B4ron

I disable both compatibility settings and tried, unfortunately no performance increase. I also reduced preload radius from 55000 to 40000, also no noticeable effect. I'm only running opentrack. Afaik opentrack can use either trackir and freetrack as communication protocol since it has no own. I chose trackir as protocol, that's why opentrack is running trackir.exe in addition to opentrack.exe. Anyway I closed trackir.exe - also no effect on performance. May it be possible that 1600x900 as native resolution is not supported by 2.7 and therefore will be calculated down from a higher resolution? If assumed that its calculated down from 1920x1080 this would pretty exactly match up with the performance loss.

Thank you for your answer. Ok, lets go through it: DXGUI_WIN_ADAPTER: ChangeDisplaySettings failed! Result is -1 This error message is not present in DCS stable 2.5.6 log (see log file attached)! Do you think this might be the problem? 1600x900 is the native resolution of my display. 1. The pagefile is already located to D:\. I already did this when building the SSD in. The actual installation is: D:\ (SSD) = DCS stable, DCS openbeta, Saved Games folder, pagefile, OpenTrack C:\ (HDD) = OS Windows 10, rest of the programs wmic pagefile list /format:list AllocatedBaseSize=6144 CurrentUsage=36 Description=D:\pagefile.sys InstallDate=20210614203138.160075+120 Name=D:\pagefile.sys PeakUsage=3136 Status= TempPageFile=FALSE 2. The GPU drivers for my graphics card are not supported by AMD anymore. The installed 6-year old version is the most actual one. I already did exactly this. I completely removed the GPU driver and installed the latest version 15.7.1 WHQL from the AMD page (date 29.07.2015). Windows 10 did another update later on the driver which is the actually installed one: "Advanced Micro Devices, Inc. driver update for AMD Mobility Radeon HD 5000 Series / Erfolgreich installiert am ‎05.‎05.‎2021" (from Windows Update history) Non of the driver versions brought a performance increase. The driver supports DX11 and works fine with DCS stable 2.5.6. Or have you changed something on the graphics engine from 2.5.6 to 2.7 that the driver does not support anymore? However if you find it useful to reinstall the driver again, I can do that. 3. I think the problem is GPU related, since the GPU utilisation is usually at 99% when running DCS (in fullscreen mode). Usually the 1GB VRAM of my graphics card is the bottleneck of the system. Have you changed anything from 2.5.6 to 2.7 that generates more traffic between GPU und VRAM? If so this might be the problem, since more overflow VRAM is allocated to RAM and further to the pagefile which might cause the performance drop? May a reduction of "preload radius" then be a sollution in order to lower RAM usage? dcs.log

No I haven't. But I only opened OB once after each update just to see if the performance improved.

The crash definitely happened after update to 2.7! I completely uninstalled OB on April 29th, renamed DCS stable folder (in order to prevent files being copied from there) and fresh installed OB 2.7. The crash happened on May 24th.

As requested I opened a new thread here: Thank you.

As requested here is the new thread with my performance issues: - after update to DCS 2.7 I recognise a fps drop of >30% compared to DCS 2.5.6 - complete clean reinstall done, game is at initial setup, no mods enabled, graphics are set at the same values as in 2.5.6 - several measures have been tried with no success: update GPU driver, Update Windows 10, update Direct X, disable clouds, ... Attached you can find the requested files + the last crash log (since DCS 2.7 is regularly crashing). If you need more information please let me know. Thank you. dcs.log dcs.log-20210524-120210.zip DxDiag.txt process_list.txt

It's amazing how ED ignores their customers problems with their product!!! 2.7 is still unplayable for me as for many others. We are just left alone in the dust, not even a serious attempt to help us with the product we payed for! I followed this post from the beginning and tried every possible advice also from other sources, including complete reinstall. Nothing worked! Meanwhile I found a solution for me: IL-2!

Update from my side. I tried everything I was able to find: - rolled back Windows K5001330 - updated to K5001330 with hotfix from Microsoft again - slow repaired DCS - deleted Saved games folder, cleaned misc shaders folder - clean reinstalled DCS openbeta (even with renaming DCS World folder, to avoid copying data from there) - rolled back AMD GPU driver - no Tacview installed - clouds set to low Nothing of this had noticeable impact on performance. Still 30-40% fps loss compared to 2.5.6 stable. Since I have a low end rig and my graphics set to almost lowest level, I cannot reduce them any further. 2.7 is not playable for me at the moment.

