I am testing for like 3 days now. I setup a flight and let the KI do the flight in order to get comparable results.
My system normally is undervolted and I run the xmp2 profile. What I found out now is no matter if I run with my bios settings or with the optimized default settings every first flight crashes right after few minutes. Then after restarting g the flightsim it gets better and after like the third or fourth crash there were no crashes for at least 3h.
Every other software, even benchmarking software (3d mark, conebench r23) runs fine with slightly above average results.
Typically, trying t access memory 0xfffffffffffffff (-1) is cause by the code expecting a memory pointer, and not detecting when it gets sent an error code instead ) ie -1
Null, (or -1) pointer errors.
If the cause of the inability to get a valid pointer cannot be found (ie Maybe a bug in a non Asobo created library ), then at least the Asobo code should be testing for Error Code returns and taking the necessary action.
It does this many case, and maybe Asobo are adding more of these exception detection throughout their code, but maybe they have not found all the occurrences when this may happen ?
usually not really “many”, of course depends a bit on the difference base vs xmp
But also remember: most of the RAM have more than one xmp profile stored in SDP and users can try another, bit slowers, xmp-profile. In special if users use 4 sticks, instead 2 sticks this might often be the solution ( also in special ddr5 ).
This is often due to very high vram usage in dx12. Microsoft needs to optimize dx12, they said they fixed the high vram usage, but unfortunately it’s still quite high!
In dx12 msfs uses 9.76 off 22 gb. That should not be too high.
I have been testing dx11 and dx12 mode now. How comes that Nvidia overlay is showing the doubled framerate in dx 12 that is shown in msfs built in fps window
So for the last 4 days I did some testing and wanted to share my results with you:
DAY1
DAY2
DAY3
Day4
3 times same test
4 times same test
2 times same test
2 times same test
Run
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Flight-Duration
5 minutes
1h
3h
10 minutes
2h
3 minutes
30 minutes
2h
2h
3 minutes
3,5h
Crash on World Map
35 minutes
0
2h
1,5h
1h
1,5h
0
Abort by
crash (memory)
crash (memory)
User
crash (unknown software exception)
User
crash (memory)
crash (memory)
User
User
crash (id1000; no error message)
User
crash (id1000; no error message)
crash (id1000; no error message)
crash (id1000; no error message)
User
User
User
User
crash (id1000; no error message)
Simconnect installed
yes
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
XMP II
yes
yes
no
yes
yes
yes
yes
yes
yes
yes
yes
no
no
yes @5600mhz
no
no
no
no
yes
XMP I
no
no
no
no
no
no
no
no
no
no
no
yes
yes
no
no
no
no
no
no
Undervolt
-0,068
-0,068
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
Hapticore installed
yes
yes
yes
yes
yes
yes
yes
no
no
no
no
no
no
no
no
no
no
yes
yes
Buttkicker plugged in
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
GAME REPAIRED
DX12
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
no
DX11
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
yes
optimized defaults
optimized defaults
optimized defaults
ram @4800
ram @4800
ram @6000
ram @6000
ram @6000
Conclusion: No matter what I did - only chance to get the flight simulator to run stable was to disable the xmp profiles.
It is really strange as there are no benchmarks or other games that show problems with my xmp or undervolting settings - the opposite is true: With my xmp II and undervolting settings I get better temps and better frames. Seems that msfs is kind of moody here
What’s about that issue? this is critical, after I update my graphic card to an RTX now using DX12 and hight settings, I have it each flight so I can not use the sim, Microsoft developers are working on that for the next updates?
Do a CLEAN install of graphics drivers and make sure you also update mainboard and chipset drivers (if available), run windows update too. Also read through this thread … particularly if yours is a factory overclocked gpu.
Today I did a memory test with MemTest86. In the first run I encountered 1 error. (Bios @default but memory @ 6000mhz which is the sticks native clock) Then I activated xmp 1 with 5600 mhz memory clock which is the max that Intel guarantees the i13900k running with. MemTest86 took about 2 hours and 45 minutes and all 4 passes hat no errors.
Now I tried to run the msfs with these settings and for now it is running for 1h and 15 minutes without any issues.
Based on my experience it should have been crashed 20 minutes ago
Edit 1: In flight I now switched to La Guardia Airport as I was somewhere in Russia before and it worked. I then switched the plane to an airbus and it still works. No crash so far after 90 minutes.
Edit 2: In the New York area I switched through every plane inflight and really tried hard to crash the sim. I then ported to EDNY Friedrichshafen und Germany and still no crash. For like 2 hours now I really tried hard to crash the sim but fortunately without success.
and together with a z790… Also dont forget: there is a difference in amount of installed sticks ( 2 vs 4 ). There is a reason why special overclocker-mainboard have often only two RAM slots.
I didn’t want to flex, I just wanted to share my findings in order to help.
Yes I do have a Z790 board but I also had severe problems with crashes and what I did today seemed to do the trick for me.
I am just saying that it is worth a try to go for xmp but within the CPUs designated range.
yep, thanks for that. It match exactly with what we try to mention since half year ( where we had similar cases, also with ddr5 ). I only wanted too add addiitional facts, in special relevant for ddr5, so users can may be check.
Its also that we not can repeat often enough, that XMP is OC… still in year 2023. As view of the mainboard and cpu is 5600 with a z790 and a 13900k now within the normal specs ( its more like a xmp profile, because former was 4800 the max ). But 6000, in special with 4 sticks, can still be a problem.
You should verify the voltage on your memory controller when running at 6000 Mhz. When running beyond 5600 Mhz for Raptor Lake, you’ll likely need to bump up IMC (or CPU VDDQ) voltage. Most Z790 motherboards should scale IMC voltage when enabling XMP and IMC voltage set on auto. Usually running at or beyond DDR5 6000 requires around 1.4v to the IMC, but this can be lower or higher depending on chip quality. For example, my 13900K can run DDR5 6400 with an IMC voltage at 1.42v.
It’s also completely possible your RAM isn’t stable at 6000 Mhz no matter what voltage you pump in the RAM or IMC. If this is the case, you might want to consider RMA. You shouldn’t have any memory errors when running XMP.
may be … that may still not work… as example from the manual of a “asrock Z790 Taichi Carrara” ( just an example ) depends on memory-configurtion:
1DPC 1R Up to 7400+ MHz (OC), 4800 MHz Natively.
1DPC 2R Up to 6000+ MHz (OC), 4400 MHz Natively.
2DPC 1R Up to 6000+ MHz (OC), 4000 MHz Natively.
2DPC 2R Up to 4800+ MHz (OC), 3600 MHz Natively.
