Jump to content

Framerate decreasing up to 16% with 2.8.0


Limaro
Go to solution Solved by Limaro,

Recommended Posts

Hi,
I want to report a general decrease of framerate by about 13%. This might be to the new cloud and lighting effects, however I think It might be useful to let you know this. Since the 2.7.6 to 2.7.7 - Performance-Loss Bug I have a small replay I regularly check before updates. These are now the differences:

2.7.18

27-10-2022, 16:05:09 DCS.exe benchmark completed, 8549 frames rendered in 101.203 s
                     Average framerate  :   84.4 FPS
                     Minimum framerate  :   56.9 FPS
                     Maximum framerate  :  109.9 FPS
                     1% low framerate   :   58.7 FPS
                     0.1% low framerate :   33.7 FPS

2.8.0

30-10-2022, 20:50:31 DCS.exe benchmark completed, 7452 frames rendered in 101.015 s
                     Average framerate  :   73.7 FPS
                     Minimum framerate  :   51.4 FPS
                     Maximum framerate  :   92.6 FPS
                     1% low framerate   :   52.8 FPS
                     0.1% low framerate :   47.7 FPS

My PC in short: Ryzen 7 3700X, AMD 6900XT (Adrenalin 22.10.2), 64GB RAM, Windows 10 21H2

Unfortunately, I can only share a actual 2.8 log file (attached). Currently I try to make space on my drive, to update the stable and -counterverify the original numbers. I will post the log for this later.

What I've done so far:

- Delete fxo
- Delete metashader2
- Removed Scripts-Folder

As soon as the stable is ready I'll try diffent settings to compare the impacts.
But meanwhile, if someone has valuable suggestions, I am open for tipps.

 

EDIT: I finally could find some place for the stable and installed it. I've created two more flight, one in very bad weather, the other without. Also changed the modules. But the results are very compareable.

30-10-2022, 22:11:55 DCS.exe benchmark completed, 5598 frames rendered in 74.672 s 2.7.18 Beirut_4_2
                     Average framerate  :   74.9 FPS
                     Minimum framerate  :   56.6 FPS
                     Maximum framerate  :   99.8 FPS
                     1% low framerate   :   56.6 FPS
                     0.1% low framerate :   55.1 FPS
30-10-2022, 22:15:30 DCS.exe benchmark completed, 6445 frames rendered in 63.859 s 2.7.18 Beirut_4_3
                     Average framerate  :  100.9 FPS
                     Minimum framerate  :   72.7 FPS
                     Maximum framerate  :  137.7 FPS
                     1% low framerate   :   72.4 FPS
                     0.1% low framerate :   70.8 FPS


30-10-2022, 22:19:27 DCS.exe benchmark completed, 4906 frames rendered in 74.469 s 2.8.0 Beirut_4_2
                     Average framerate  :   65.8 FPS
                     Minimum framerate  :   52.2 FPS
                     Maximum framerate  :   87.3 FPS
                     1% low framerate   :   51.6 FPS
                     0.1% low framerate :   47.5 FPS
30-10-2022, 22:21:04 DCS.exe benchmark completed, 5590 frames rendered in 63.438 s 2.8.0 Beirut_4_3
                     Average framerate  :   88.1 FPS
                     Minimum framerate  :   60.8 FPS
                     Maximum framerate  :  115.7 FPS
                     1% low framerate   :   60.1 FPS
                     0.1% low framerate :   58.6 FPS

EDIT2: Updated Title

EDIT3: Corrected GPU Number

dcs.log

Performance_Fly_4_1_Beirut.trk Performance_Fly_4_2_Beirut.trk Performance_Fly_4_3_Beirut.trk dcs_2.7.18.log


Edited by Limaro
  • Like 9
  • Thanks 2

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

Thank you for posting this. I had done testing with 2.7.18 vs 2.8.0 and saw similar results. Below is one of my test runs from PG with clear weather. People really need to do an actual benchmark test before claiming performance is better/worse/same. In 2.8 it's certainly worse and not related to VR.