Just 1 thing to discuss: In another thread a user helped himself with a clean win10 reinstall. To the guys still having problems: When switching to win10, did you update win10 from win8, or have you clean installed win10?

clean Win10 install - thats what I haven't tried yet. What exactly did you do? Settings / recovery / recover this PC / remove everything?

I uninstalled KB5001330, did a new clean DCS install, switched off windows hags and/or fso. Nothing had effect on 2.7 performance. Still 30-40% fps loss compared to stable with the same settings (even with clouds on low and details at 0.1).

The problem is not only VR! I have a low end rig without VR and also recognizing about 30% fps drop. I did a fresh install with deleting the saved games, rolling back windows 1330 update, switched of windows hags and fso - all of which with no effect. So maybe the problem is more based on screen resolution, since only about 15% of the users are recognizing this performance drop?

Hi ciribob. Since a week or so I have significant CPU usage by SRS when in MP and connected to the SRS channel. SRS uses around 20-25% of CPU which significantly influences DCS performance. This has not been the case before, CPU utilization was around 5%. I also installed Tacview, but deinstalled it meanwhile, since I initially thought this is the problem. But the problem definitely is SRS, when I disconnect the utilization goes back to 2-3%. Is there anything known which may get wrong?

Here you are: 1vs1 against AI P51. AI overheating.trk

Hi. When I fly SP against WWii AI and force them to climb after me at low speeds, they are going down after a while without being hit. This occurs after update to the new DM. It seems to me that they are overheating their engines. Is this a known problem?

I also updated stable version to the new damage model yesterday and what I saw in the first test flights was AMAZING! Weapons effects are now modelled to closely meet the historic content and the visual damage effects are superb! I doubted that the new damage model will come with the costs of a performance decrease, but this is absolutely not the case. Thanks to the developers. Chapeau!

Hi, if the value "condition" for AI aircrafts is set to <100%, the aircraft will be immediately destroyed straight after mission start. I assume that this value is supposed to set a "pre damage" for the AI aircraft?! Greats B@ron

As Nightstorm posted in this threat: https://forums.eagle.ru/showthread.php?t=274099&page=11, ED seems to have taken back the encryption of these files in open beta again. I hope this will also get into the stable version with the next update...

Just one more question: Why did you encrypt or compile the data in these files then with update 2.5.6?

As I see this seems to lead to nothing, I will upload the modded aircraft.lua files into the user files for download as soon as they are hopefully editable with the next stable update. I also did some modification on the aiming ability of AI opponents in order to behave more human like. So everyone can decide whether to download or not...

I agree. But for possibly historically based context I like the missions and campaigns that are also available as free user files. Why not doing those improvements (or bug fixes whatever you call it) when it's just simple settings?

They do definitely effect flight behaviour. weight settings: The total weight of AI and player aircrafts is calculated in the mission editor: M_empty+total fuel weight+total weight of ammunition (calculated from "round_mass" from weapons.lua)+weight of additional weapons (bombs and rockets). If you add an editional fuel tank or a bomb or reduce fuel, flight behaviour will change (especially the climb rate). Ergo also M_empty must have an effect! It can be tested by changing the value in stable version 2.5.5 or open beta (the aircraft.lua files have been encrypted or compiled in stable version 2.5.6 => this change has been taken back in open beta and hopefully will also be for stable version). If you set the value lower than your maximum thrust available, you'll even be able to climb vertically straight from the start (that's where those nonsense videos on youtube come from and likely the reason for encrypting the files). M_nominal and M_max are only values for comparison to the actual total weight IMO. weapons effect: If you set the value "life" from 18 to 9 you'll need 2 MK108 hits in average to get a P51 down (which is far closer to historic data and the player DM). If you set the value to 1, the AI aircraft will go straight back to the airfield for landing. You'll need only 1 MG131 hit to bring it down then. So this value will not effect weapons effects directly, but will set the "damage bar" for AI aircrafts to a lower initial value. MK108 rounds for Bf109 motor cannon: You can count the total ammunition by firing your whole ammo stock and count the number of tracer bullets. In standard setting every 3rd bullet is a tracer one. You'll also be able to see it on ammunition weight. So this setting definitely influences the ammo count for the player flight model.

