Jump to content

2.8.1 still has serious issues with terrain objects shadows


some1

Recommended Posts

@some1 What seems strange about your results is that you don’t see a performance difference in 2.8.1 between Object Shadows Default and Off (you even gained a minuscule 2/131 fps) despite the increase in triangles and objects. Looking at this another way means we’ve now got Default shadows with no performance cost i.e. now we get free shadows, which isn’t a problem. But I gather that most people see a benefit from turning the shadows off. I see you’re running this on the High preset. What if you set everything to the Low preset except the shadows in order to isolate them? Right now it looks like you’re being capped by some other variable. 

i9-14900KS | ASUS ROG MAXIMUS Z790 HERO | 64GB DDR5 5600MHz | iCUE H150i Liquid CPU Cooler | 24GB GeForce RTX 4090 | Windows 11 Home | 2TB Samsung 980 PRO NVMe | Corsair RM1000x | LG 48GQ900-B 4K OLED Monitor | CH Fighterstick | Ch Pro Throttle | CH Pro Pedals | TrackIR 5

Link to comment
Share on other sites

Default shadows are not free, they put more load on GPU. In this particular spot and with an RTX4090, the GPU has extra power to spare, so they don't make a big dent in FPS. Doesn't mean it's always like that.

You can see this difference between various options better in the screenshots from VR that I posted on the previous page. Gpu frametimes are the worst with Default shadows, Flat shadows reduce GPU frametimes but increase CPU frametimes. Setting the shadows to OFF reduces both, but not by much. CPU Frametimes without objects shadows in 2.8 are still higher than in 2.7 with shadows on.

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

45 minutes ago, some1 said:

Default shadows are not free

Sure I get that. But the test here somehow doesn’t reveal a cost for them. It shows this giant increase in triangles and objects without a corresponding drop in frame rate. 


Edited by SharpeXB

i9-14900KS | ASUS ROG MAXIMUS Z790 HERO | 64GB DDR5 5600MHz | iCUE H150i Liquid CPU Cooler | 24GB GeForce RTX 4090 | Windows 11 Home | 2TB Samsung 980 PRO NVMe | Corsair RM1000x | LG 48GQ900-B 4K OLED Monitor | CH Fighterstick | Ch Pro Throttle | CH Pro Pedals | TrackIR 5

Link to comment
Share on other sites

In a situation where you are cpu limited, cranking up a gpu intensive option will not have a noticeable effect. I could increase resolution or msaa, and it also wouldn't decrease framerate, at least not in that spot.

 

Besides, it's not that Default option has no penalty on the cpu, more likely the cost of extra objects added with shadows OFF negates the gains from not displaying shadows. Maybe tomorrow I'll check how are the numbers with all shadows off, including self shadowing and cockpit. That should reduce the object count.

 

 

 

  • Like 1

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

Ok, I tested the same spot on High preset but with All shadows disabled (both terrain objects and main shadows setting to OFF), and the performance of both 2.7 and 2.8 is the same, hovering around 180 FPS. And looks like I'm still on CPU limit, so GPU is not skewing the result.

I guess that is a good news, it shows that the shadows are indeed the main culprit of increased CPU usage in 2.8. 

2.7 off 180 fps avg.jpg

  • Like 2

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

19 minutes ago, some1 said:

 

I guess that is a good news, it shows that the shadows are indeed the main culprit of increased CPU usage in 2.8. 

 

Indeed, lets hope they can go back to halving the perf hit rather than doubling it with flat shadows, and giving us some options for the new cockpit shadows... 

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

YOUTUBE CHANNEL: @speed-of-heat

1569924735_WildcardsBadgerFAASig.jpg.dbb8c2a337e37c2bfb12855f86d70fd5.jpg

Link to comment
Share on other sites

Been having an absolute blast last few months with stable build 2.7 falling in love with the WWII birds while running at 3240x1920 at stable 60 fps @ 60Hz with mostly all maxed detail settings. I was very disappointed to get updated to 2.8.1 and see a solid and almost consistent 10 fps drop from where I was with absolutely no visual improvement to account for it. Granted I don't have a super high end system, 2.7 ran great. Both 2.7 and 2.8.1 run at the same fps with all terrain shadows turned off in 2.8.1 while 2.7 was set at default. Turning on flat shadows craters my fps in 2.8.1

ED, all of my respects to you and the phenomenal work you do. But please keep unstable and unoptimized implementations in the experimental branch until they have been thoroughly tested and proved fully functional to be promoted to the stable build. These are basic product development practices. 


