Captain Slow
-
Posts
53 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Posts posted by Captain Slow
-
-
30 minutes ago, speed-of-heat said:
mirror view wont help you, it takes no resources.
you need to enable reprojection.
there is no hardware that enable 90 in all circumstances, try my settings in my sig
I realize it is unlikely anyone will 90 in all circumstances, however I want to make sure I have understood the implementations of g2 reprojection correctly (as my first post) and am curious about the stuttering without it.
-
So I finally got both the 3090 and Reverb G2, but unfortunately my machine isn't able to produce 90 fps in dcs in *all* situations.
So, Ive been trying to understand reprojection, and the conclusion and experience I'm currently seeing is:
* If I enable reprojection, due to limitations in WMR + SteamVR, it will go to 45 fps (but still feels kind of like 90 fps) when the game starts rendering below 90 fps. Many times I dont even realize reprojection kicks in
, but somtimes I do (wild rolling causes certain hud elements to show some ghosting artifacts)
* If I dont enable reprojection, head rotation will still have reprojection, but pretty much everything else (head translation, aircraft movement, aircraft rotation..) will move at game framerate, for example 70fps - and for some reason it causes some super wild stutters (not a framerate issue, but rather just plain old stuttering), pretty much making the game feel unplayable.
Have I understood my options correctly? And also that there might be a better reprojection option coming in the future for us Reverb G2 users, which is basically option 2 without the stuttering?
For now I am sticking with reprojection enabled, because the second option/mega stuttering is just unbearable for me
I don't really understand why option 2 starts stuttering so madly below 90 :/. Normal desktop games with regular vsync on an triple buffering does introduce some stuttering when slightly below target fps, but nowhere near this
also, different question: Is there ANY way to disable the desktop mirroring of DCS VR view?
-
same issue here. 100% reproducibility. target box/diamond is just straight ahead regardless of where target is - and the angle off indication is locked at 4 degrees :)
-
LTWS TDC-over information does not move with contact if turning
I wish to inform you that this report is valid again, now that LTWS has officially been added:
https://forums.eagle.ru/showthread.php?t=233537
symbology is obviously a little different, but the principle is exactly the same.
You can basically have your tdc on the left side of the display while the target you are supposedly keeping the tdc over is on the right :)
-
-
SA page and RDR page both introduce very significant performance hits for me as well.
Basically 2-3 ms of frame time added by just those pages combined. The more contacts, the more impact. They dont seem very well optimized.
-
yup - having the same issue. The function has been removed I suppose ? (or was never supposed to be introduced yet).
-
You can still drop the azimuth width and point the nose, just can’t stay on the gimbles like that.
If you are super good, you can still command mini-raster with the designate button, but I think that’s like a 3-5* sweep. :)
pointing the nose means giving the opponent a better shot. I would avoid it if possible, so very much looking forward to being able to point the radar instead of my nose :).
mini-raster? you mean when you tell it to try to lock the first best target which comes into the highlight volume? I need something with more control in a multi target environment.
-
FYI: The IDLE problem was solved for me by disabling the game's built in fps limiter.
It can be found in <DCS_HOME>/Config/graphics.lua line 350
-
My situation has improved significantly by disabling the game's own fps limiter.
You can edit it in <DCS_HOME>/Config/graphics.lua line 350.
I just set it to 10 000 :). Before that I had some rather crazy issues where my GPU actually decided to go into "idle" power profile.
If you fly the F18, also check so that you havent run into this issue I fought against for a few days:
https://forums.eagle.ru/showthread.php?t=233582
NOTE : The solution in the thread is actually wrong. .net had nothing to do with it.
I dont know what the solution is yet, but after reinstalling windows and removing the fps limiter it seems to work better.
-
I'm 90% certain we will get HAFUs on radar screen (because there are TODOs in the lua code about it ;) ), even if we might not be able to link radar tdc with sa tdc.
