v0.5.4.0-exp Dynamic FG with RTSS Trial

[ResetXPDR]

Background: https://www.avsim.com/forums/topic/642527-msfs-auto-fps-app/page/301/#findComment-5815994

MSFS Compatibility: 1.39.12.0, 1.8.14.0.

Release notes for 0.5.4.0-exp

Exp1:

• Removed restriction of being able to use RTSS as the FPS source for Dynamic FG, thus allowing the Dynamic FG Target FPS to set as a post-FG value.
• Added calculation of the current frame generation multiplier when Dynamic FG is enabled and RTSS is used as the FPS source.
• Added display of the current frame generation multiplier on the app status line when Dynamic FG is enabled and RTSS is used as the FPS source.
• Expanded logic for determining the FPS colour to display when comparing target FPS with current FPS to cover when Dynamic FG, MFG and Manual FG are inactive, including when RTSS is used as the FPS source.
• Disabled all TLOD changes when Dynamic FG is enabled and RTSS is used as the FPS source, since an appropriate pre-FG target FPS cannot be reliably determined in this specific configuration.
• Updated Target FPS Type and value UI control tooltips to reflect the new Dynamic FG and RTSS FPS source configuration.
• Updated logging to include the current frame generation multiplier when Dynamic FG is enabled and RTSS is used as the FPS source.
• Fixed issue where the intial spike reduction event was not being logged.

---

This discussion was created from the release v0.5.4.0-exp Dynamic FG with RTSS Trial.

[MrJigs7]

Hi. I got chance to update to this new test version and load in MSFS at the gate whilst playing around with the auto fps settings to trigger some lod changes, everything seemed to work as expected. I'm using Dynamic FG in sim and RTSS as source FPS in AutoFPS. Log is attached, I can post full flight log on Monday as well. 1 very minor point, the mouse over tool tip for the current status in AutoFPS still shows "NFR" not sure if thats intended.
MSFS_AutoFPS20260822.log

[MrJigs7]

Yep I'm mrlip over at avsim. I updated to the new version and again just loaded in at gate, attached is the log.

Regarding my multiplier always being 6, I have vsync enabled at driver level and I believe there is a known issue (bug?) with DFG that when vsync is enabled it will always set the highest multiplier. Nvidia acknowledged it and I believe they plan to add full vsync support later. Source - https://www.pcguide.com/news/nvidia-confirms-dynamic-multi-frame-gen-doesnt-work-with-v-sync-as-users-report-compatibility-issues/

Monitor is 240HZ

MSFS_AutoFPS20260823.log

[ResetXPDR]

Thanks for the new data. As expected, the log now shows both the pre and post FG FPS and for the most part is always 6, most likely for the vsync issue reason you provided the link for. Have you tried dynamic FG without having vsync enabled, or is screen tearing unacceptable when you do?

Data analysis aside, is the app now behaving how you thought it should be behaving now that you can set the target FPS in post-FG FPS? It will probably take you doing an actual flight to answer this properly, so please do so after you have done that - no rush at all.

BTW, an interesting thing came up from your log to do with with spike detection. I know you are not using spike protection, but it will still detect spikes when RTSS is running and yours is showing a lot of false positives because of the high FG multiplier and the median frame time method I have been using for this. In short, the median frame time is very low because most frames are frame generated and that makes the much larger real frame frame time look like a spike in comparison, when it is really not. As such, I will have to revise where this reference frame time comes from. This will not impact your use of the app but this data will help me show whether I am on the right track with any fixes.

[MrJigs7]

New log attached of full flight, was a long haul so apologies for the length!
MSFS_AutoFPS20260824.log

I disabled vsync at driver level on this flight so you see some changes to the multiplier. Unfortunately it was nearly all at night so tlod was only tlod base. Having said that, the app is behaving how I expect it to when using RTSS as source and targeting post FG fps that are generated using a dynamic multiplier.

[ResetXPDR]

Wow, you are not wrong about a long flight - 10 hours+!

Nonetheless, there is no such thing as too much data when testing and yours had some interesting bits with what we weren't specifically testing that showed me my spike detection code was still not robust enough to handle all cases of the 6X FG data your scenario threw at it. I have have now tightened it up so that it behaves correctly with everything it didn't and theoretically everything else should still behave, but software being software will prove me wrong I am sure 😆

Regarding Dynamic FG with RTSS that is the main focus of this release, as you say everything looks good and there were some occasions when something other than 6X was present, so that bit works too. You may have also seen some out of range values like 7X briefly immediately after switching back to MSFS having the focus. I have now capped the multiplier range to have to lay between 1X and 6X, which is the max range of any type on nVidia FG, so values like 7X will no longer appear.

I have tweaked 0.5.4.0 with aforementioned improvements, plus a couple more. Since the concept of setting a post-FG target FPS when using Dynamic FG and RTSS as the FPS source seems proven now, I have advanced the app to test status. The only impact this will have on you is that now you will have to explicitly check Log+ for the extra logging, as it is only forced in experimental builds, so make sure you do that for any further logs you provide me.

Thanks for keeping up with me in providing timely logs for the changes I have made in response to previous logs. This allows for much quicker development and, more importantly, correct implementation of requested features. Your logs have also allowed me to fix spike detection with high FG multiplier RTSS frame time streams, something I didn't even know was broken until I saw your data, so thanks for that too!

