Jump to content

Phaëthon

Members
  • Posts

    45
  • Joined

  • Last visited

Posts posted by Phaëthon

  1. Doesn't seem to do much for me...

    here's a preliminary test recorded with MSI afterburner, the .trk I used (free flight over nevada).

    Everything is in the attachment. Results are kind of good meaning I don't loose a lot going from low to high settings. I don't really see that much of a difference from Vsync off, though, which I used to run before.

    The whole data is in the attached file, including the .trk

     

    I might test the same against Vsync completely off...

     

    with my usual settings

    [ATTACH=JSON]{"alt":"Click image for larger version Name:\tLow Settings.jpg Views:\t0 Size:\t98.0 KB ID:\t7151863","data-align":"none","data-attachmentid":"7151863","data-size":"thumb"}[/ATTACH]

    Average FPS = 128

    No. Samples = 644

     

    -----------------------------------

     

    With "High" settings

    [ATTACH=JSON]{"alt":"Click image for larger version Name:\tHigh Settings.jpg Views:\t0 Size:\t93.5 KB ID:\t7151865","data-align":"none","data-attachmentid":"7151865","data-size":"thumb"}[/ATTACH]

     

    Average FPS = 121

    No. Samples = 644

     

    system is 9700k@5GHz, 64GB RAM@3200MHz, 2080Ti FE

    Thank you for posting it the results! Just to check with you, you did test with vsync off ingame and vsync turned to FAST in nvidia Cork panel, right? That option inside the nvidia control panel was what made all the difference for me.

  2. @tagomago You are on the path to DCS enlightenment. Try Texture filtering quality = High Quality, Texture Filtering Anisotropic Sample Optimization = Off, Antialiasing Mode = Enhance the application setting, Antialiasing Setting = 2X, Antialiasing Transparency = Multisample, Texture Filtering Trilinear Optimization = Off.

     

    I use vsync ON in Nvidia control panel, not DCS and fly at a rock solid 60fps. If it's dropping below 60 it means you are still trying to squeeze too much into your graphics card. Adjust visibility range to HIGH and Preload ratios, trees, and grass to about 2/3 of the slider range or until the fps drops go away.

     

    And when the next beautiful update comes out, rinse and repeat. :)

    Thank you for the suggestions @glide , I'm always interested in settings that produce better overall quality (but not to the point of not meeting the target fps of the monitor's refresh rate). You're post however also misses the point: like I've previously mentioned the settings I'm running are in no way shape or form the best settings, some of them may even be conterproductive in terms of performance or quality. I'm just referring to them so that we're able to repeat the same conditions and work our way from there. This forum post isn't a guide on the best quality/performance settings for DCS, but a report of an abnormal fps gain from setting the vsync Nvidia option to 'Fast' + dcs vsync to 'off', and at best asks other users to try it out and report the results.

     

    Having said that, I'll try your suggested settings and tell you what I think. Cheers!

  3. The flat terrain shadows is terrible in quality and it is much better to switch it to default, however, if you do so, 60 fps is impossible at the low altitude where there are thousands of trees.

    Thank you for the suggestion! Honestly, I don't think the extra cost in fps is worth it for the visual quality it gives. Although it's a very noticeable feature if you're looking at static images and scenery, when you're actually flying missions in a dynamic environment it will be practically negligible. This however is entirely subjective and it's up to each one's personal preferences and gpu headroom.

    You are however missing the point of this post: have you tried the suggested settings and gained any fps?

  4. It all depends on individual hardware indeed. I use 60 Hz monitor and on that one, fast sync has been always inducing microstuttering mess at low altitudes, even without TrackIR.

     

    With standard V-Sync, a bit of input lag (not noticeable in flight simulators in my opinion) is a price I'm ready to pay for otherwise buttersmooth rendering, but it's subjective thing, obviously.

    No worries, you can probably test it in whatever map you want as long as you have a base of reference. The settings I'm running are in no way shape or form the best settings, some of them may even be conterproductive. I'm just referring to them so that we're able to repeat the same conditions and work our way from there.

    I've previously compared MSAA on and off and the impact wasn't significant (I could be mistaken, but I had the weird impression that without MSAA performed worse than with 2x), at least for the RTX 2070 (surely hardware differences can play a big part in this), so I decided to let it stay on as it nicer visually.

     

    It would be good to replicate the settings and have a comparison between:

    (vsync off ingame + Nvidia vsync "let app choose") VS (vsync off ingame + Nvidia "Fast")

  5. It all depends on individual hardware indeed. I use 60 Hz monitor and on that one, fast sync has been always inducing microstuttering mess at low altitudes, even without TrackIR.

