Jump to content

Performance (FPS) loss from 2.7.6 to 2.7.7


Limaro
Go to solution Solved by Limaro,

Recommended Posts

Just now, speed-of-heat said:

I always recommend turning off asw/motion smoothing to measure performance for the same reason you recommend turning off vsync

You will understand the data impalor provided much better than I. From what limited understanding I have of the VR world however it appears to me his results are affected by such witchcraft. 

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

1 hour ago, Boosterdog said:

From what I can see from the data, and my very limited (read zero) understanding of VR, your system is getting cained pre and post 2.7.7/8.  Steam VR is doing its "magic" and reprojecting- a lot- to present even the minimum 1/2 rate of 45 fps to you. Not sure if you change the repro manually or its automatic, I assume its automatic.

If thats correct, its the VR equivalent of me and my flatscreen with vstink running say 55 fps in 2.7.6 and 40 fps in 2.7.7/8.  In both cases the vsync software will change down a gear to half refresh rate (30 fps) because my system cant produce the 60 fps vsync requires just as yours it half stepping to the 45 fps.- as such my perceived performance will not have changed in the same way yours hasnt. I'd be getting 30 fps like i always did. In reality however the loss is there, I just cant see it cos Im already at a low state and I wont see until I cant make 30 fps.  Difference here is that Stream VR has further tricks up it sleeve in the shape of reprojection to help keep things bearable. 

What Im saying is that you cant make a meaningful comparison with fps alone in such circumstances as there there are more factors, external to DCS, at play. As an Analogy, you cant tell a Fiat Punto (2.7.8) has an overall slower top speed than an Ferarri (2.7.6) by driving them both at a speed limit of 45 mph (half rate) with cruise control (Steam VR) on. 

Thats why we switch off vsync when comparing. In your case I think you are in that ballpark where the perfomance difference is "soaked up" by an already choked output. If you pushed up your settings so stream VR algorithms couldnt cope or were previously on the cusp, I'd venture you'd experience a difference. 

Forgive if Im wrong but I suspect your set up is masking the lost performance at this time. As it probably is for many others and likely the reason why "where the F has my performance gone" posts tend to drip into the forums and not tsunami as each player hits that magic mark in their own time.  

 

If you scroll down on the Config page you will see that I fly with no reprojection and at 90. My max FPS in these benches was higher than 45, and on lighter missions it goes to 60 or so, so there is no capping at 45. 

 

But I cannot know for sure. Maybe SVR or WMR indeed make 45 a 'sticky' point even with reprojection turned off, to make my experience smoother. I can run again on flat screen 1080p and with no FPS cap to make you happy!


Edited by impalor

12900KF@5.4, 32GB DDR4@4000cl14g1, 4090, M.2, W10 Pro, Warthog HOTAS, ButtKicker, Reverb G2/OpenXR

Link to comment
Share on other sites

45 minutes ago, impalor said:

If you scroll down on the Config page you will see that I fly with no reprojection and at 90. My max FPS in these benches was higher than 45, and on lighter missions it goes to 60 or so, so there is no capping at 45. 

 

But I cannot know for sure. Maybe SVR or WMR indeed make 45 a 'sticky' point even with reprojection turned off, to make my experience smoother. I can run again on flat screen 1080p and with no FPS cap to make you happy!

 

As I say I do not know the VR software or how it works. From the FPS VR1.21 charts you linked I saw that Reprojection was noted at somewhere between 36% and 59% in the tests and so assumed that reprojection is a factor. Not sure what you mean by your config but, as Im not VR savvy, I dont know if what your are setting in one place is being overridden by another program (Steam VR?) There was also a figure in there of 45 (second line down) which appears to be a target FPS. Overall my limited knoweldge is my limiting factor here. The differening lengths of the runs from build to build also add to my confusion. Whithout any description of methodology its all assumtion on my part.

At the end of the day, you have nothing to prove to me. If you feel your tests are accurate and represent a true comparison of the builds without external interference its all good and goes to show not everyone is affected. 

