flyinion

huge FPS difference full screen vs windowed, very confused

Recommended Posts

Posted (edited)

 Hi guys, working on getting my FSX:SE install back up and running after a new computer build and I'm seeing some very weird behavior.  First a bit about the environment, main monitor is a 1440p Asus GSync with a max refresh of 144Hz, secondary is a 60Hz Dell at 1080p.  I mention this because last time I installed I was only using the Dell and did not notice this issue.  I later added the Asus monitor and after seeing just even worse FPS on my old rig, assuming due to the higher res, I just stopped flying for a while.  Now I'm wondering if it was the same problem (though I still needed an upgrade)

 

So here's the weirdness.  Originally had used my standard Nvidia Inspector profile which was set to 1/2 refresh rate (based on 60Hz monitor and capping FPS at 31 in dense areas).  Loaded in and was only getting 25fps with framerate on unlimited and just base FSX install.  No scenery addons etc.  FPS should be wayyyy higher.  Realized maybe it was a vsync issue, disabled Gsync and set inspector to "force off" for vsync.  Still, 55-60fps.  HOWEVER, if I go to windowed mode but a full size window (maximized) I'm getting like 220FPS.  What's going on here?  Do I need to start flying in Windowed mode?  I guess I'm ok with that since one of the reasons for dual monitor was so I could reference stuff.  I guess I'm just trying to understand why there's a huge FPS difference first and if I'm losing anything by going to windowed mode?  SGSS etc all seems to be working in windowed.

Edited by flyinion

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

This behavior is well known to the MS FlighSim series.  Some computers run FSX better in full screen mode, some run FSX better in windowed mode, some will show no difference in either mode.  It really depends upon the hardware combination and requires experimenting to find which mode works best.

Share this post


Link to post
Share on other sites
6 hours ago, stans said:

This behavior is well known to the MS FlighSim series.  Some computers run FSX better in full screen mode, some run FSX better in windowed mode, some will show no difference in either mode.  It really depends upon the hardware combination and requires experimenting to find which mode works best.

I did some more testing last night and it's definitely something weird with the multi-monitor config.  If I run full screen with both monitors on it's like it's enabling VSync or some internal FPS limiter even though it's forced off, and making it run at the max refresh of the slower monitor.  If I shut the older monitor off, it then runs with no cap in full screen.  Since I realized I want to have multi-monitor functionality anyway, I settled on of course windowed/maxmized and with GSync on and unlike my previous configs will force VSync off instead of 1/2 refresh/etc.  That way I should be able to still use the internal FPS limit in FSX as desired and GSync should result in keeping things smooth. 

 

Starting to think it may be time to move to P3D v4.x if I get back into simming seriously.  Seems like all the major addons I use, or the companies I like to buy them from at least, support that platform more than XP11.  I think at this point I set FSX back up more as an OCD part of reinstalling Windows/etc. now lol.  I also ran into another problem last night of realizing I can't reinstall FS Global 2010 FTX due to apparently having bought it from FSS as well.  I contacted Pilot's and unlike REX and SteveFX they were zero help in getting a working installer.  At least I have the mesh folder to copy over and try to import manually.  I just won't be able to use the flattener tool.  I figure if I get into simming enough again that I'm having issues with airport flattening I can just upgrade to the apparently more efficient 2018 version anyway.

Share this post


Link to post
Share on other sites

Hi Charles;

I just out of curiosity, what were your settings for your  60Hz Dell at 1080p in windowed mode?

NVI:

Frame Rate Limiter:

GSYNC:

Vertical Sync:

FSX:

Do you set a limiter in the FSX Graphics setting in or is it set to 31 or unlimited.

 

The reason I am asking is because I have the same dilemma with my 60Hz  ASUS 1080P I get the same performance in Windowed (Maximize) and Full Screen mode. I want to experiment with some setting to see how I can get the most out of my windowed screen. Since i didn't see a difference between the two I pretty much left my NVI settings as if it was in Full screen mode. I want to try some other settings as a true Window mode so see I can get a little more FPS flying. I get pretty good performance but any additional always helps in FSX

Thanks.

Forshaw.

