]

1 hour ago, Cpt_Piett said:

IIf you purchased a PC (i.e. you didn't build it yourself) you may have XMP enabled in the BIOS as well as a factory OC'd GPU. This can cause stability issues with MSFS. 

I suspect that very few PC suppliers, let alone home builders, would have a niche flight simulator that is intolerant of some overclocks but not others at the forefront of their minds when setting up the device.
If I were Asobo, I would be trying to make my niche simulator work across the board, instead of only on some devices.
There is, after all, anecdotal evidence that in some cases, it doesn't even work consistently on an Xbox.
The difference seems to be that even on an Xbox, first-party but addon products developed by someone else and third-party addons are allowed.

Edited August 31, 2022 by Reader

[Bernard Ducret 790]

18 hours ago, Cpt_Piett said:

I remember reading that post of yours. This is an important point, and somewhat unfair to the end user. If you purchased a PC (i.e. you didn't build it yourself) you may have XMP enabled in the BIOS as well as a factory OC'd GPU. This can cause stability issues with MSFS. 

Another factor following the most recent Windows 11 update introduced a USB time control designed to put "unused" USB hardware into Standby automatically. If you have any USB item (and you probably have) some or even only one of them (if not designed to go on Standby) will create a hardware conflict and CTD. This can be avoided by disabling selective standby of USB devices on Windows 11, that generates many major problems and crashes.

I see that you have DDR5 RAM, make sure it is not exceeding 4800 MHz and is not OC (XML).

Good luck.

[Cpt_Piett 4,634]

@Bernard Ducret Thanks for the advice. I’m running the DDR5 at 5800 (XMP is 6000). Guess I could always reduce it to 4800, but I haven’t seen any other info re: that.

I think I’ve remembered to disable USB standby. For others that have not, it can be done Er på vei! is command prompt or control panel: 

https://www.windowsdispatch.com/disable-usb-selective-suspending-windows-setting/#:~:text=Look for the Generic USB,this device to save power.

[Bernard Ducret 790]

3 hours ago, Cpt_Piett said:

I haven’t seen any other info re: that.

Here is what a quick Google search gave me to support my suggestion above: Techreviewer (excerpts)

When using DDR5 memory, the Core i9 12900K CPU officially supports memory speeds up to 4800 MT/s. This maximum 

speed means that stock DDR5 performance will be maximized using DDR5-4800 memory.

I hope this helps you.

Edited September 1, 2022 by Bernard Ducret

[Maladoror 89]

20 minutes ago, Bernard Ducret said:

Here is what a quick Google search gave me to support my suggestion above: Techreviewer (excerpts)

When using DDR5 memory, the Core i9 12900K CPU officially supports memory speeds up to 4800 MT/s. This maximum 

speed means that stock DDR5 performance will be maximized using DDR5-4800 memory.

I hope this helps you.

Actually, from Intel:

If the motherboard has two memory slots total, then the maximum support is DDR5-4800 in any configuration.
If the motherboard has four memory slots total, then the maximum support is DDR5-4400 when two slots are filled with any memory.
If all four memory slots are filled, single rank memory will support up to DDR5-4000.
If all four memory slots are filled, dual-rank memory will support up to DDR5-3600.

[Alpine Scenery 1,609]

Having been in computers, programming, and PC's for years and years, and I used to always make fun of people that used to say "my computer is possessed"...

Well it finally happened, my only explanation is possession.

I can go days without an issue, then I can get BSOD's with my AMD 6900xt out of the blue 3-4x in a row which are likely triggered by MSFS leaking pointers onto the heap, and once the first BSOD happens, things get haywire for a few reboots. There is some type of persistent state issue being triggered by MSFS and the AMD card.

I am however forced to replace the AMD with an Nvidia 3080 just to be sure.

BTW, this is my 3rd (yes 3rd AMD), I have replaced the memory with Corsair, I am on a CyberPower 1500VA UPS, I have replaced my old 1000 watt power supply with an EVGA 1200 Watt Platinum. My motherboard is a very standard well known ASROCK B450-Pro with an AMD 5800x. Nothing connects the dots or makes sense, I can go days without a single error, then in one hour I might experience 7 crashes suddenly and out fo the blue. Oh I do get CTD's in Dev mode sometimes, but that is unrelated.