     

    With standard V-Sync, a bit of input lag (not noticeable in flight simulators in my opinion) is a price I'm ready to pay for otherwise buttersmooth rendering, but it's subjective thing, obviously.

    @Art-J and @Boosterdog, you have similar specs in terms of processing capability as my setup; if you can, for the sake of repeating the experiment, can you try replicating with all the settings I mentioned (I moved the Win 10, Nvidia and DCS settings to the 1st post).

    When I started my adventure in DCS (not too long ago) I ran into all kinds of videos and guides on the internet claiming miraculous settings and my conclusion in the end I had not gained one single fps. Probably some are more beneficial to lower end machines, and I'm not denying that some people may find better performance following those guides. They didn't do it for me and have since left most of things back to default.

    Just wanted to make it clear that I'm not trying to sell you that kind of snake oil :) I'm just reporting this finding, which I believe exceeds what vsync is intended for, and possibly asking other users to experiment and verify if it works for them, so that we can have a better picture of what's going on and how.

  6. I tried disabeling vsync in game, and enableing in nvidia control panel...My fps went from 30 to 60...But now problewm is on my second screen - i use ipad to control MFD screens, but now my left and right MFD video screens on ipad are flcikering...When i turno vsyinc inside DCS problem gone, but fps loss is back.. Did anyone else noticed that?

    Hmm, sorry to hear that. Maybe some technical limitation with how Nvidia fast sync works, but I wouldn't know for sure as that is beyond my knowledge and the scope of this finding :)

  7. It all depends on individual hardware indeed. I use 60 Hz monitor and on that one, fast sync has been always inducing microstuttering mess at low altitudes, even without TrackIR.

     

    With standard V-Sync, a bit of input lag (not noticeable in flight simulators in my opinion) is a price I'm ready to pay for otherwise buttersmooth rendering, but it's subjective thing, obviously.

    I think you’re right, but what I’m suspecting with my finding is not about whichever vsync method you choose working as intended. Vsync working properly would result in smooth motion with no screen tearing.

    I am however reporting a gain in game performance which isn’t exactly what vsync is about.

  8. I tried disabeling vsync in game, and enableing in nvidia control panel...My fps went from 30 to 60...But now problewm is on my second screen - i use ipad to control MFD screens, but now my left and right MFD video screens on ipad are flcikering...When i turno vsyinc inside DCS problem gone, but fps loss is back.. Did anyone else noticed that?

    I haven’t tried a multiple monitor setup. However can you check what’s the refresh rate of each of the displays you’re using? Do they all have the same refresh rate? (Possibly not but check it out and report back)

  9. I have tested all the maps and at least in my configuration the heaviest maps are indeed the Channel and Syria - they're the most recent maps and I feel there's a lot of investment in terms of the sheer amount of assets, textures & detail, etc. I haven't compared fps between them, or any kind of scientific approach to it, but from a purely subjective observation I would say they run some 20% slower, and are the only maps where I constantly get under 60fps. However, my latest entry in this forum might interest you:

    https://forums.eagle.ru/forum/english/digital-combat-simulator/dcs-world-2-5/7148299-dcs-big-fps-increase-vsync-off-ingame-vsync-fast-in-nvidia-ctrl-panel

    • Like 2
  10. I always test things like this (or almost always) when i read them so i again did now.

    The FPS rate did go up with about 90% (Syria map, F/A-18 free flight), going from a synced 60 to roughly 110 FPS. All cool.

    But then i added TrackIR into the equation and things got quite stuttery when looking sideways when the FPS is not on the monitors refreshrate (or synced) or on an exact division of it (running 120Hz monitor so either 120, 60, or 30).

    After seeing the stutter i turned VSync back on, stutter gone, all smooth ride. No questions asked.

    Hmm interesting. What TrackIR are you running? My testing was done with TrackIR 5 and didn’t notice those stutters. Tested with the spitfire landing on the Channel which was giving me problems.

     

    Edit: I would be cautious not to rule it out because of TrackIR tho, as the equipments sureley throwing more complexity into the equation. The problem with the TrackIR stutter could very well come from a different origin, even if triggered by the vsync option.

  11. To answer your comment it is not producing more per se its just taking off the brakes that standard vsync imposes. The part of your testing where you couldnt hit 60fps even after turning off vsync in game was likely something in the nvidia driver settings carried over from when you had it enabled. Ive experienced similar when doing similar.

     

    Having had the "benefit" of DCS for 5 years I can pretty much say I know what works for me and I know what doesnt. That said, I also confidently told myself that Id never see a Eurofighter so Ill give it another go and report back.