Windows 11 22H2 (latest updates), Nvidia Driver 516.94, I9-10900K, 64G RAM, RTX 3090

30-10-2022, 12:06:21 DCS.exe benchmark completed, 6735 frames rendered in 89.234 s (Open Beta 2.8.0)
                     Average framerate  :   75.4 FPS
                     Minimum framerate  :   62.4 FPS
                     Maximum framerate  :   92.1 FPS
                     1% low framerate   :   62.7 FPS
                     0.1% low framerate :   61.7 FPS
30-10-2022, 12:09:50 DCS.exe benchmark completed, 7888 frames rendered in 92.141 s (Stable 2.7.18)
                     Average framerate  :   85.6 FPS
                     Minimum framerate  :   69.7 FPS
                     Maximum framerate  :  105.4 FPS
                     1% low framerate   :   70.1 FPS
                     0.1% low framerate :   68.8 FPS

 

 


Edited by bandit648
  • Like 6
  • Thanks 2

PC Hardware: Asus Maximus XIII Hero / i9-10900K / 64 GB DDR4 3600 MHz / NVIDIA RTX 3090 FE / Samsung 960 NVMe / LG OLED48CX / Reverb G2

Flight Controls: Thrustmaster HOTAS Warthog / VKB T-RUDDER MKIV / Cougar MFD Set

DCS Modules: F-14 Tomcat / F/A-18C Hornet / AV-8B NA / F-5E Tiger II / A-10C Warthog / F-16C Viper / UH-1 Huey / LOFC3 / Spitfire LF Mk. IX / P-51D Mustang / WWII Assets Pack / Normandy 1944 / NTTR Map / Persian Gulf Map / Syria Map

Link to comment
Share on other sites

I've made some more tests and also compared some modules and many maps, but the results show all the same. Only Channel and Nevada are less affected.

All Runs were about 80seconds long. Full data attached, here a overview of average data:

Caucasus 2.7.18: 106.7 FPS
Caucasus 2.8.00:  90.0 FPS  -15.6%

Marianas 2.7.18:  74.7 FPS
Marianas 2.8.00:  64.6 FPS  -13.5%

Nevada: 2.7.18: 103.1 FPS
Nevata: 2.8.00:  96.7 FPS   -6,2%

Persian Golf: 2.7.18: 108.8 FPS
Persian Golf: 2.8.00:  91.5 FPS -16.2%

Channel: 2.7.18: 91.4 FPS
Channel: 2.8.00: 83.6 FPS -8,5%

As a side-node: I believe this is the same thing as discussed in here:

And It feels like the same flat-rate performance loss we had last year.

I just hope, that we must not wait three patches again.

@Flappie  I count on you again to reproduce this and forwarding it to the dev-Team 😊

Benchmark.txt Performance_Fly_7_Caucasus.trk Performance_Fly_7_Channel.trk Performance_Fly_7_Marianas.trk Performance_Fly_7_Nevada.trk Performance_Fly_7_PersGolf.trk

  • Like 5
  • Thanks 2

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 changed the title to Framerate decreasing up to 16% with 2.8.0

Hi Limaro, I have not tested 2.8 performance yet, but another tester who did told me you were right aout the 13-15% decrease in performance in 2.8 initial release. I know we currently have issues with shadows and that might be playing a role. I'm pretty busy reporting fresh bugs at the moment, but I'll try to check next week and will report (unless the performance issue is fixed in between).

  • Like 4
  • Thanks 4

Don't accept indie game testing requests from friends in Discord. Ever.

Link to comment
Share on other sites

After 2.8 i lost near 30 FPS in different missions. one example is single player empty mission with no units where i could get 130 FPS in 2.7.18 but now in 2.8 i get 100 FPS. GPU usage stay at 99% so i don't have cpu bottleneck there. changing shadows from high to low had no significant effect on FPS.

Is it the same problem we had in 2.7.7 "water surface rendering effect" ?

