7 hours ago, Mace said:

Also, I would test with an empty Community folder and use a default airplane.  Also make sure nothing is being loaded in the exe.xml.  Goal is to test with a completely stripped-down sim.

This is crucial, I had stutters and discovered a few add-ons where the guilty ones. One was a mod for taxi, landing and center lights, it made them 3D and introduced lots of stutters. Another one was a tree height mod, all of them free or payware will give you performance issues and stutters (see martinboehme's post in this thread

). If you're using one of these two, remove them from comm folder and you'll have no more stutters, that's my experience anyways. Hope it helped

15 hours ago, alr5100 said:

I have it set to all 3D VCache Cores in Process lasso 1-15, and all other apps on 16-32. I looked at XBox Game Bar and it shows 0%

 

Do you happen to know where in the Game Bar it can set to a specific core?

1) Try setting MSFS to only core 1 (not core 0!) in process lasso. Set affinity to high for flightsim.exe. Leave all the other affinities at normal. (Windows operating system will always mainly use core 0 and that can't be changed).

 

1 minute ago, SN737 said:

This is crucial, I had stutters and discovered a few add-ons where the guilty ones. One was a mod for taxi, landing and center lights, it made them 3D and introduced lots of stutters. Another one was a tree height mod, all of them free or payware will give you performance issues and stutters (see martinboehme's post in this thread

). If you're using one of these two, remove them from comm folder and you'll have no more stutters, that's my experience anyways. Hope it helped

Thanks I was thinking of buying that tree height mod, sad to hear it caused performance issues. 

1 hour ago, Sethos said:

Well, assuming we're talking about the actual framerate limiter; none of them as async introduces stutter and latency, where as the others just introduce massive latency, in excess of 100ms with FG. Even CapFrameX advises against using it;

https://twitter.com/CapFrameX/status/1620511329416265728

Scanline sync I don't know, assume that's not an actual cap.

Either of the Edge Syncs deliver this, and no other limiting method does, with the testing I did ~11/2022.  This equates to unparalleled smooth animation, and the frame rate of 35 was done w/ my prior 9900K fully maxed settings.  This holds up with DLSS FG as well, though this was w/o w/ the prior 3080Ti:

 

2 minutes ago, Noel said:

Either of the Edge Syncs deliver this, and no other limiting method does, with the testing I did ~11/2022.  This equates to unparalleled smooth animation, and the frame rate of 35 was done w/ my prior 9900K fully maxed settings.  This holds up with DLSS FG as well, though this was w/o w/ the prior 3080Ti:

 

That is strange I get massive latency with Edge Sync, and Frame limiting through RTSS, but no latency if I cap frames in NVCP. I have FG on. 

17 hours ago, F1Fan said:

I have a similar system PC that I upgraded a few months back and was having the same until I saw a few videos explaining how to get a stable frame rate.  One video that help was Capt. Pieit and another was from  2020 fs'ers on Youtube....check it out here if you haven't seen it..It is long but worth the watch.

 

I believe this may have been the answer I was searching for. I got the closest I possibly could to balancing the CPU latency with the GPU latency and it seems much smoother. I will be doing a couple VATSIM flight this weekend to make sure this solves the issue I am facing. It seems like a lot of people in the comments on the video with the combination of a 4090 and 7950X3D had the same issue I was experiencing, and I can already notice quite a large difference by employing the methods outlined in the video. Sadly I can only really strike the balance by bringing my TLOD to 50 and below, but I prefer to not have stutters...I had to bring my render scaling to 130-140 to match the CPU latency on the GPU and all GPU settings to ultra. 

1 minute ago, Noel said:

Either of the Edge Syncs deliver this, and no other limiting method does, with the testing I did ~11/2022.  This equates to unparalleled smooth animation, and the frame rate of 35 was done w/ my prior 9900K fully maxed settings.  This holds up with DLSS FG as well, though this was w/o w/ the prior 3080Ti:

 

I don't care about non-FG data, as that's not what we are talking about. I question the validity of the actual output captured by CapFrameX when the developers themselves state it's not compatible and it's quite universally known across every tech forum that RTSS simply isn't working correctly with Nvidia's Frame Generation. Open any technical overlay and look at the current frame latency with any of the edge syncs, even async, you'll see it's close to or exceeds 100ms. I mean, you don't even need to see an overlay, you can feel it right away that you have massive added latency, as just moving the mouse feels like it's moving through molasses

2 hours ago, tfm said:

Also have you tried without the NCP framerate limiter?