Share this post


Link to post
Share on other sites

FSX windowed mode or maximized window mode, vertical sync is under control of the desktop. So for example on a 60Hz monitor it will run in Unlimited fashion, whatever fps it can do. In P3D there is a Display Setting function to control the output to the monitor frequency, VSync=On, which sets the frame time to the monitor. The fps wobbles around the monitor refresh frequency. In FSX it may be better to use the Frame Lock (slider), or an external limiter.

FSX exclusive fullscreen mode, vertical sync can be asserted with the GPU profile tool example Nvidia Control Panel or Nvidia Profile Inspector. So the vertical sync can be set to half on a 60Hz monitor giving around 30Hz. P3D exclusive mode does not exist, in fullscreen the simulation continues as if maximized borderless and captionless and so continues to require VSync=On to control the fps output..

Edited by SteveW

Share this post


Link to post
Share on other sites
3 hours ago, forshaw said:

Hi Charles;

I just out of curiosity, what were your settings for your  60Hz Dell at 1080p in windowed mode?

NVI:

Frame Rate Limiter:

GSYNC:

Vertical Sync:

FSX:

Do you set a limiter in the FSX Graphics setting in or is it set to 31 or unlimited.

 

The reason I am asking is because I have the same dilemma with my 60Hz  ASUS 1080P I get the same performance in Windowed (Maximize) and Full Screen mode. I want to experiment with some setting to see how I can get the most out of my windowed screen. Since i didn't see a difference between the two I pretty much left my NVI settings as if it was in Full screen mode. I want to try some other settings as a true Window mode so see I can get a little more FPS flying. I get pretty good performance but any additional always helps in FSX

Thanks.

Forshaw.

When I was on 1080p 60Hz I ran full screen, vsync at 1/2 refresh in Inspector, and then 31 or unlocked in FSX depending on area.  I found heavy dense city areas to be more stable at  unlocked (would still cap due to the 1/2 refresh rate setting) and would go locked when outside that.

Share this post


Link to post
Share on other sites
2 hours ago, SteveW said:

FSX windowed mode or maximized window mode, vertical sync is under control of the desktop. So for example on a 60Hz monitor it will run in Unlimited fashion, whatever fps it can do. In P3D there is a Display Setting function to control the output to the monitor frequency, VSync=On, which sets the frame time to the monitor. The fps wobbles around the monitor refresh frequency. In FSX it may be better to use the Frame Lock (slider), or an external limiter.

FSX exclusive fullscreen mode, vertical sync can be asserted with the GPU profile tool example Nvidia Control Panel or Nvidia Profile Inspector. So the vertical sync can be set to half on a 60Hz monitor giving around 30Hz. P3D exclusive mode does not exist, in fullscreen the simulation continues as if maximized borderless and captionless and so continues to require VSync=On to control the fps output..

Yeah I don't know what was going on still with the refresh rate other than maybe something about running FSX in full screen was forcing my 144Hz monitor to match the refresh of the 60Hz one, but I had vsync forced off in Inspector either way so it really didn't make sense that it was acting like it was actually on.  I know my monitors are capable of running at independent refresh rates however since as I mentioned when I run FSX in windowed, and before I installed all my scenery etc. (kills FPS), with GSync set to "desktop and full screen apps" the FPS would cap at 144 while in windowed vs 200+ with GSync off.

Share this post


Link to post
Share on other sites

Locked fps settings stress the system more because they build look ahead frames.

Any wobble in fps flow draws on the reserve frames and these must be replenished. If you can do double the fps with testing Unlimited, say you can see 50fps all the time at least, then you should be able to set say 24fps locked or so.

Share this post


Link to post
Share on other sites
2 hours ago, SteveW said:

FSX windowed mode or maximized window mode, vertical sync is under control of the desktop. So for example on a 60Hz monitor it will run in Unlimited fashion, whatever fps it can do. In P3D there is a Display Setting function to control the output to the monitor frequency, VSync=On, which sets the frame time to the monitor. The fps wobbles around the monitor refresh frequency. In FSX it may be better to use the Frame Lock (slider), or an external limiter.

FSX exclusive fullscreen mode, vertical sync can be asserted with the GPU profile tool example Nvidia Control Panel or Nvidia Profile Inspector. So the vertical sync can be set to half on a 60Hz monitor giving around 30Hz. P3D exclusive mode does not exist, in fullscreen the simulation continues as if maximized borderless and captionless and so continues to require VSync=On to control the fps output..