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

@BoosterdogThe tests are accurate to the best of my ability. All ran exactly 60 seconds as the Pause button was used to start FRAPS. Then FRAPS stopped automatically in 60 seconds. I ran additional tests when the first two results were noisy, i.e. much different. I compared averages with averages in the end. Only Release/HDD 2.7.6 vs 2.7.7 makes sense to compare, as OB 2.7.8 runs from NVMe on my PC.

 

Now I have added more results for 1080p (same link), and between 2.7.6 and 2.7.7 I see a similar %difference as for VR. The 2.7.7 run does show substantially more CPU utilization, and uses somewhat more RAM and GPU.

 

Also, when launching DCS from HDD (Disk 1 or S on my screenshots), it mattered WHEN I pressed the Pause key to start the test. The longer I waited, the smoother was the run (no stutters) as DCS could preload more data upfront, resulting in higher average FPS. 

@BIGNEWY Maybe these benchmarks can help your devs to explain us the 2.7.6 vs 2.7.7 performance? It seems the textures are heavier or something like that?


Edited by impalor
  • Thanks 2

12900KF@5.4, 32GB DDR4@4000cl14g1, 4090, M.2, W10 Pro, Warthog HOTAS, ButtKicker, Reverb G2/OpenXR

Link to comment
Share on other sites

18 minutes ago, impalor said:

Only Release/HDD 2.7.6 vs 2.7.7 makes sense to compare, as OB 2.7.8 runs from NVMe on my PC.

No it's not, it doesn't really matter where it's located. There should be no difference when located on HD or SSD (or NVMe for that matter).
It will only load faster from SSD and that count also from stuff thats need to be sideloaded when a mission is already running (which is only minor). It should not have influence on overall FPS. There is also no need to run a very complex mission to see difference in FPS delivery, just fly around alone is more then enough (in that case there is no sideload at all...).

Win11 Pro 64-bit, Ryzen 5800X3D, Corsair H115i, Gigabyte X570S UD, EVGA 3080Ti XC3 Ultra 12GB, 64 GB DDR4 G.Skill 3600. Monitors: LG 27GL850-B27 2560x1440 + Samsung SyncMaster 2443 1920x1200, HOTAS: Warthog with Virpil WarBRD base, MFG Crosswind combat pedals, TrackIR4, Rift-S.

Personal Wish List: A6 Intruder, Vietnam theater, decent ATC module, better VR performance!

Link to comment
Share on other sites

28 minutes ago, impalor said:

@BoosterdogThe tests are accurate to the best of my ability. All ran exactly 60 seconds as the Pause button was used to start FRAPS. Then FRAPS stopped automatically in 60 seconds. I ran additional tests when the first two results were noisy, i.e. much different. I compared averages with averages in the end. Only Release/HDD 2.7.6 and 2.7.7 makes sense to compare, as OB 2.7.8 runs from NVMe on my PC.

 

Now I have added more results for 1080p (same link), and between 2.7.6 and 2.7.7 I see a similar %difference as for VR. The 2.7.7 run does show substantially more CPU utilization, and uses somewhat more RAM and GPU.

 

Also, when launching DCS from HDD (Disk 1 or S on my screenshots), it mattered WHEN I pressed the Pause key to start the test. The longer I waited, the smoother was the run (no stutters) as DCS could preload more data upfront, resulting in higher average FPS. 

@BIGNEWY Maybe these benchmarks can help your devs to explain us the 2.7.6 vs 2.7.7 performance? It seems the textures are heavier or something like that?

