[REPORTED]2.5.6 Performance

Just an update.

After several without DCS and half a day downloading, I have reverted back 2 patches (41964?). It seems to have made DCS playable again. It is not as good as it was before the 2 updates (choppier) , but it is functioning.

I also had submitted a ticket regarding the fps, and have already had a good response from ED. Nothing specific but I figure maybe I can submit some files or help via the ticket.

If anyone else's DCS is broke perhaps a rollback to 41962 is in order. I tried, DCS repairs, cleaning files, drivers, Bios update and Clean update of Windows 10. These helped a little with smoothness and small fps increases but reverting back was the only solution.

Make no mistake, some may have been lucky with last 2 patches, there is still a major problem being experienced by VR users. Apparently I hear from above server traffic has been lower. If this is indeed the case than its quite possible there are more out there with this issue not reporting it.

Thanks for the help all

I do not seems to be having any fps or stuttering issues in both 2D and vr for the instant action mission I played in 2.5.6.

Everything played really good.

My only issue is that I still get logged out of dcs the last time for someone logging in from a different machine.

Otherwise 2.5.6 is okay.

In VR on a medium to low graphics settings, dcs PD of 1.0 and steamVR SS at 179%, game was wasn't different from 2.5.5. I still get averagely 36fps in vr.

The are occasional micro stuttering but not enough to stop me from playing.

I am wondering why it really unplayable for others who oftentimes have beefier system.

Are there other factors at play beside the issues with 2.5.6 update?

I´m not an expert, but i think something is broken with vsync. Prior to 2.5.6 i had solid 60 fps in SP as well as in MP and after the update my fps dropped to 25-35 in average. I read something about vsync and that it sometimes doesen´t perfectly work and then can cause fps drops in games, so i wanted to give that a try and turned it off yesterday. Now in rare cases i still have fps drops, but not down to 25-30 and my average fps went up to 60-80 i´d say. And it looks the same, means no typical non-vsync issues... wtf? 2 days ago i would have stated that the game has become almost unplayble for me, but now i´d say the total opposite. Without vsync, it "feels" even better than before the update...

Shut the front door... Can that be confirmed?

Based on the fps hit sounds about right... if not 3x the polys....

 

I think its interesting how FPS bumps up to (almost) normal when I pause mission (esc) and look around (VR) and then unpuase it becomes a god awful nightmare

 

I'm still waiting for a rolback patch to download.... 3 hours and counting.

My performance is still abysmal after the new patch, maybe a little worse even. Now I'm getting stutters and crashes too. As a mission maker I re-test the same missions on the same system all the time, so I have a good benchmark. DCS has never ran so bad.

 

SkateZilla, if you're right, that should explain everything. Fingers crossed for a quick fix!

It shows in the screenshots, but I doubt it's the root cause because users without Ka-50 are having issues w/ Performance.

I´m not an expert, but i think something is broken with vsync. Prior to 2.5.6 i had solid 60 fps in SP as well as in MP and after the update my fps dropped to 25-35 in average. I read something about vsync and that it sometimes doesen´t perfectly work and then can cause fps drops in games, so i wanted to give that a try and turned it off yesterday. Now in rare cases i still have fps drops, but not down to 25-30 and my average fps went up to 60-80 i´d say. And it looks the same, means no typical non-vsync issues... wtf? 2 days ago i would have stated that the game has become almost unplayble for me, but now i´d say the total opposite. Without vsync, it "feels" even better than before the update...

Having said that, it's normal for V-sync to create some issues some times. It eliminates tearing at the cost of introducing input latency and (potentially) costing a few FPS, making the game feel less responsive at times. Were you using V-sync from within the game, in the options? If yes and then you disabled it, it's normal to see higher FPS and responsiveness.

The reason you're not seeing a difference in image tearing etc could be because of your graphics card default options that are applying another method, as long as the in-game one is disabled.

Dear DCS.

Please send me a RTX 2080 Ti for free? Please.

Dear DCS.

 

Please send me a RTX 2080 Ti for free? Please.