     

    EDIT - as predicted - a horrible jerky stutterfest.

    Damn. Well, here are my specs just for reference, if it helps anyone:

    Ryzen 7 3700X / Nvidia RTX 2070 Super / 32gb ram

  12. To answer your comment it is not producing more per se its just taking off the brakes that standard vsync imposes. The part of your testing where you couldnt hit 60fps even after turning off vsync in game was likely something in the nvidia driver settings carried over from when you had it enabled. Ive experienced similar when doing similar.

     

    Having had the "benefit" of DCS for 5 years I can pretty much say I know what works for me and I know what doesnt. That said, I also confidently told myself that Id never see a Eurofighter so Ill give it another go and report back.

     

    EDIT - as predicted - a horrible jerky stutterfest.

    Hahaha, totally onboard with you. And I also don't know of a vsync variant that can prevent the issues if frames fall lower than the refresh rate. From what I could read normal vsync suffers a bit when it happens, and other variants try to mitigate that problem in a number of ways.

    But what boggled me was the comparison between the (ingame vsync off + nvidia "let app choose"), which fluctuated between ~30 and ~50, and the (ingame vsync off + nvidia "fast vsync") fluctuating between ~70 and ~110.

    For the ingame fps counter, both methods should display the full fps output and wouldn't have a reason for such disparity. I know that we're talking about the full output of the gpu and not what actually gets displayed, which is capped at the tv refresh rate (in this case 60hz). You may have something there, when you mention the possible buffer effect, or the "close to double the frame rate". Could also be the carry over you mention or the result of a combination of several things, I don't really know for sure.

    But the sudden jump in total output of the gpu (at least in the fps counter) and the smoothness that was very real kind of tells me that there's something weirder going on.

    Curious to check your results!

  13. TLDR: disable vsync option ingame and turn it to "Fast" in Nvidia Control Panel Vsync.

     

    Guys, wanted to report something I've found following our discussion in discord regarding vsync.

    I was under the impression that having the VSYNC option turned on ingmae (for any game, for that matter) could not impact negatively the game performance as it would only drop the frames that were in excess of the monitor's refresh rate. That turns out not to be true (entirely):

    When the gpu is able to output more frames than those required, that's what happens - excess frames are dropped and screen tearing avoided.

    However when the frames suddenly drop below the refresh rate ingame, the Vsync actually struggles to sync things and it actually enforces additional stutter and lag. I didn't knew this, but I'm sure most of you did.

    This however was just half of the finding. I bought the Channel today and spent all afternoon testing the performance (vsync left on) with the same mission over and over and I was getting miserable results, the fps not hitting the 60fps (to match the tv sync rate) and getting horrible stutter. My average fps was around 50, with many drops to ~30.

    I decided to try with the vsync off and it didn't make any improvement but only added the screen tearing when ocasionally the gpu was able to go above the 60fps.

    I'm not trying to be a smart guy but this is not really any revelation but a common to all 3D programs (not only games) fact about VSync which is known since many years. VSync from a technology perspective is an abomination that comes with a price of tremendous input lag and decrease in performance, including drops to half of monitor refresh rate when GPU can't sustain full rate. If I could point a setting that must always be changed, forcing VSync off in global settings would be the first one.

     

     

    Input has a negative affect on any input device. Not only the mouse movement but also keyboard and TrackIR. If you get a constant feeling of a delay in screen response comparing to a head movement with TrackIR, turn the VSync off.

    From my post you can assume that I agree that ingame vsync option should be turned off.

    Do try the "Fast" vsync option in Nvidia Control Panel to see if it makes a difference.

  14. Fast sync works for some not for others. You are not getting 110fps displayed. The fps counter simply takes an average of frames over a second that the game engine is outputting. Where vsync limits that output and attempts to precisely control the pacing of output (i doesnt drop frames), fast sync doesnt and lets the engine produce all it can. Fast sync then selects the most recent complete frame to use in the display buffer whilst honoring the limits of the sync cycle. So you're still getting 60fps on screen. What you perhaps are also getting is the benefit of having something in the bag in terms of frames stored in the "flip flop" buffers Fast Sync uses. Potentially this might help over standand vsync when the output drops below the required rate (60 in your case) to fill the gap. Given the unlimited output of fast sync it probably avoids the 1/2 refresh rate enforcement to when the output drops below the target to.

     

    Its designed to work best when your rig is producing close to double the number of frames to the refresh rate but doesnt seem to be totally reliant on it. The lower that number goes however, the less fluid the image and the more frame pacing issues prevail.