I don't want AI opponents to be easier to kill, that's not the point. I'm frequently winning air combats against AI planes on high or excelent skill level. The point is that DCS is a flight simulator with a claim to be as close to historic realism as possible. This should also be true for AI planes since you are interacting with them. This involves to set up correct data, in order to be at least as close as possible to the historic role model. 1. weight settings in aircraft.lua files: This is the settings that will be shown in mission editor. These weight settings influence the flight behaviour of both AI and player aircrafts. I tested it! In order to have a simulation closer to reality, those weight settings should be according to historical data. 2. weapons effects: Historical tests and flight records show specific damage for certain weapon hits. If a AI P51 is still flying after 6 rounds of MK108 ammunition hits (there's different ones, I know) this can't be close to historic realism. Also if you consider AOA and approach speed (which by the way shouldn't have to have much effect on HE ammunition with a filling of 85g of TNT). 6 hits would have pretty shurely brought a B17 down! The current player DM is way more close to realism than the AI one. This can be adjusted with the "life" setting in the aircraft.lua. 3. amount of ammunition This is just a bug in the Bf109K-4.lua file. The command "count = 65" for MK108 is missing. If you insert it, the setting will be correct. This is also valid for both AI and player aircraft. I know that there's an improved DM supposed to come, but I doubt that this will solve all problems with AI.

Dear DCS WWII developers, DCS-World WWII collection is one of the most realistic available flight simulators of WWII aircrafts. The player flight model is developed with a lot of effort to meet the historic role model. Annoyingly there is some incorrect data in the simplified AI flight model which lead to historical incorrect behaviour of AI aircrafts. There are a lot of discussions in the forum about AI flight model and I understand that the PFM cannot be used for AI in order of performance reasons. What can be done IMO is to adjust the SFM to historically correct data. I used to adjust that data by myself but since update 2.5.6 for the stable version all aircraft.lua files are encrypted. Following data affecting the flight performance of AI aircraft need to be corrected in the aircraft.lua files in order to refer to historical data: 1. Adjust weight settings: Bf109K-4: M_empty = 2947kg -- was 2900.65kg (calculated from M_nominal) M_nominal = 3400kg -- was 3100kg (Report A/IV/294/44 app. 5026/28 of OBB Forschungsanstalt Oberammergau) M_max = 3900kg -- was 3400kg (max. bomb load) => with 3400kg the plane will be overloaded with the 500kg bomb! [ATTACH]239416[/ATTACH] FW190A-8: M_empty = 3640kg -- OK M_nominal = 4300kg -- was 4273kg (document nJ.190.801.-051 of Focke-Wulf Flugzeugbau G.m.b.H) M_max = 4851kg -- was 4840kg (max. bomb load) M_fuel_max = 388kg -- was 409kg (main tank = total fuel 640l - rear tank 115l) FW190D-9: M_empty = 3613kg -- was 3600 (calculated from M_nominal) M_nominal = 4273kg -- OK (document FW5232 of Focke-Wulf Flugzeugbau G.m.b.H) P-51D: M_empty = 3576kg -- was 3549 (calculated from M_nominal) M_nominal = 4593kg -- was 4000 (Flight Tests on the North American P-51D Airplane, AAF No. 44-15342 + 60gal fuel capacity left) M_max = 5260kg -- was 4800 (wikipedia.org) SpitfireLFMkIX (Merlin 66): M_empty = 3004kg -- was 2914.0 (calculated from M_nominal) M_nominal = 3395kg -- was 3100 (spitfireperformance.com) M_max = 3837kg -- was 3400 (max. bomb load) 2. Adjust weapon effects Weapon effects on AI airplanes are generally to low due to historical data. For german weapons the following average number of hits have been reported to kill a plane: MK108 30mm: 1-2 hits for single engine fighter aircrafts (british trials of German 30mm ammunition) 1-2 hits for twin engine medium bombers (british trials of German 30mm ammunition) A very good summary with historical data for the test can be found on: https://forum.il2sturmovik.com/topic/41665-3cm-mk108-historical-test-data-vs-ingame-test-data/ MG151 20mm: 4-5 hits for single engine fighter aircrafts 20-30 hits for heavy bombers (wikipedia.org / "Messerschmitt Bf 110: Die Rehabilitierung eines Flugzeugs" ISBN 978-3837022896) To get a closer to real weapon effect the value "lifebar" should be adjusted to: life = 9; -- lifebar -- was 18 This is still too low compared to historical weapons tests IMO but on about the same level like for player flight model. 3. Bf109K-4 MK108 rounds MK108 motor cannon was equipped with 65 rounds (wikipedia.org + several more sources). The value "count" is missing in Bf109K-4.lua for MK108 => the count will be taken from weapons.lua which is set to 60. (This is true for MK108 wing mounted guns.) There are more incorrect data to be found in the aircraft.lua files which however do not have affect to the AI flight behaviour according to my testing. Thanks B@ron

Ok, thanks so far. Then I'll consider switching to OB.