I love MSFS, it is the sim I have always dreamed of, but I swear they have some super nasty pointer errors that are cross-leaking into critical OS processes at times, it could even corrupt your Windows if your unlucky. 

That said, they mysteriously disappeared the past few days, so maybe it is witchcraft or possession, at this point it is possible.


 

Edited September 1, 2022 by Alpine Scenery

[fogboundturtle 2,708]

2 minutes ago, Alpine Scenery said:

Having been in computers, programming, and PC's for years and years, and I used to always make fun of people that used to say "my computer is possessed"...

Well it finally happened, my only explanation is possession.

I can go days without an issue, then I can get BSOD's with my AMD 6900xt out of the blue 3-4x in a row which are likely triggered by MSFS leaking pointers onto the heap, and once the first BSOD happens, things get haywire for a few reboots. There is some type of persistent state issue being triggered by MSFS and the AMD card.

BTW, this is my 3rd (yes 3rd AMD), I have replaced the memory with Corsair, I am on a CyberPower 1500VA UPS, I have replaced my old 1000 watt power supply with an EVGA 1200 Watt Platinum. My motherboard is a very standard well known ASROCK B450-Pro with an AMD 5800x. 

I love MSFS, it is the sim I have always dreamed of, but I swear they have some super nasty pointer errors that are cross-leaking into critical OS processes at times, it could even corrupt your Windows if your unlucky. 

That said, they mysteriously disappeared the past few days, so maybe it is witchcraft or possession, at this point it is possible.


 

have you ever think that maybe the AMD Drivers for your 6900XT is not very good ? and BSOD are 100% hardware related and not MSFS cause.  Come to the green side and get an Nvidia GPU.

[Alpine Scenery 1,609]

1 minute ago, fogboundturtle said:

have you ever think that maybe the AMD Drivers for your 6900XT is not very good ? and BSOD are 100% hardware related and not MSFS cause.  Come to the green side and get an Nvidia GPU.

Yah, that's why I'm exchanging it for the Nvidia just to make sure. I know the AMD drivers aren't perfect but I'm running the most stable driver.

It's just weird as heck that I can go days, even a week without an issue (and I'm a very very heavy SIM user), then out of the blue, one particular hour it feels like my PC is about to just stop working there are so many BSOD's.

Edited September 1, 2022 by Alpine Scenery

[wim123 761]

3 minutes ago, Alpine Scenery said:

Yah, that's why I'm exchanging it for the Nvidia just to make sure. I know the AMD drivers aren't perfect but I'm running the most stable driver.

It's just weird as heck that I can go days, even a week without an issue (and I'm a very very heavy SIM user), then out of the blue, one particular hour it feels like my PC is about to just stop working there are so many BSOD's.

a lot depends where you fly, certainly with 3rd party sceneries, maybe ai traffic, mp closeby..there are so many variables,

[Alpine Scenery 1,609]

2 minutes ago, wim123 said:

a lot depends where you fly, certainly with 3rd party sceneries, maybe ai traffic, mp closeby..there are so many variables,

True, but the errors seem to happen to me regardless, doesn't matter. They've even happened at my own airports, or when I'm testing a clean BLANK file build (as in running a blank package in Non-Dev mode), so even outside of development of a very remote airport in the middle of nowhere. So I know it's not really location specific for me, but it might be for some, but for me I kind of ruled that out.

Plus I have been flying the same places over and over the past 2 weeks, heck I don't even think I've left Colorado or Utah in 3 weeks except maybe twice.

Edited September 1, 2022 by Alpine Scenery

[Maxis 1,861]

3 minutes ago, fogboundturtle said:

have you ever think that maybe the AMD Drivers for your 6900XT is not very good ? and BSOD are 100% hardware related and not MSFS cause.  Come to the green side and get an Nvidia GPU.

That's not been my experience. ALL the drivers installed (and i always update the drivers)  since purchasing an AMD gpu have been solid in the sim .. and the the card is also heavily overclocked. No CTD on my end

[Alpine Scenery 1,609]

9 minutes ago, Maxis said:

That's not been my experience. ALL the drivers installed (and i always update the drivers)  since purchasing an AMD gpu have been solid in the sim .. and the the card is also heavily overclocked. No CTD on my end

It could be USB device related might explain some of it, but I have my doubts that it's my issue.