It won't help :( I have stutter fests randomly flying, or on bases cold starting alot. I turned settings down to get to 45 when flying mostly alone in the PG map, but going in for landings its half that at best.

The more concerning value on all these screenshots is "Simulation" on the main thread. In my tests this value has the biggest impact on framerate. Sometimes it's crazy going up and down, what is this? Physics calculations?

It won't help :(

I am willing to give it a try :-)

It won't help...

lol

I think there are multiple problems with 2.5.6. The interesting Thing is, that some missions or campaigns seem to be more affected than others. E.g.: The arab spring campaign is playable in 2.5.5 but also has a bad GPU usage (50 -70 %) same for the Taskforce Challenger Campaign!

In 2.5.6 these campaigns are unplayable / low 20's FPS.

Other campaigns are nearly as good as in 2.5.5 e.g.: Eastern Friendship

framrates are nearly as good as in 2.5.5

Also the Flaming cliffs free flight missions have good FPS in 2.5.6

At the moment I think there is a "dual Performance loss" in 2.5.6

1) Slight loss through the new lightening

2) Massive loss through AI problems / high CPU low GPU usage. It does not matter if you have a 2080ti or a 1080 because the GPU usage is in the low 50's / the GPU is not maxed out. And there I think this is because with 2.5.6 the game has maxed out the CPU so that the GPU can not be maxed out

Just tested again a Taskforce Challenger Mission:

2.5.5: GPU usage 55 - 60 %

2.5.6: GPU usage low 30's %

For some reason 2.5.6 does not use the whole GPU power!

CPU cores where nearly the same

Settings are equal!

Was trying to pinpoint the reason or one of the reasons why I in certain missions would have dramatically less fps in 2.5.6 compared to 2.5.5. After a lot of trial and error with removing units or certain unit types, I came across the following:

Infantry, infantry close to uncontrolled AI aircraft totally tanks the performance in 2.5.6. I tried putting them close to other unit types but never got the same result.

In the attached test mission its just eight IL-76 uncontrolled AI on the ramp at Gudauta, and after 10 seconds a group of 8 infantry activate close to the first aircraft, 5 seconds later, group 2 activates and it keeps going. Notice the fps. Here from group 6 it really starts taking a dive.

If I move the infantry outside the the planes 'square footprint' then fps remain unchanged.

So its something with 'personal space'. If that is all infantry, or units, or just certain, and only certain against certain. I dont know, but they certainly make the difference for me, and in the mission I have been testing 2.5.6 against 2.5.5, well I have infantry groups close to uncontrolled aircraft, and removing those gives me twice the framerate (still 30 short, but I have a lot of units of various type close to each other).

FPS AI Test.miz

Sounds an awful lot like the new pathfinding is going for a spin if/when it tries to manoeuvre around placed obstacles…

I am finding there are also performance issues around the AI planes that are flyable modules, maybe the new lighting is affecting higher LOD modeled planes?

Flying the single mission included with the F16 module at the Nellis range completely tanked my FPS in VR especially when the A10s taxid past. i've also noticed that when an F16 is near me even if im flying another module has the same effect.

I'm no expert but i think there might be something going on here with the models that have been reworked, this would also include any of the ground units. When using free look and moving around the older models i see no impact on frames. When doing the same thing with any reworked higher fidelity model i get the a much larger than normal FPS hit.

-Could the new lighting and the higher poly count models be causing issues?

-Are LOD levels scaling correctly?

I noticed that in VR i can now spot planes when using the VR zoom function out to 40+ miles and even a bit further when using the spyglass zoom. Im not complaining about this one as its a massive improvement and it saves me having to use the dot labels which is very cool! Could there be a relationship between the LOD and distance?

So can we all agree here that 2.5.6 has brought a drop in overall FPS performance.

If I fly my F-16 around in a solo mission on PG on 2.5.5 when coming in to land I get zero stutter. When I use 2.5.6 with the settings the same, the mission the same etc I get major stutter when coming into land.

32Gb RAM, SSD, 2070s, TrackIR.