Edited by ExGreyFox
  • Like 3

Intel i7-8700k Coffee Lake@stock

G.Skill TridentZ RGB 32GB (2x16) 3200MHz

ASUS ROG STRIX GTX 1080 Ti

Maximus X FORMULA Z370 BIOS 1401

NZXT Kraken x62

SAMSUNG 970 EVO M.2 2280 500GB (OS)

SAMSUNG 970 EVO M.2 2280 1TB (Games)

Corsair AX1200i 1200 Watt

Cooler Master HAF X

HOTAS Warthog | VKB MCG Pro GunFighter Mk. II | MFG Crosswind | TrackIR 5

Heatware

Link to comment
Share on other sites

7 minutes ago, ExGreyFox said:

ED, all of my respects to you and the phenomenal work you do. But please keep unstable and unoptimized implementations in the experimental branch until they have been thoroughly tested and proved fully functional to be promoted to the stable build. These are basic product development practices. 

 

I must agree. I am extremely dissappointed by stable moving to 2.81. It feels like mockery to me. Stable version has always been a bit of a meme, but at least it avoided the biggest issues. Now with stable on 2.81 it has become a complete farce.


Edited by twistking
  • Like 7
  • Thanks 2
Link to comment
Share on other sites

Apologies for going off topic but is there an easy  way to revert back to 2.7 without having to do a full blown uninstall/re-install?

 

2.8.1 with it's terrain shadows fps drain is currently unacceptable.

Intel i7-8700k Coffee Lake@stock

G.Skill TridentZ RGB 32GB (2x16) 3200MHz

ASUS ROG STRIX GTX 1080 Ti

Maximus X FORMULA Z370 BIOS 1401

NZXT Kraken x62

SAMSUNG 970 EVO M.2 2280 500GB (OS)

SAMSUNG 970 EVO M.2 2280 1TB (Games)

Corsair AX1200i 1200 Watt

Cooler Master HAF X

HOTAS Warthog | VKB MCG Pro GunFighter Mk. II | MFG Crosswind | TrackIR 5

Heatware

Link to comment
Share on other sites

13 hours ago, some1 said:

Ok, I tested the same spot on High preset but with All shadows disabled (both terrain objects and main shadows setting to OFF), and the performance of both 2.7 and 2.8 is the same, hovering around 180 FPS. And looks like I'm still on CPU limit, so GPU is not skewing the result.

I guess that is a good news, it shows that the shadows are indeed the main culprit of increased CPU usage in 2.8.

Hi @some1. I've been reading this threads with passion and I did the same  tests. I'm reacting now because I don't get the same result regarding this last test of yours: even with no shadows (no secondary shadows, no preimary shadows, no terrain shadows), I still get a 15% loss of FPS, and I don't quite understand why. Any idea?

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

Link to comment
Share on other sites

What's your gpu usage with all shadows off? 100% or less?

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

In this case you're testing GPU limits, not CPU. Not a bad thing to do, just different. 

For me the GPU usage in this spot does not ever reach 100%, so my FPS measurements are limited by CPU. 

Hardware: VPForce Rhino, FSSB R3 Ultra, Virpil T-50CM, Hotas Warthog, Winwing F15EX, Slaw Rudder, GVL224 Trio Throttle, Thrustmaster MFDs, Saitek Trim wheel, Trackir 5, Quest Pro

Link to comment
Share on other sites

On 1/1/2023 at 8:44 AM, some1 said:

Ok, I tested the same spot on High preset but with All shadows disabled (both terrain objects and main shadows setting to OFF), and the performance of both 2.7 and 2.8 is the same, hovering around 180 FPS. And looks like I'm still on CPU limit, so GPU is not skewing the result.

I guess that is a good news, it shows that the shadows are indeed the main culprit of increased CPU usage in 2.8. 