 


Edited by Alireza.kh
  • Like 2
Link to comment
Share on other sites

I had dreadful performance with 2.8 which was virtually impossible to play.  My FPS was down to the 20's and frame time up to near 40!

After checking my 6800xt I noticed the usage was always around 95 % but temp was going over 90 degrees.  I've now set a manual fan curve which has brought the temp back to around 65-70 and frametimes back to 15 ish.

Not sure if this was conicidence but it might be worth checking.

Link to comment
Share on other sites

Might make sense in your case. But I dont understand why people try to compensate the performance drop. Even MSFS2020 starts to be playable in VR due to optimisations. And it started with really awful frame rates in VR.

Number 1 should be performance in a military sim. Also MP performance should be in second place.

And there was no worse time for the performance drop than now. Some people purchased an RTX 4090 to get higher frame rates. And a few weeks later 2.8 smashed the performance boost from the new hardware back to 2.7 or even lower.


Edited by darkman222
  • Like 1
  • Thanks 2
Link to comment
Share on other sites

so i dont know if this is my memory but there seems to be a much crisper shadow rolling around inside the canopy glass now, which i don't remember seeing before

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

5 minutes ago, speed-of-heat said:

so i dont know if this is my memory but there seems to be a much crisper shadow rolling around inside the canopy glass now, which i don't remember seeing before

Yup shadows on the glass is a new one, as are a lot of reflections within the cockpit - the shkval in the KA50 now floods dynamic green light onto the surrounding panel surfaces which flickers as the screen is slewed around.

I'm not a million miles off my 2.7 performance, getting frametimes of 12-13ms, the new flat terrain shadows occasionally cause issues though but generally it's keeping above 75fps 🙂


Edited by edmuss

Ryzen7 7800X3D / RTX3080ti / 64GB DDR5 4800 / Varjo Aero / Leap Motion / Kinect Headtracking
TM 28" Warthog Deltasim Hotas / DIY Pendular Rudders / DIY Cyclic Maglock Trimmer / DIY Abris / TM TX 599 evo wheel / TM T3PA pro / DIY 7+1+Sequential Shifter / DIY Handbrake / Cobra Clubman Seat
Shoehorned into a 43" x 43" cupboard.

Link to comment
Share on other sites

1 minute ago, edmuss said:

Yup shadows on the glass is a new one, as are a lot of reflections within the cockpit - the shkval in the KA50 now floods dynamic green light onto the surrounding panel surfaces which flickers as the screen is slewed around.

ah cool, glad it's no just me!  it s almost like dynamic cockpit lighting got left on and some turned the shadows to high ...  i guess if they have add new graphical effects that explains the perf difference,  but please can i have an off switch... 

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

Just now, speed-of-heat said:

ah cool, glad it's no just me!  it s almost like dynamic cockpit lighting got left on and some turned the shadows to high ...  i guess if they have add new graphical effects that explains the perf difference,  but please can i have an off switch... 

Yeah, the new shark cockpit looks really nice, but it would be nice to be able to toggle off some of the bells and whistles if possible.  Soft edges on flat shadows being one of them, look nice but have the same performance hit of 2.7 default shadows

Ryzen7 7800X3D / RTX3080ti / 64GB DDR5 4800 / Varjo Aero / Leap Motion / Kinect Headtracking
TM 28" Warthog Deltasim Hotas / DIY Pendular Rudders / DIY Cyclic Maglock Trimmer / DIY Abris / TM TX 599 evo wheel / TM T3PA pro / DIY 7+1+Sequential Shifter / DIY Handbrake / Cobra Clubman Seat
Shoehorned into a 43" x 43" cupboard.

Link to comment
Share on other sites

ah cool, glad it's no just me!  it s almost like dynamic cockpit lighting got left on and some turned the shadows to high ...  i guess if they have add new graphical effects that explains the perf difference,  but please can i have an off switch... 
Noticed this too, mainly in the harrier and hornet. Canopy reflections off but these are like you say more "shadows" . Also evident in the Hornet where the frame of the HUD casts a shadow on the canopy. Pretty sure it wasn't there before. And likely not helping with perfmance. An off switch would be appreciated.