     

    Personally with my rig (6600K, 1080 @ 1080p) fast sync has never produced anything but issues. Ive also stood over the shoulder of friends who swear its working great when what im seeing is a stuttering mess. Ive never found a better soluton to vsync and a options set up that ensures the ouput never drops below the refresh rate on my TV bound and old rig. Im very sensitive to uneven frame pacing though.

     

    Its worth a try for sure, but try it with a decent mission with a few assets loaded to check it works well.

    I know how it works and I agree with you. Not entirely sure why a vsync option would be responsible for a fps boost. However more users have reported that things did become a lot smoother and got around +30fps. I don't know if this is configuration dependent, if it relies on a combination of other settings, but still think it's worth the try. If you do try, I'm curious to know your results. Hope it helps!

     

    Note: my computer doesn't particularly struggle with any of the maps, except for the case I mentioned in the Channel. Do test it on a decent mission with lots of assets and compare.

    I'm also very sensitive to uneven frame pacing and the kind of smoothing I got isn't the 'vsync smoothing".

  15. TLDR: disable vsync option ingame and turn it to "Fast" in Nvidia Control Panel Vsync.

     

    Guys, wanted to report something I've found following our discussion in discord regarding Vsync. I was under the impression that having the Vsync option turned on ingame (for any game or 3d application, for that matter) could not impact negatively the game's performance as it would only drop the frames that were in excess of the monitor's refresh rate. That turns out not to be (entirely) true:

    When the gpu is able to output more frames than those required, that's what happens - excess frames are dropped and screen tearing avoided. However when the frames suddenly drop below the refresh rate, the Vsync actually struggles and ends up causing additional stutter and lag. I didn't knew this, but I'm sure most of you did.

     

    This however was just half of the finding. I bought the Channel map today and spent all afternoon testing the performance (Vsync left on) with the same mission over and over (Instant Action: Spitfire -> Landing -> Channel) and I was getting miserable results, the fps not hitting the 60fps (to match the tv's 60hz refresh rate) and getting horrible stutter. My average fps was around 50, with many drops to ~30.

    I decided to try with the Vsync off and it didn't make any improvement but only added the screen tearing when ocasionally the gpu was able to go above the 60fps.

    But then I decided to keep the ingame Vsync option off and activate the Nvidia Control Panel vertical sync option to "Fast", leaving that work to the gpu.

    I have absolutely no idea how it could have made such a difference, but I amazed with the results: no tearing, absolutely fluid, hitting average 110fps. That is a HUGE difference. I hope you find this useful and check if it helps your performance. Gaining a pair of fps always makes me smile.. but we're talking a night and day difference of more than 50fps in some cases.

     

    (Sorry for the lengthy msg)

    Fly safe!

     

     

    Note: as of the time of writing this post, several more users have been able to confirm these fps increase :) you might as well give it a try!

     

     

    Edit: just in case it is dependent on a combination of settings and configuration, I'm leaving my specs and settings here just for reference:

     

    - Specs:

    Windows 10 Pro 64-bit | AMD Ryzen 7 3700X | Asus ROG Strix B550-F Gaming | Asus GeForce RTX 2070 SUPER Dual EVO OC 8GB | HyperX 32GB DDR4 3200MHz | SSD Samsung 970 EVO Plus 500GB M.2 NVMe | TrackIR 5 | Oculus Rift S

     

    - DCS version:

    DCS 2.5.6.57949 Open Beta

     

    - Windows 10:

    Game Mode: On

     

    - Nvidia Control Panel:

    image_263424.jpg.0c81895fa15b9847b8de42d47b9f7883.jpg

     

    - DCS Graphical Options:

    image_263425.thumb.jpg.1032f565c491b91bc06510575bee02d5.jpg

    • Like 1
    • Thanks 2
  16. When I saw the wip video of the new kneeboard features I was happy to see those changes implemented, however I was critical of how the window stretch behavior is applied. Resizing the window to a different aspect ratio shouldn’t distort and change the aspect ratio of its contents. It should act as a frame, and the contents fit to width or height with the option to drag inside if it exceeds one of the dimensions. It’s been a staple of UX. Imagine that something as simple and ubiquitous like pinching to zoom a picture or webpage in a smartphone would change the aspect ratio and scale unevenly. Oh, and please don’t get me wrong, I don’t mean for this criticism to be unfair, things are probably still in wip in that regard. Consider that it’s just my OCD talking :)

     

    edit: A nice feature to complement that feature would be to add a button to lock/unlock the aspect ratio when resizing the window.

    • Like 2
×
×
  • Create New...