This is really nice work and thorough test results. I was really curious why you have nearlly no frame drop.
so I've tried your exact settings and run your track and these are the results:

22-11-2021, 14:30:08 DCS.exe (Stable 2.7.6) benchmark completed, 7342 frames rendered in 65.515 s
                     Average framerate  :  112.0 FPS
                     Minimum framerate  :   87.9 FPS
                     Maximum framerate  :  138.8 FPS
                     1% low framerate   :   25.5 FPS
                     0.1% low framerate :    3.7 FPS
22-11-2021, 14:34:27 DCS.exe (Open Beta 2.7.8) benchmark completed, 7354 frames rendered in 65.985 s
                     Average framerate  :  111.4 FPS
                     Minimum framerate  :   89.0 FPS
                     Maximum framerate  :  136.8 FPS
                     1% low framerate   :   39.9 FPS
                     0.1% low framerate :    3.7 FPS

The averate framerate just drop by .6 FPS!!
It's the first time, I've seen nearly the same data in both versions.

I will repeat the tests again , to exclude the propability that I missed something.

Windows 11 Pro 64Bit - AMD Ryzen 5800X3D - AMD 6900 XT - 64 GB RAM

VPC MongoosT-50CM2 Throttle, VPC WarBRD Base, VPC MongoosT-50CM2 Grip

Link to comment
Share on other sites

Mission I made in east Syria west Iraq,  with a cloud preset and flying the f18 my gpu usage was in mid 90’s. And would drop below my ASW 36 in quest 2,  so I went in and got rid of the clouds so no cloud preset and set at 0.  Flying in f18 gpu usage was down in 60 percent usage up to mid 70 percent.  Frame rate locked at ASW 36 whole time.  In 2.7.6 the clouds didn’t hammer my 1080ti like this.

   Heck even the hind fly’s smooth now in this mission with no clouds.  Granted Hind at times is up to 20 percent higher gpu usage then f18.  Figure I’d share from my mission testing last night.  
 

Another mission to test is f18 free flight Syria,  in 2.7.6 ran smooth as butter gpu usuage in 70’s with clouds,  now it’s mid 90’s and drops below 36 ASW.  Will try removing all clouds and see results.

Intel 8700k @5ghz, 32gb ram, 1080ti, Rift S

Link to comment
Share on other sites

Just ran impalor's test track, played the Pause game.

Results for me are the same as before, quite a drop in OB performance, both run from the same SSD. Shots +/- taken at the same moment, when the F-18 dives under the clouds and levels out on the coast line. But the difference is there throughout the entire track. Pancake mode.

Stable 2.7.6                                                                      Open Beta 2.7.8

image.png     image.png

One thing: when running someone else's track, does it run with my settings or with the creators settings?
If it runs with my settings, i would like to retest with impalor's settings but i can't find these (the google drive link doesn't show these or do i miss something?).


Edited by Lange_666

Win11 Pro 64-bit, Ryzen 5800X3D, Corsair H115i, Gigabyte X570S UD, EVGA 3080Ti XC3 Ultra 12GB, 64 GB DDR4 G.Skill 3600. Monitors: LG 27GL850-B27 2560x1440 + Samsung SyncMaster 2443 1920x1200, HOTAS: Warthog with Virpil WarBRD base, MFG Crosswind combat pedals, TrackIR4, Rift-S.

Personal Wish List: A6 Intruder, Vietnam theater, decent ATC module, better VR performance!

Link to comment
Share on other sites

1 hour ago, Lange_666 said:

No it's not, it doesn't really matter where it's located. There should be no difference when located on HD or SSD (or NVMe for that matter).
It will only load faster from SSD and that count also from stuff thats need to be sideloaded when a mission is already running (which is only minor). It should not have influence on overall FPS. There is also no need to run a very complex mission to see difference in FPS delivery, just fly around alone is more then enough (in that case there is no sideload at all...).

My terrain preload radius is pretty low, because I have only 16gb of RAM for VR in MP. DCS keeps loading terrain inflight resulting in stutters (0 FPS moments) when on HDD. On SSD also. Only NVMe does it quickly enough. My pagefile is also there. I modified my BIOS to enable it for 4670k, but not as a system drive.

36 minutes ago, Lange_666 said:

Just ran impalor's test track, played the Pause game.

Results for me are the same as before, quite a drop in OB performance, both run from the same SSD. Shots +/- taken at the same moment, when the F-18 dives under the clouds and levels out on the coast line. But the difference is there throughout the entire track. Pancake mode.