12900KF | Maximus Hero Z690 | ASUS 4090 TUF OC | 64GB DDR5 5200 | DCS on 2TB NVMe | WarBRD+Warthog Stick | CM3 | TM TPR's | Varjo Aero

Link to comment
Share on other sites

Mainly for mud pushers. If you are prepared to fiddle a little and use the high shadow setting routinely you can pull about 8-10 fps back in some scenarios (maybe a few more) by reducing the shadow map size from 4096 to 3072 in the shadows.lua

This keeps the up close shadows from flickering too much and doesn’t have such a defined  cut off at altitude. 
 

channel and perhaps Mariana’s  can recover about 12 fps on my rig by reducing the scenery detail to 0.75 with no noticeable effect on graphics. Doesn’t work for other maps. 

  • Like 3

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 month later...

Update for 2.8.1: Still the same performance drop compared to 2.7.18

Example Beitut 2 -Click from first post:

2.7.18: 74.9 FPS
2.8.0: 65.8 FPS
2.8.1: 66.4 FPS

  • Like 4

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 just tested and have to agree that there is a massive performance impact in VR especially the CPU frametimes suffer:

F18 caucasus ready on the ramp:

2.7.18: CPU 12-14ms

2.8.1 CPU 14-18 ms

and this on my machine 😱 5900x / 4090. Just switched back to stable. It was smooth and also looks good. The whole mess began with 2.8 and dynamic weather.

Where does the up to 4 ms CPU frametime hit come.

I don't think, that just some have a performance loss in VR, I think that only some check their frametimes and only this folks are crying out loud now. The others who don't measure the frametimes are quiet

There is a massive loss of FPS on same settings, and this does not make me happy ED!

Insert an optional button for dynamic weather or do something to bring back some FPS. There is a promise to bring a VR performance increase since 4 years, but it never hit the ground. This is very frustrating! I mean a 4090 and a 5900X should be fine to play 45 FPS under all circumstances. You read right, i don't even demand 90 FPS, I am fine with smooth 45 under all circumstances. But not even this is possible.

Example: Sedlos Eastern Friendship

under 2.7: smooth

under 2.8.1: stuttering in the ground attack in the first mission

 

Do you recognize something? Smooth under 2.7 - not smooth under 2.8.1. Something is messed up in 2.8.1!!!!!

  • Like 2

SYS: MSI MPG X570 Gaming Edge Wifi || Ryzen 5900x || Gainward RTX4090 || 2x16 Gb Crucial Ballistix RGB 3200@3800 || XPG Core Reactor 850 Watt PSU || Kingston Fury 2 TB NVME SSD || WD SN850 1TB NVME || 1 x 500 GB Crucial MX300 SATA SSD || 2 x HDD 3TB || Thrustmaster F-16 & F-18 Stick on Virpil War BRD Bases || WinWing Orion2 F-16EX Viper Throttle Combo || WinWing Orion2 F/A-18 Hornet Throttle Combo (With Finger Lift) || WinWing Takeoff Panel II MFG Crosswind Pedals w. Damper || 3 x Thrustmaster Cougar MFD || Multipurpose UFC ||  Wheel Stand Pro II

 VR: HP Reverb G2

Link to comment
Share on other sites

The thing is, if there is features that will cause a drop of CPU frame time, which is already a very very valuable good, there should be simply an option to deactivate.

Moving clouds, okay, needs to be synced between clients.

But new lighting, or shadows does not need to be in sync between clients. So if there is a weak or a VR client, let this person just switch the eye candy off , and the other client playing in 2D with a strong machine, he can keep the eye candy on.

  • Like 2
Link to comment
Share on other sites

  • 2 months later...

Warming up this thread with 2.8.3.37556 (Multi Threading Update) with some news:

1. I got a new CPU (5800X3D) some time ago. My first task in DCS was to remeasure the FPS from the very first post of me - and the results were identical. This did not surprise me, as they are very light missions with no CPU load. Thats Why I did not mention it since. I mention in now because I measured again with 2.8.3, once with multi threading once without.