-
Sounds like you are referring to something different than the OP is. He is referring to being able to select an off-center scan pattern during search. You are referring to the scan pattern when attempting to designate from the looks of it.
Yeah, given the answer by Anonymous User I think we are still waiting for it to be implemented.
Re-engaging after being on the defensive right now is quite dangerous and slow. Need to be able to reposition scan center.
-
My 2080ti does the same quite often, and I have yet to figure out what is the cause.
I have verified with gpu-z that indeed it sometimes goes to idle in DCS! No idea why.
I thought I was the only one... But apparently I am not, and this isn't a single user case.
My system is 9900k, 32 GB ram, Palit 2080ti gamingpro OC.
What about you?
This is on a fresh windows installation! (yesterday) with zero usb input devices attached except mouse and keyboard
Setting the card to "prefer max perf" has absolutely no effect on the issue.
Screenshots attached
Flapping idle/full perf in game
Idle in menus:
Idle looking at sky:
-
Reproducible: 100%
Date: 2018-03-03 1940 Z
Sim version: 2.5.4.28090 Open Beta
Mission: Any with air target (Track recorded in F18 instant action aim120 training mission)
Steps to reproduce:
1. Angle your radar to scan the targets
2. Look on SA page
3. Position TDC around contacts
Problems:
* TDC should be drawn on top of contacts, so we can read altitude and mach number - right now it is the opposite
* same issue with SA page RWR info
* Draw order among contacts order is reversed (prio 1 should prob be drawn topmost/last - right now it is the opposite)
Track: attached.
Screenshot: attached.
Desired solution:
* TDC should be drawn on top of/after contacts
* TDC should be drawn on top of/after SA page's rwr symbols
* draw order among contacts should favor highest prio
-
You can decenter the scan, but it does not follow the TDC like in the F15/16. You depress the TDC to set the center of the scan.
Holding the TDC for >0.5sec enters ‘spotlight’ mode centered on where you have your cursor at the time of depression.
This is not yet implemented, right?
-
same issue. star not present if locking PPLIs
-
Here is a quick question to the devs:
Are you aware that the range cues (120B, 120C, possibly others as well) on the HUD are completely wrong at the moment, do do you want a bug report with tracks etc for it? Asking before repeating details for something already known.
(Raero is not drawn at all)
(Rmax is wildly off/drawn factor 2 too far away tail chase against non maneuvering target)
(Rne is wildly off against head on target: appears at roughly 80-90% of Rmax)
(Rmin is drawn at R=0/merge)
(NIRD 12oclock does not represent R=0, in fact, 12oclock is never reached)
-
What is the behavior in the real F18 in bvr modes when azimuth limits are set to less than 140?
Can you slew your radar scan limits left/right manually like the russian jets?
Do the radar scan limits follow the tdc like in the f15/f16?
Or is the current behavior where it only scans around your own heading the only real option...
I am hoping not,...
-
Okay, so: I have now verified the issue to be significantly reduced after reinstalling the game.
However what is interesting here is how the reinstalled folder differs from the old folder. Basically there is almost no difference. The following is a recursive diff comparing the two installations (produced with diff inside babun shell).
Files /c/DCS/autoupdate.cfg and /c/Program Files/Eagle Dynamics.bak/DCS World/autoupdate.cfg differ Files /c/DCS/autoupdate.dat and /c/Program Files/Eagle Dynamics.bak/DCS World/autoupdate.dat differ Files /c/DCS/autoupdate_log.txt and /c/Program Files/Eagle Dynamics.bak/DCS World/autoupdate_log.txt differ Files /c/DCS/autoupdate_log_old.txt and /c/Program Files/Eagle Dynamics.bak/DCS World/autoupdate_log_old.txt differ Only in /c/DCS/: dcs_variant.txt Only in /c/Program Files/Eagle Dynamics.bak/DCS World/Mods/aircraft: A-10C Only in /c/Program Files/Eagle Dynamics.bak/DCS World/Mods/aircraft: Flaming Cliffs Only in /c/Program Files/Eagle Dynamics.bak/DCS World/Mods/terrains: PersianGulf Files /c/DCS/unins000.dat and /c/Program Files/Eagle Dynamics.bak/DCS World/unins000.dat differ Files /c/DCS/unins000.exe and /c/Program Files/Eagle Dynamics.bak/DCS World/unins000.exe differ
No significant binaries differ, and neither do the configs (if you check the autoupdate.cfg for example the only diff is I had more aircraft modules installed in the old installation).