Stable 2.7.6                                                                      Open Beta 2.7.8

image.png     image.png

One thing: when running someone else's track, does it run with my settings or with the creators settings?
If it runs with my settings, i would like to retest with impalor's settings but i can't find these (the google drive link doesn't show these or do i miss something?).

 

I included my settings as Preset 1. Replace your Custom1.lua in Saved Games with mine, then load in DCS.

Edit. Sorry! Just updated the preset, because it was originally for VR and not with 1080 resolution!


Edited by impalor
  • Like 1

12900KF@5.4, 32GB DDR4@4000cl14g1, 4090, M.2, W10 Pro, Warthog HOTAS, ButtKicker, Reverb G2/OpenXR

Link to comment
Share on other sites

@impalor @Limaro   difference remains for me with or with out the custom settings (changed the screen res to my native 1080P) and ran the straight free flight mission

2.7.6 Frames: 7559 - Time: 60000ms - Avg: 125.983 - Min: 49 - Max: 169

2.7.8 Frames: 7003 - Time: 60000ms - Avg: 116.717 - Min: 112 - Max: 120 

 

The min FR differences can be accounted for having to alt tab in the old stable to get fraps to register the game screen

There is less of a difference but equally less stain on the GPU which I could hear with the lower fan noise. I did observe odd behaviour with Fraps in that, without alt/Enter being used it would fixate on a figure and stick with it regrdless of what was going on.  This produced very similar results if not altered but these where in the low 60s and not the higher frames seen elswhere so i duount its a factor in anyone elses figures. 

 

I then went and took the same settings into the Mi24 weapons range mission

2.7.6 Frames: 8091 - Time: 60000ms - Avg: 134.850 - Min: 134 - Max: 136

2.7.8 Frames: 6119 - Time: 60000ms - Avg: 101.983 - Min: 100 - Max: 103

 

Go figure. 😞

 

 

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

Same here, running with impalor's settings copied over into my options.lua (i normally run a 2560x1440 screen) i still have the same performance difference between Stable 2.7.6 and OB 2.7.8. So with his settings on my monitor i now ran the test in sort of a small window.

At first there was no difference but the FPS didn't fluctuated anymore, not even a little bit, 119FPS for both. Then it occurred to me that G-Sync played a roll and with the monitor refresh set to 120 Hz it was spot on that G-Sync played a limiting role, limiting the FPS in DCS to 119.

Next step was to disable G-Sync and put the monitors refresh rate back to default which is 144 Hz (i set this to 120Hz because of TrackIR).
When this was done, FPS could go unlimlited (above the clouds in Stable it hit 200 briefly).

Result (don't look at the CPU 1-4 values, they jump up and down all the time):

Stable 2.7.6                                                                Open Beta 2.7.8

image.png  image.png


Edited by Lange_666
  • Like 1

Win11 Pro 64-bit, Ryzen 5800X3D, Corsair H115i, Gigabyte X570S UD, EVGA 3080Ti XC3 Ultra 12GB, 64 GB DDR4 G.Skill 3600. Monitors: LG 27GL850-B27 2560x1440 + Samsung SyncMaster 2443 1920x1200, HOTAS: Warthog with Virpil WarBRD base, MFG Crosswind combat pedals, TrackIR4, Rift-S.

Personal Wish List: A6 Intruder, Vietnam theater, decent ATC module, better VR performance!

Link to comment
Share on other sites

This time measured with FRAPS (couldn't get the FPS Min,Max, etc not to show on MSI Afterburner for some reason):

On my settings:

Stable 2.7.6:    Frames: 5917 - Time: 60000ms - Avg: 98.617 - Min: 87 - Max: 108

OB 2.7.8:          Frames: 5154 - Time: 60000ms - Avg: 85.900 - Min: 74 - Max: 93

 

Win11 Pro 64-bit, Ryzen 5800X3D, Corsair H115i, Gigabyte X570S UD, EVGA 3080Ti XC3 Ultra 12GB, 64 GB DDR4 G.Skill 3600. Monitors: LG 27GL850-B27 2560x1440 + Samsung SyncMaster 2443 1920x1200, HOTAS: Warthog with Virpil WarBRD base, MFG Crosswind combat pedals, TrackIR4, Rift-S.