2.7 off 180 fps avg.jpg

Been doing lots of before/after comparisons lately. I would like to add that the shadows on/off performance hit has something else going on as well I think. It gets progressively worse the longer a server is running, independant of how many other players are currently connected. After the server resets, it goes back to baseline performance again.

You can bruteforce a reduction in the parseobj spikes by reducing the object draw distance in the lua settings, but this simply reduces the range that all ground forces / planes etc are drawn at, which is far from ideal. But dropping it from 80km to 50km or so can make a big difference. Some quick before/after reset comparisons, with shadows on flat and all other shadows (including terrain object) off:

 

 

I'm at the point where I just disable shadows and enjoy the frames. Having to constantly work around the progressive performance degradation with shadows on is too much of a headache, and reducing the object draw distance is harmful to spotting. But as far as I can tell, having shadows flat or above results in DCS trying to run extra calculcations on EVERY ground and plane unit in an 80km range in the direction you're looking.


Edited by MoleUK
Link to comment
Share on other sites

I did some tests with 5600x/3070 and there is not perceptible difference between ground shadows off and default. Flat seem to be worse than those two. 
I also tried to limit the visibility of objects by altering the graphics.lua LodMut (or whatever it is called) parameter but as I fly relatively simple missions with not many objects, there was no change. 

There seems to be some deeper problem than just shadows. Even if I set all settings to minimum, the game is far from fluid. The game WAS totally fluid above big cities with very high settings in 2.7 until I reinstalled Windows Media Reality, then it got such bad that even 2.8 update did not make any change. 

Link to comment
Share on other sites

30 minutes ago, jurinko said:

I did some tests with 5600x/3070 and there is not perceptible difference between ground shadows off and default. Flat seem to be worse than those two. 
I also tried to limit the visibility of objects by altering the graphics.lua LodMut (or whatever it is called) parameter but as I fly relatively simple missions with not many objects, there was no change. 

There seems to be some deeper problem than just shadows. Even if I set all settings to minimum, the game is far from fluid. The game WAS totally fluid above big cities with very high settings in 2.7 until I reinstalled Windows Media Reality, then it got such bad that even 2.8 update did not make any change. 

The LODmult can help in other ways, if you want to reduce the draw distance for objects you have to change this line in the graphics lua. Search for "camera" line, you will see different values for all the different draw distance settings. Alter the second value of the "objects" line that you have your view distance set to.

So if you're on medium view distance, change it from 3000, 80000 to say 3000, 50000. If you're on high view distance change it from 5000, 80000 to 5000, 50000. etc etc. This is a very bruteforce method and reduces the distance all those planes/vehicles render in, so keep that in mind.

It does help reduce the parseobj spike significantly, but shadows completely off is just easier. Eliminates the problem entirely without having to reduce visibility. There might well be some other stuff going on as well in 2.8, but shadows has had by far the biggest impact for me.


Edited by MoleUK
Link to comment
Share on other sites

SSAO = Off

Ryzen 7 5800X3D | 64GB DDR4 3600| MSI RTX 4080 16GB Ventus 3X OC  | Samsung 970 Evo 2TB NVME | HP Reverb G2 | DIY Head Tracker Cap | Logitech X-56 throttle | VKB NXT Premium |  Win 11

"Any sufficiently advanced technology is indistinguishable from magic."

--Arthur C Clark

Link to comment
Share on other sites

8 hours ago, SkateZilla said:

out of curiosity what are everyone's SSAO Settings?

Off like most of us. Too Big fps impact for me.


Edited by YoYo

Webmaster of http://www.yoyosims.pl

Yoyosimsbanner.gif

Win 10 64, i9-13900 KF, RTX  4090 24Gb OC, RAM 64Gb Corsair Vengeance LED OC@3600MHz,, 3xSSD+3xSSD M.2 NVMe, Predator XB271HU res.2560x1440 27'' G-sync, Sound Blaster Z + 5.1, TiR5, [MSFS, P3Dv5, DCS, RoF, Condor2, IL-2 CoD/BoX] VR fly only: Meta Quest Pro

Link to comment
Share on other sites

SSAO = off

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

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

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