[MrJigs7]

great thanks very much! I've updated to 0.5.4.0-test1 version and will let you know if I see anything unexpected or have any issues

[ResetXPDR]

FYI, 0.5.4.0-exp2 has been released with the following changes:

Exp2:

• Changed Log+ to be force enabled in experimental test builds or when ExperimentalSettings is enabled.
• Added logging of the raw MSFS‑sourced FPS and the calculated dynamic FG multiplier to the performance‑status log line when Log+ is enabled and in‑game Dynamic FG is active with RTSS as the FPS source.
• Removed status‑line tooltip because its explanation of NFR is no longer necessary given how long NFR has been part of the app.

[ResetXPDR]

FYI, 0.5.4.0-exp3 has been released with the following changes:

• Changed RTSS frame time median calculation from the median of the entire data set to automatically detecting active frame generation and calculating the median of only real frames, cleanly separating out generated frames and isolated stutter spikes.

[ResetXPDR]

FYI, 0.5.4.0-test1 has been released with the following changes:

• Refined real frame time detection using log-transformed clustering to prevent severe freeze spikes from corrupting frame generation detection and causing false fallbacks to generated frame rates.
• Updated periodic spike detection early‑exit logic to require <5s elapsed AND <90% buffer fullness, preventing premature skips at very high FPS where the buffer fills rapidly.
• Capped auto‑derived dynamic FG multipliers to the 1–6 range for both manual‑FG and FG‑mode paths.
• Disabled failure logging for individual CPU core performance counters, as the subsequent 'unable to add any counters' log message is sufficient.
• Changed auto-update process to now allow updating from a release version to an experimental version, provided that Exp+ is the chosen update channel.

[jerryyyyy]

Hi, saw you had a new version. This is actually two flights, long followed by a short graphics only VR test.

New MSFS_AutoFPS20260825.log

PS here is the result of USB3 VD on the Q3... > 5000 Mbps....

[ResetXPDR]

You got your first ever periodic spike detection when you were in 2D 🥳 and spike protection kicked in as a result, which is good:

2026-08-26 12:52:36.118 [DBG] [ ServiceController:DetectPeriodic ] PeriodicDetect medianUs=43528 averageUs=50923 minUs=41787 maxUs=311385 sampleCount=197 multiplier=2.0 thresholdUs=87056 spikeCount=6 
2026-08-26 12:52:36.118 [DBG] [ ServiceController:DetectPeriodic ] SortedIntervals count=9 min=1136938 max=2516392 
2026-08-26 12:52:36.118 [DBG] [ ServiceController:DetectPeriodic ] LargestCluster size=4 first=1136938 last=1165430 periodicityUs=1164326 
2026-08-26 12:52:36.118 [DBG] [ ServiceController:DetectPeriodic ] PeriodicCheck earliestIdx=0 earliestElapsedUs=188107 clusterPeriodUs=1164326 clusterMaxIntervalUs=1165430 
2026-08-26 12:52:36.118 [DBG] [ ServiceController:DetectPeriodic ] FrameTimesCSV: 188107,41886,41805,41845,41929,41787,42084,41960,42038,41958,42115,42195,42117,42238,42206,42142,42552,42424,42396,42342,42522,42674,76017,204198,42312,42312,42359,42346,42540,42422,42527,42585,42552,42509,42946,42712,42793,42835,42659,43031,42861,42970,43018,43022,70989,240026,42855,42801,42597,42567,43224,43107,42986,42875,43122,43021,43091,43132,43213,43186,42904,43200,43478,43294,43380,44241,178407,311385,43158,42900,43133,42720,43357,43145,43201,43070,43207,43190,43351,43285,43276,43372,43450,43322,43527,43563,43516,43598,77480,194117,43193,43512,43147,45268,43325,43522,43203,43658,43444,43599,43528,43577,43496,43763,43626,43705,43725,43694,100391,278276,43427,43304,43372,43078,43704,43527,43458,43432,43415,43815,43498,43622,43516,43606,43809,43508,43760,43630,43888,43761,43836,43375,44221,43838,43811,43815,43784,44147,43934,43871,44189,43918,43632,43982,44490,44184,44011,43835,44402,44098,44007,44231,44063,43834,44492,44244,44160,44490,44203,44061,44472,43950,44561,44470,44342,44326,44308,44285,44159,44637,44307,44325,44613,44652,44617,44343,44583,44516,44436,44539,45226,47023,44431,44492,44617,44519,44318,44631,44804,46792,44279,45703,44564,45231,54021,47266,48675 
2026-08-26 12:52:36.118 [DBG] [ ServiceController:DetectPeriodic ] 4 periodic spikes, 1.2s apart 
2026-08-26 12:52:36.118 [INF] [ LODController:RunTick            ] Automatic periodic spike TLOD reduction activated @ TLOD 640 

After it kicked in, your spikes cleared up - also good. You sat for the rest of that 2D session on the ground, so recovery never had a chance to occur, with TLOD capped to 580 but your actual TLOD after that was always under that anyway. When you switched to VR, no more spiking was detected.

[jerryyyyy]

That P51 is a weird aircraft. I was not getting anything erroneous with the others. I dual task and sometimes have to pause the simulator. Tomorrow a long flight in the Kodiak, which is very reliable.

I also upgraded to the latest NVIDIA driver. There have also been some changes in RayTracing (which I have turned off) and DLSS.

[jerryyyyy]

This is another long flight in the KitFox and seemed to work well. Think the plane has quite the influence.

New3 MSFS_AutoFPS20260827.log

[jerryyyyy]

Your new test version(s).

New Test MSFS_AutoFPS20260829.log

[jerryyyyy]

Cub now in Alaska. Mostly VR. Seemed to perform well.
MSFS_AutoFPS20260830.log