Personal Wish List: A6 Intruder, Vietnam theater, decent ATC module, better VR performance!

Link to comment
Share on other sites

42 minutes ago, Lange_666 said:

Same here, running with impalor's settings copied over into my options.lua (i normally run a 2560x1440 screen) i still have the same performance difference between Stable 2.7.6 and OB 2.7.8. So with his settings on my monitor i now ran the test in sort of a small window.

At first there was no difference but the FPS didn't fluctuated anymore, not even a little bit, 119FPS for both. Then it occurred to me that G-Sync played a roll and with the monitor refresh set to 120 Hz it was spot on that G-Sync played a limiting role, limiting the FPS in DCS to 119.

Next step was to disable G-Sync and put the monitors refresh rate back to default which is 144 Hz (i set this to 120Hz because of TrackIR).
When this was done, FPS could go unlimlited (above the clouds in Stable it hit 200 briefly).

Result (don't look at the CPU 1-4 values, they jump up and down all the time):

Stable 2.7.6                                                                Open Beta 2.7.8

image.png  image.png

 

I dont think I need to fear capping my frames at 120 or even 60 for that matter!

TBH, Im pretty dejected with the whole thing. There is a problem. 2.7 eats FPS like Elvis on a toilet but noone seems to care. Less so now -  even the ED mods havent made an appearance this time around to grunt exasperatedly at us daring not to accept that this particular parrot has been snuffed out under  a "reported" sign.

I put aside £800 this month to go towards a new GPU at christmas - its now turned into a Deering Banjo. Next months will likely turn into another Tele (the number of telecasters you need is precisely one more telecaster than you have - fact). The Apache my missues was buying me as a birthday gift has turned into a snazzy strap and some strings. All decisions I may come to regret when Im asked to play "that Burt Reynolds song" for the nth time or when multicore hits but there's an increasng trend towards jam tomorrow at present which i frankly dont have the patience for or the faith in.


Edited by Boosterdog
  • Like 1

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

Just for shitsngiggles, I went in IL-2 and turned on 4k textures(thinking that this may be the problem with DCS of maybe increasing textures and me not knowing it)...and to my surprise it made no difference. Solid (external view) 60 fps (50 in the pit)with AA on. So I just don't get it why I can get twice the performance with twice the eye candy turned on. What makes il-2 so performance friendly on my older system compared to DCS?? 

MODUALS OWNED       AH-64D APACHE, Ka-50, UH-1H, Mi-8MTV2, Mi-24,Gazelle, FC3, A-10C, A-10CII, Mirage 2000C, F-14 TOMCAT, F/A-18C HORNET, F-16C VIPER, AV-8B/NA, F-15 E, F-4 Phantom, MiG-21Bis, L-39, F-5E, AJS 37 Viggen, MiG-19, F-86, MiG-15Bis, Spitfire IX, Bf-109K, Fw-190D, P-51D, CA, SYRIA, NEVADA, NORMANDY, PERSIAN GULF, MARIANA ISLANDS,SUPER CARRIER, WORLD WAR II ASSETS PACK, HAWK T1

SYSTEM SPECS            AMD  7600X 4.7 Ghz CPU , MSI RX 6750 12 gig GPU ,32 gig ram on Win11 64bit.