Hi Steve;

Thanks for the clarification. So let me see if I understand what you are suggesting, sticking to FSX (not P3D I know people complain about the true Full Screen mode) since that's what i am using and Window mode.

I set NVI Vertical Sync to:" Force ON"

NVI Frame Rate Limiter: "Off" Since i will set the limiter in FSX

FSX Frame lock Slicer to 30

Did I understand you correctly?

Thanks again.

Cheers.

Forshaw.

  • Like 1

Share this post


Link to post
Share on other sites

With window mode vertical sync cannot be controlled on the desktop. In windowed mode use the limiter in NCP or the Frame Lock in FSX.

 

Regarding GSync setups, vertical sync frequency climbs above the natural frequency to meet the fps as it increases. So generally that's good for fast fps games.

FSX and P3D run at a more sedate pace with lots of scenery. But say when gliding it may be preferred to set the scenery a bit more bald, so the fps is higher and more fluent.

Share this post


Link to post
Share on other sites

...we need to see 60 all the time or more to maintain 30 locked reliably. But anyway the locked fps setting is meant for low fps, try 20-25 because as the fps rises above 30 there's very short times between frames. Since the flow requires consistency rather than high fps as time between frames reduces the apparent consistency increases. Go for smooth, rather than fast.

Share this post


Link to post
Share on other sites

Also very relevant to fps is having an appropriate affinity mask, especially if hyperthreading is enabled. Can be losing 40% with HT enabled and no AM.

Another thing to increase fps is to make sure addon exe apps are corralled away from the core with the main process. 20-40% available there depending on system.

Another is to use the smallpartrejectradius setting can get you another 2-4%

 

Share this post


Link to post
Share on other sites
15 minutes ago, SteveW said:

Also very relevant to fps is having an appropriate affinity mask, especially if hyperthreading is enabled. Can be losing 40% with HT enabled and no AM.

Another thing to increase fps is to make sure addon exe apps are corralled away from the core with the main process. 20-40% available there depending on system.

Another is to use the smallpartrejectradius setting can get you another 2-4%

 

I never used to use a mask, always got worse performance on it in disc and SE versions on my 4790K.  Yeah that definitely changed with my 8 core 16 thread Ryzen 3000 series cpu.  I was getting some pretty bad FPS in the NGX 737 out of Seattle with OrbX PNW.  Locked a mask to cores 1-7 (left 0 open) and no HT cores and I literally gained a 50%+ increase in FPS off the end of the runway in spot view coming off 34R at KSEA where it's typically the worst.  Still need to play around with different masks, and maybe see if any upcoming BIOS/driver updates help things, but yeah I'm now getting better performance than my 4790K there, at 1440p with the default boost behavior hanging around 4.15-4.2Ghz, vs. the 4790K at 4.4Ghz all core at 1080p.

Edited by flyinion

Share this post


Link to post
Share on other sites

AMs are a mathematical thing and produce the results we want when we use them correctly. They can only restrict use of cores and logical processors. If the AM produces worse performance it has basically been chosen incorrectly.

Share this post


Link to post
Share on other sites
12 hours ago, flyinion said:

I never used to use a mask, always got worse performance on it in disc and SE versions on my 4790K.  Yeah that definitely changed with my 8 core 16 thread Ryzen 3000 series cpu.  I was getting some pretty bad FPS in the NGX 737 out of Seattle with OrbX PNW.  Locked a mask to cores 1-7 (left 0 open) and no HT cores and I literally gained a 50%+ increase in FPS off the end of the runway in spot view coming off 34R at KSEA where it's typically the worst.  Still need to play around with different masks, and maybe see if any upcoming BIOS/driver updates help things, but yeah I'm now getting better performance than my 4790K there, at 1440p with the default boost behavior hanging around 4.15-4.2Ghz, vs. the 4790K at 4.4Ghz all core at 1080p.

Just thinking what you said about HT cores:

The problem with FSX (and P3D) is that they will run two critical processes per core with HT enabled. That means the main process is forced to share a core when the core of that process should be reserved for the main process.

With HT disabled they put one critical process per core.

Switching on HT requires an appropriate AM. That is why we see such confusion about AMs and HT performance.

The no AM crowd should know that there is always an AM, a 4 core CPU is AM 15 no way out of that.

The no HT crowd simply have their setup wrong.

Edited by SteveW

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now