I think people that have issues on clean builds, it comes down to MB compatibility with video card driver, TDR interop with malloc and "IRQ like" type crud, even though IRQ's aren't the same there is still shared critical processes that can conflict just like old IRQ conflicts from the old days. This stuff gets complex, and AMD certainly isn't perfect in the TDR department with all hardware. So it's probably interoperability luck, there are so many variables in this stuff. The pointer errors I rarely but occasionally see from MSFS are most likely triggered by an error handler in code, so they aren't common triggers, that would explain most of it. What triggers the error handler is probably timing issues, so that is how server issues might make it more likely for some to get errors, as error handling code seems to be bugged out in MSFS.

Eliminating the AMD can help even if someone ELSE has perfect luck with an AMD, because it may be the AMD TDR or similar isn't timing out as well on their hardware causing issues.

BTW, my operating temps of the GPU average in the 50's and 60's, my voltage is conditioned and always 120.

Edited September 1, 2022 by Alpine Scenery

[Bob Scott 7,720]

I'm not at all convinced that MSFS has some mysterious special coding x-factor that will cause a properly overclocked PC to crash.

A couple of possible explanations come to mind:

When it comes to RAM, MSFS uses a lot of RAM, more than most other programs, so if you have an issue in the higher address regions that don't get used often (or at all) by other programs, then MSFS might find that lurking issue and go boom.  RAM modules are rated for specific speeds, timings, and voltage.  Often, due to manufacturing variations, the actual voltage applied at the pins of the DIMM slots may be lower than the Vdimm set in the BIOS.  Adding just a few hundredths of a volt in the BIOS setting can make the difference between RAM running stable or throwing the occasional error.  An overnight run of MemTest 86+ is really needed to effectively root out a memory overclock that's right on the ragged edge of stability or to locate a bad block in the largely unused upper reaches of the address space.  If your RAM and IMC can pass a 12-hour run of MemTest 86+, I am really sceptical that MSFS can do something that stresses the RAM even more.

Regarding CPU overclocks, lots of folks don't pay adequate attention to AVX/AVX2 when overclocking--code that uses those extended instruction sets seriously increases the amount of work being done in the silicon and, as a result, the amount of heat being dissipated and stability under load.  An overclock tested with a utility that doesn't also heavily stress the CPU with AVX workloads will not necessarily be stable if running a program heavy in AVX code.  Most BIOSes have an AVX offset that reduces the clock multiplier when AVX instructions are loaded--increasing that offset might help.  I prefer to overclock with a zero AVX offset and accept the overclock I can get with the clock mult dialed back a notch or two to allow for heavy AVX workload without resorting to jacking the CPU multiplier around.  I believe (but haven't done the analysis to know for sure) that MSFS, as a relatively modern code base, probably does include a lot of AVX code.

 

[Alpine Scenery 1,609]

Not overclocking anything here, though your points are valid for SOME people, they don't apply to me.

I'm running Corsair DDR-3600 using a 3000 and 3200 profile (tried both) which is known to be like the MOST stable possible memory config for my specific motherboard, it's even listed at the very top. I used other software which is more intensive than MSFS memory wise without issue (for instance, the Visual Studio debugger when debugging a leaky memory process, I've literally had it use 50 of 64 of my GB of memory with no stability issues). So the idea that MSFS is using far away memory that never gets used for me doesn't apply either.

Everything runs without issue except MSFS (every game), which pretty much tells me it's their error handling code that's bugged. That's why it comes and goes with rarity and unreproducible. I've reinstalled MSFS with completely brand new hardware with the same result, the only thing I haven't replaced is the motherboard, which is why this is almost for certain an malloc interop issue with hardware. My suspicion is that if I simplified my config enough (removed drives and as many USB devices as possible), that the TDR timings would have less interrupted chances of memory conflicts. That said, TDR isn't a USB issue directly, but anything to free up potential "memory holds".

Even rendering software like Substance Painter and Quixel require TDR changes in the registry with AMD cards, so it's not a surprise to me, they recommend 60, though I have mine at 30, default is like 2-5 (or something, I forget).

Edited September 1, 2022 by Alpine Scenery