 

Link to comment
Share on other sites

I run some additional tests, comparing settings, resolution and track files.

this is the outcome:

impalor track, original settings from impalor (1080p Settings)
2.7.6    Average framerate  :  112.0 FPS
2.7.8    Average framerate  :  111.4 FPS
                    
impalor track, 1440p Settings impalor
2.7.6    Average framerate  :  109.6 FPS
2.7.8    Average framerate  :  108.1 FPS
 
impalor track, Own Settinges @ 1440p
2.7.6    Average framerate  :   86.1 FPS
2.7.8    Average framerate  :   84.8 FPS

impalor track, Own Settinges @ 1400p Only recorded Under Clouds
2.7.6    Average framerate  :   76.3 FPS
2.7.8    Average framerate  :   74.3 FPS

My original track, Own settings 1440p
2.7.6    Average framerate  :  106.2 FPS
2.7.8    Average framerate  :  101.2 FPS

My original track, Own settings 4k
2.7.6    Average framerate  :   84.9 FPS
2.7.8    Average framerate  :   75.8 FPS
                    
impalor track, Own settings 4k
2.7.6    Average framerate  :   79.7 FPS
2.7.8    Average framerate  :   73.3 FPS
                    
impalor track, Own settings 1080p
2.7.6    Average framerate  :   87.7 FPS
2.7.8    Average framerate  :   86.6 FPS
                    
My original Track, Impalor settings:
2.7.6    Average framerate  :  116.0 FPS
2.7.8    Average framerate  :  115.6 FPS

 

Espacially last run, of my track and @impalor settings, shows, that somehow it is settings related. However, I also assume that resolution has special impact comparing 2.7.6 and 2.7.8.

But also the other tackfile shows less differences in general, comparing the versions... this is all odd

EDIT: my settings are on my initial post.


Edited by Limaro

Windows 11 Pro 64Bit - AMD Ryzen 5800X3D - AMD 6900 XT - 64 GB RAM

VPC MongoosT-50CM2 Throttle, VPC WarBRD Base, VPC MongoosT-50CM2 Grip

Link to comment
Share on other sites

I think the reason is that my settings were found by trial and error to push my PC for maximum utilization of both its CPU and GPU, given RAM and VRAM that I have, in scenarios where I fly most. Applying these settings to beefier PCs or ones with different CPU/GPU balance gives different results. But we all see that higher versions demand higher hardware utilization, and depending on where bottlecks are, can punish FPS more or less.

12900KF@5.4, 32GB DDR4@4000cl14g1, 4090, M.2, W10 Pro, Warthog HOTAS, ButtKicker, Reverb G2/OpenXR

Link to comment
Share on other sites

One more thing that differs from PC to PC is RAM speed and latency. My earlier tests showed that DCS is very sensitive to it, and maybe even more now (2.7.8) than before (2.7.6). 


Edited by impalor

12900KF@5.4, 32GB DDR4@4000cl14g1, 4090, M.2, W10 Pro, Warthog HOTAS, ButtKicker, Reverb G2/OpenXR

Link to comment
Share on other sites

33 minutes ago, impalor said:

One more thing that differs from PC to PC is RAM speed and latency. My earlier tests showed that DCS is very sensitive to it, and maybe even more now (2.7.8) than before (2.7.6). 

 

My PC was changed - from an I5 6600K running meh Corsair 3000hz LPX 4X8GB to stupidly expensive 2X16GB Sammy B Die 3600 16 16 16 32 dual stack stuff all tied up to Ryzen's infinate whatnot - if it has an effect it, along with the undervolted 5600X,reduces the lows in heavy scenes but it doesnt alter the highs save for a few frames. That was last month's £800.


Edited by Boosterdog

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

1 hour ago, Limaro said:

shows, that somehow it is settings related. 

Being very unscientific, my view is that whatever DCS used to cain the CPU with has gone home and fetched its baseball bat. Everything takes just that little bit more. I dont think its settings, or a golden ratio. I cant spend any more time on it.  Its something we cant effect in a GUI in any meaningful way and the CPU use is certainly different. I havent seen the matching effect remember. I still had the differential throughout. One that actually increased.

Only ED knows what got changed under the hood and they dont seemed to be that bothered to find out. To be honest, I cant say I blame them. There are more people spouting off at Heatblur for missing a Viggen update and not painting a boat to their liking than there are complaining about performance loss. If you have to drop a plate now and then pick one with the least food on it. 