2. I've measured now following data for

Beirut 4_2:

2.7.18:    74.9 FPS (lower CPU)
2.8.0:     65.8 FPS
2.8.1:     66.4 FPS

2.8.3:     68.0 FPS
(got 2 FPS back)
2.8.3 MT   73.8 FPS (almost back to pre 2.8)

Beirut 4_3:

2.7.18:   100.9 FPS (lower CPU)
2.8.0:     88.1 FPS

2.8.3:     92.3 FPS
  (got 4 FPS back)
2.8.3 MT: 103.4 FPS (more than 2.7.18)

A surprising result for me. My mindset was, that multi threading would have no effect on my scenarios. They are located near the city of Beirut (large) but only my airplane in single player. Not any other AI aircraft or ground unit.

Dispite the good MT results, I'd still like to not mark here anything as resolved. Just imagine performance of 2.7.18 having multi threading 🤤.

Anyway, ED, thanks for the improvements and keep up the good work 💪

 

EDIT:

I could reproduce these results with my second PC. Spec is Intel i5-4670, 32GB DDR3, Geforce 980 (4GB)

Caucasus Scenario:
2.7.18:   66.4 FPS
2.8.3 ST: 57.4 FPS
2.8.3 MT: 68.0 FPS

Marianas Scenario:
2.7.18:   44.2 FPS
2.8.3 ST: 42.6 FPS
2.8.3 MT: 49.5 FPS

dcs.log

PC2 dcs.log_2.7.18.old PC2 dcs.log


Edited by Limaro
typo
  • Like 2

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

  • 1 month later...

Update with 2.8.4

Beirut 4_2:

2.7.18     74.9 FPS
2.8.0      65.8 FPS
2.8.1      66.4 FPS
2.8.3 ST   68.0 FPS

2.8.3 MT   73.8 FPS

2.8.4 ST   65.4 FPS
2.8.4 MT   70.3 FPS

Beirut 4_3:

2.7.18    100.9 FPS
2.8.0      88.1 FPS
2.8.3 ST   92.3 FPS

2.8.3 MT: 103.4 FPS

2.8.4 ST   89.3 FPS
2.8.4 MT  102.1 FPS

Looks like we lost a bit of performance.

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

  • 2 weeks later...

Atleast its not just me, So my findings in general (No numbers)

 

After the update 2 to 3 hours of play and my fps halfs from 60 to 20fps, I thought it was just the new Normandy map buts its happening in PG to, Does not matter on the population of the server or its unit count.

Link to comment
Share on other sites

Not sure if your issue is related. But here is a thread I started about performance degradation over time:

 

And also there is a thread that identified the issue to come from craters that never disappear and add on CPU frame time. This issue is claimed to be resoved in the next update. Lets see if it fixes your and my issues too:

 

Link to comment
Share on other sites

  • 1 month later...

Update with 2.8.6. Meanwhile AMD Adrenalin 23.5.2.
Seams a small drop again, but this can be due to measurement inaccuracy.

Beirut 4_2:

2.7.18     74.9 FPS
2.8.0      65.8 FPS
2.8.1      66.4 FPS
2.8.3 ST   68.0 FPS

2.8.3 MT   73.8 FPS
2.8.4 ST   65.4 FPS
2.8.4 MT   70.3 FPS
2.8.6 ST   65.9 FPS
2.8.6 MT   71.2 FPS

Beirut 4_3:

2.7.18    100.9 FPS
2.8.0      88.1 FPS
2.8.3 ST   92.3 FPS

2.8.3 MT: 103.4 FPS
2.8.4 ST   89.3 FPS
2.8.4 MT  102.1 FPS
2.8.6 ST   88.6 FPS
2.8.6 MT   99.9 FPS

Still performance is not back on 2.7.18 level comparing single thread versions. Hope when we get back this, MT will be even greater 🙂

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

  • Recently Browsing   0 members

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