Yes I just split tested, it seems to do better with the NCP frame limiter for my system. I cap it 2-3 frames below the hz of the monitor. 

2 hours ago, ieydis said:

My suggestion - get rid of Process Lasso.  This is really not needed in MSFS and for me it caused more stuttering.  I also highly recommend using the AutoFPS tool if you haven't tried that yet.  It will automatically lower your LODs on approach and improve performance/stutters.

Ilya 

I just did some split testing seems like lasso the way I have it set up on my system brought the latency down on the main thread down about 1ms and gave some more stability. I am going to need to use AutoLOD because the most I can squeeze out of my system in dence airport scenery and traffic is about 50 TLOD. 

17 minutes ago, Sethos said:

I don't care about non-FG data, as that's not what we are talking about. I question the validity of the actual output captured by CapFrameX when the developers themselves state it's not compatible and it's quite universally known across every tech forum that RTSS simply isn't working correctly with Nvidia's Frame Generation. Open any technical overlay and look at the current frame latency with any of the edge syncs, even async, you'll see it's close to or exceeds 100ms. I mean, you don't even need to see an overlay, you can feel it right away that you have massive added latency, as just moving the mouse feels like it's moving through molasses

Well, not true here with or without DLSS FG on.  I leave it off because of visual artifacts not worth the additional frames.  A few others tried this here and were duly struck by the subjective smoothness of animation, including Cpt_Piett and a few others willing to try it.  No such thing as massive latency here can't relate in the slightest and none of the people mentioned ever reported anything about input latency. 

I know this is a properly banalt but CPU throttling possible ? I noticed is offcause on my system for a couple of days ago and now all is good again even with AIG

Michael Moe

1 minute ago, Michael Moe said:

I know this is a properly banalt but CPU throttling possible ? I noticed is offcause on my system for a couple of days ago and now all is good again even with AIG

Michael Moe

I'm not a hardware expert but I would think you would want to increase the CPU to get the latency lower. How were you able to throttle your CPU to bring the latency down?

6 minutes ago, Noel said:

Well, not true here with or without DLSS FG on.  I leave it off because of visual artifacts not worth the additional frames.  A few others tried this here and were duly struck by the subjective smoothness of animation, including Cpt_Piett and a few others willing to try it.  No such thing as massive latency here can't relate in the slightest and none of the people mentioned ever reported anything about input latency. 

Well, it would be the easiest thing to confirm with an overlay that captures the direct frame latency, Nvidia's own works, whilst also showing something confirming FG to be active. OP just confirmed exactly what I was saying mere 4 minutes before I posted it.

Just now, alr5100 said:

I'm not a hardware expert but I would think you would want to increase the CPU to get the latency lower. How were you able to throttle your CPU to bring the latency down?

No this was meant only  as a OP question to monitor CPU temps during the events of stutters. 

Michael Moe

Just now, Sethos said:

Well, it would be the easiest thing to confirm with an overlay that captures the direct frame latency, Nvidia's own works, whilst also showing something confirming FG to be active. OP just confirmed exactly what I was saying mere 4 minutes before I posted it.

Then something else is amiss on both of your systems as there is no subjective latency whatsoever.  RTSS async is wretched, and no other locking schemes produced the subjective impression w/ edge sync does and that's not just my take.   This 'as just moving the mouse feels like it's moving through molasses' is just flat out laughable.

Not sure what's different for you and the OP, but none of that happens here, with or without FG.  Here's another anecdote from someone who tried Edge Sync for the first time and there are others, using FG.  I wonder what's different for you and the OP...

 

>>Hey Noel,

Hope all is well and that you are enjoying your new 4090....

RTSS

FPS - 80

Front Edge Sync

I just departed Liverpool airport in the Fenix with FSLTL and ran CapFrameX for 15 minutes, Overall I am pretty pleased. The smoothness is beyond exceptional locked at 80FPS.

One thing I did stumble on is that when I enabled the NVIDIA filters it messed with my FTV so I turned them back off

•  
• Quote