 

 

  • Like 2

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

22 hours ago, Boosterdog said:

Being very unscientific, my view is that whatever DCS used to cain the CPU with has gone home and fetched its baseball bat. Everything takes just that little bit more. I dont think its settings, or a golden ratio. I cant spend any more time on it.  Its something we cant effect in a GUI in any meaningful way and the CPU use is certainly different. I havent seen the matching effect remember. I still had the differential throughout. One that actually increased.

Only ED knows what got changed under the hood and they dont seemed to be that bothered to find out. To be honest, I cant say I blame them. There are more people spouting off at Heatblur for missing a Viggen update and not painting a boat to their liking than there are complaining about performance loss. If you have to drop a plate now and then pick one with the least food on it. 

 

 

I still cannot believe they dont care. But priorization is always a tricky thing. that's why I asked in this thread again everyone to follow the advice from ED, to make a own thread, to allow better tracking of the different aspects.

I also made some hopefully last tests, but I think they point on the issue, in one or another way. I truely believe that most damage caused something with shadows and resolution.
Here is my last data:

- I run everything with settings from @impalor (mostly everything is set to low).
- Then i've changed specific settings
- Run both versions (stable 2.7.6 and OB 2.7.8)
- I calculated the total loss compared to the "original run"
- I made the tests in the order listed here. The text in bold is the changed setting.

And there are two very outstanding impacs:

                  
My original Track, Impalor settings (Origial Run to compare settings changes with)
2.7.6    Average framerate  :  116.0 FPS
2.7.8    Average framerate  :  115.6 FPS    
> 2.7.6 vs. 2.7.8: -0,34%

My Track, Impalor Settings, + Low Shadows (previously off)
2.7.6    Average framerate  :  103.0 FPS -11,2%
2.7.6    Average framerate  :   97.8 FPS -15,4% 
> 2.7.6 vs. 2.7.8: -5,0%

My Track, Impalor Settings, + Low Shadows + Terrain High, Shadow Terrain Default, Global Illu - Performs better than the previous, I guess because of terrain shadow. I mentioned this effect in another thread)
2.7.6    Average framerate  :  103.4 FPS -10,9%
2.7.8    Average framerate  :  100.8 FPS -12,8% 
> 2.7.6 vs. 2.7.8: -2,5%

My Track, Impalor Settings, + Low Shadows + Terrain High, Shadow Terrain Default, Global Illu - @ 4k
2.7.6    Average framerate  :   83.9 FPS -27,7%
2.7.8    Average framerate  :   74.5 FPS -35,6% 
> 2.7.6 vs. 2.7.8: -11,2%


Edited by Limaro
  • Like 2
  • Thanks 1

Windows 11 Pro 64Bit - AMD Ryzen 5800X3D - AMD 6900 XT - 64 GB RAM

VPC MongoosT-50CM2 Throttle, VPC WarBRD Base, VPC MongoosT-50CM2 Grip

Link to comment
Share on other sites

@Limaro your perseverence is a credit to you. There was some mention of 2.7.7> locking out the the flat shadows remove mod from accesing the shaders. If something has changed there that affects the other shadows perhaps it would make sense. As I say , went home and got it its baseball bat. Either way, Ill enjoy what I can in DCS but no more investment in hardware or product at this time.

  • Like 5

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

FPS LOST yes, but when refly a mission...

Hi,

I have big big FPS lost every time I do this : fly the missoin 11 of UH1 World appart. Fly 75%, so I do not achieve the mission. I glue at the stage where I should shoot down infantery near the group ALPHA.

So I click "Re-fly" and then... oh oh... same place FPS get down from a nice and stable 60 (limited to) to a bad 6-8 FPS... It happens shortly after I fly into the left valley and the copi calls Berlin Control.

If I want to retrieve the normal FPS I have to reboot the computer. And it works well again.

So if ED will repreduce it, I thing it's a way to start with.

If it helps...