Sent from my iPhone using Tapatalk

I noticed a lot of stuttering, and it was unplayable for me in 2.5.6, so I have reverted to the latest 2.5.5 version, and it seems to be back to normal. I use DCS in VR exclusively, but some others using a monitor only seemed to report an increase in performance after the 2.5.6 update. It may depend on the mission(s) being flown at the time. I have been testing in the Hornet Aggressors BFM campaign and with a self made mission with a number of ground and air units, both of which I was using quite regularly before the update.

Something I noticed yesterday as I was installing the shader mod, and thus doing a little performance data gathering, is that performance on my system (I play in VR) is not actually GPU bound like most VR games seem to be. Instead, oddly, I'm CPU bound. The CPU is just barely fast enough to get around 90FPS while sitting still on the ground in a cold and dark Hornet. As soon as I start turning things on, the CPU frame time goes up to around 12ms, which is too long for 90FPS and I get reprojection. When I'm airborne, it's generally right around 12-13.

I should note that I'm running the game on a system with a 2080 Ti and a 9900K clocked at 5GHz. This is the fastest setup it's practical to have (excluding the RTX Titan). What's really interesting to me, though, is that the game appears to be VERY CPU bound, even on the fastest CPU on the market. Edit: The way this appears to be happening is that the game is running only a couple of threads, and the single core performance of my CPU isn't high enough to get the frame time down below 11ms.

Something I noticed yesterday as I was installing the shader mod, and thus doing a little performance data gathering, is that performance on my system (I play in VR) is not actually GPU bound like most VR games seem to be. Instead, oddly, I'm CPU bound. The CPU is just barely fast enough to get around 90FPS while sitting still on the ground in a cold and dark Hornet. As soon as I start turning things on, the CPU frame time goes up to around 12ms, which is too long for 90FPS and I get reprojection. When I'm airborne, it's generally right around 12-13.

 

I should note that I'm running the game on a system with a 2080 Ti and a 9900K clocked at 5GHz. This is the fastest setup it's practical to have (excluding the RTX Titan). What's really interesting to me, though, is that the game appears to be VERY CPU bound, even on the fastest CPU on the market. Edit: The way this appears to be happening is that the game is running only a couple of threads, and the single core performance of my CPU isn't high enough to get the frame time down below 11ms.

Nice run down there TheChafing. I too have found that my CPU on 2.5.6 is maxed out whereas my GPU is now hardly being tickled.

On 2.5.5 it was the other way round - the GPU was doing all the work and the CPU was working hard sure, but not to the point of bottle neck.

It’s just plain a simple - nothing has changed on my computer, my DCS world or how I play - performance is woefully less on 2.5.6 v 2.5.5. Simples.

Sent from my iPhone using Tapatalk

yep concur, i get way higher CPU frame times on average than i used to on 2.5.5, whilst there is also higher GPU frametimes the GPU usage increase is staggering

I set up a CAS mission over Iran in the Hornet. Flying the mission by myself- butter smooth.

I added 1 wingman- and my performance got around halved (Locked at 40 fps). The difference is severe.

Meanwhile I can run 20 Bf 109's vs 40 B-17 and run it butter smooth.

Something pointing to AI indeed. I can upload the mission if that is useful.

I run a pretty solid rig: Gtx 1080ti, 5gz CPU, 32 gigs ram, SSD.

I set up a CAS mission over Iran in the Hornet. Flying the mission by myself- butter smooth.

 

I added 1 wingman- and my performance got around halved (Locked at 40 fps). The difference is severe.

 

Meanwhile I can run 20 Bf 109's vs 40 B-17 and run it butter smooth.

 

Something pointing to AI indeed. I can upload the mission if that is useful.

 

I run a pretty solid rig: Gtx 1080ti, 5gz CPU, 32 gigs ram, SSD.

please upload both.

Will do!

I just tested a simple scene in the editor, 20 Bf 109's on a turning point besides the airport and me starting from the runway in a christen eagle. Framerate was tanked right from the beginning, simulation value was extremely high (around 80) frametime also extremely high (around 70). When i left the area and was around 10 miles away, the frametimes started to get better but still to high to hold 45 fps (30 fps, VR with forced motion smoothing).