However, after reinstalling: My old installation also runs faster (!)
So I think it is safe to conclude it had nothing to do with the actual installed game contents.
I am getting somewhat of a PEBKAC feeling here, due to the installation opening the following window :) (reason: the old installation was a copy-pasted folder from a previous computer's dcs-installation)
I am surprised the game even ran before, if some libraries were missing. I am not a .net (or windows) dev myself so I dont know how windows handles version mismatching in dynamically loaded libraries, but it could have tried to execute certain functions from an older/newer or compatibility version of the library causing certain calls to run significantly slower
BUT
And this is the important part. The stuttering still exists, even though it is greatly reduced, I actually believe most people are having it, even though most are not sensitive enough to feel it.
I do not think a single mfd page of very simple symbology (having built software for several of these instruments myself) is "worthy" of being a significant performance sink in the game/sim. Therefor I would still hope you will look into making the SA page rendering a lot faster (for dynamic objects)
See the new performance results below:
MP:
SP:
-
Deleted the following folders without any improvement:
- Saved Games/DCS
- %APPDATA%/Local/DCS
- %APPDATA%/Local/Temp/DCS
Reinstalling DCS to see if that helps. If it helps, I will track changes to all non-binary files in my DCS installation using a git repo or something :P
- Saved Games/DCS
-
Now that's a thorough report there... and I first thought it might be "low hardware" until I saw your specs which are basically full AB.
Would like to know how to get those frame time graphs though
I use a free program called "Rivatuner Statistics Server". I asked a squad mate to install the same and he had no stutters in his graph ;(
-
Yeah, thats the tricky one, my machine is lower spec than yours and I am not seeing this as well.
Have you tried the usual suspects like deleting the shader folders and all that?
Fxo and metashaders? Anything else?
-
I have concluded that the code causing the stutters in my single player tests is called by
function SA_FF_Symbol(parent, isOutlined) . . . end
The SA_FF_Symbol function resides in Cockpit/Scripts/Multipurpose_Display_Group/Common/indicator/Pages/MPD/MPD_page_defs.lua
If I comment out that function's contents all microstuttering in my single player test stops. This of course also stops any rendering of tracks on my SA page, so it is pretty far from a solution :).
And it doesnt exclude the possibility that for example PPLI drawing might still cause stuttering (The single player test mission/instant action aim120 training only has you on blue side, so no other PPLIs available)
What I really dont understand is why my machine (which is far more powerful than some of my squad mates) stutters badly in ONLY the SA page ... while everyone else seems to be able to play just fine..
-
* It seems like the stutters are tied to the rendering of the SA page's targets. If I am in a continous turn the targets are only repositioned on the SA screen sometimes (like 4 Hz), and it seems to match exactly the frequency of the stutters.
* Windowed/fullscreen/gsync/vsync settings have no effect on the situation
G2 VR newbie in need of reprojection advice
in Virtual Reality
Posted · Edited by Captain Slow
Thanks, good to know. That is pretty much the same machine I have (9900k and 3090).
The biggest contributor to slowdowns, by far looking at just frame times, seems to be the category "Simulation". Is there any explanation somewhere what the different frame time categories include more in detail? "Simulation" normally sits around 5 ms, but when the slowdowns start it jumps to above 10 ms every few frames, probably even quite a bit more. Thus usually happens when flying closer to the ground and near cities. (example numbers taken from simple instant action testing)