Cheers

Dan

Link to comment
Share on other sites

On 11/22/2021 at 9:39 PM, Boosterdog said:

There are more people spouting off at Heatblur for missing a Viggen update and not painting a boat to their liking than there are complaining about performance loss. If you have to 

Truer words were never spoken. With the exception that I would replace Heatblur with Eagle Dynamics (since Heatblur has nothing to do with core engine developments). 

Specs: Win10, i5-13600KF, 32GB DDR4 RAM 3200XMP, 1 TB M2 NVMe SSD, KFA2 RTX3090, VR G2 Headset, Warthog Throttle+Saitek Pedals+MSFFB2  Joystick. 

Link to comment
Share on other sites

I honestly believe that ED do care and that they are taking it seriously.  But this is a tough engineering problem to actually resolve, especially as it does not seem to impact everyone in the same way... the fact that others have solved it is functionally irrelevant. Ultimately it comes down to this if the devs can't repo it reliably its almost impossible for them to fix.  

Please don't misunderstand me I see the frametime loss too, but even if it were "simple" to fix, I dont think a few weeks would be likely to get a sensible engineering response.


Edited by speed-of-heat

SYSTEM SPECS: Hardware Intel Corei7-12700KF @ 5.1/5.3p & 3.8e GHz, 64Gb RAM, 4090 FE, Dell S2716DG, Virpil T50CM3 Throttle, WinWIng Orion 2 & F-16EX + MFG Crosswinds V2, Varjo Aero
SOFTWARE: Microsoft Windows 11, VoiceAttack & VAICOM PRO

1569924735_WildcardsBadgerFAASig.jpg.dbb8c2a337e37c2bfb12855f86d70fd5.jpg

Link to comment
Share on other sites

32 minutes ago, speed-of-heat said:

I honestly believe that ED do care and that they are taking it seriously.  But this is a tough engineering problem to actually resolve, especially as it does not seem to impact everyone in the same way... the fact that others have solved it is functionally irrelevant. Ultimately it comes down to this if the devs can't repo it reliably its almost impossible for them to fix.  

Please don't misunderstand me I see the frametime loss too, but even if it were "simple" to fix, I dont think a few weeks would be likely to get a sensible engineering response.

 

I agree in the main. I think my words were overly harsh and should have been more "not able at this point". If you have a finite resource you need to focus it where the overall difference matters. I get that. These boards are not jammed full of people screaming about performance loss so this particular issue must sit in line. However, that this issue has not been replicated within ED, I just find hard to understand (and yes this is based soley on the fact that Ive scored 3/3 in replication over 3 machines). It appears harder to replicate impalors results that it does the loss and, frankly I prefer DCS not look like Minecraft.

The best shadows I can run with the minimal amount of jaggles are an essential for me and there is a bar within my settings, albeit not what I consider to be a very high one (2X MSAA, Default terrain shadows, Medium Shadows, Textures and view distance, High ground textures, Medium Water, no civilian traffic, Forest 85%, everything else off) that DCS should be (and was pre 2.7.7) acheiving and now isnt at 1080P on my rig.  Immersion can be aheived without VR if the 2D scene is immersive enough. 2.7.7 fails massively in that respect for me for seemingly no graphical or technical gain and its not something I can redress without outlaying the best part of £1500 at this time (having already spent £800 going Ryzen).

I will take BIGNEWY's advice. I will wait for MC and Vulkan as I have no choice. 

 

 

MSI Tomahawk X570 Mobo, Ryzen 5600X undervolted on Artic Freezer E34 Cooler, RTX3080 FE, 32GB (2x16GB Dual Ranked) GSkil 3600 CL16 Trident Neo RAM, 2X 4th Gen M2 SSDs, Corsair RM850x PSU, Lancool 215 Case. 

Gear: MFG Crosswinds, Warthog Throttle, Virpil T50CM gen 1 stick, TIR5, Cougar MFD (OOA), D-link H7/B powered USB 2.0 Hub all strapped to a butchered Wheel stand pro, Cushion to bang head on, wall to scream at.  

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...