Jump to content

Severe input lag with F-14 when force feedback on with MS Force Feedback 2 stick in use EDIT: Direct Input SOLUTION!


Aries144

Recommended Posts

EDIT: I determined conclusively that this is caused by activating force feedback. Turning off force feedback eliminates the issue completely. It seems likely that this is caused either by something introduced in the last couple of patches or it's related to my new AMD 6900XT.

EDIT 2: I have an update that I think completely solves the issue without using the "input = { threaded_ffb = false }" line in the autoexec.cfg file! This new solution is better because it comes without the negative performance aspect of using that autoexec entry. 

I stumbled onto this by happenstance while researching a completely unrelated problem. All credit goes to user l3VGV from the Russian language IL2 forums. The solution is simply adding the linked dinput8.dll file to the DCS World\bin folder.

https://github.com/l3VGV/FixFFBStutterH/releases/tag/v1.02

No lag, no measurable negative impact to performance. Force feedback forces appear to all work correctly. Problem solved.

Bug: 0.5 second control input lag when force feedback is on and Microsoft Force Feedback 2 stick is in use. All controls including keyboard affected. Also affects opening menus ex. Esc to open menu.

Can I reproduce it 100%: Yes. 100%. Just turn on force feedback and play any mission with the F-14A or B with Microsoft Force Feedback 2 joystick in use.

 

How to reproduce/ description: Turn on Force Feedback in DCS Settings Menu. Easiest way to observe is to launch F-14 Dogfight from Instant Action. Input lag may not be immediately noticed by moving joystick. Repeatedly pressing the escape key to open and close the game menu makes this obvious, as it will lag anywhere from a half second to a full two seconds. This lag is also present in all other inputs, most importantly, flight and weapon controls.

Turning off force feedback completely eliminates the issue.

 

DCS Version: OpenBeta 2.5.6.60966

 

System Specs:
Intel 7700k

AMD 6900XT

48GB RAM

SSD

OS: Win10

 

Peripherals:

Joystick: MS Force Feedback 2 + CH Fighterstick for buttons + Leo Bodner board for axis inputs 

Throttle: Thrustmaster Warthog

Pedals: Saitek rudder pedals

 

Headtracker: HP Reverb G2


Edited by Aries144
  • Like 1
  • Thanks 2
Link to comment
Share on other sites

I had major lag once, so all I did was remove ALL my graphic  drivers reboot then re install the latest graphic drivers.

It worked on that occasion but not all situations are the same.

 

Rapid 


Edited by rapid

Asus ROG Crosshair Hero VIII , Ryzen 3900X, Nzxt Kraken Z73, Vengence RBG Pro DDR4 3600mhz 32 GB, 2x Corsair MP 600 pcie4 M.2 2 TB , 2x Samsung Qvo SSD 2x TB, RTX 3090 FE, EVGA PSU 800watt, Steelseries Apex Pro. TM WartHog,TM TPR, Track IR, TM 2 x MFD, Asus VG289Q, Virpil Control Panel#2

Link to comment
Share on other sites

  • Aries144 changed the title to Severe Input Lag Only with F-14 Module When Force Feedback On

Updated: It's caused by Force Feedback. Turning off force feedback completely eliminates the issue. Also, if something happens to interrupt force feedback, i.e. the stick stops receiving force data and goes limp, the input lag goes away. 

This is new, as I've been using force feedback since the F-14 launch and never had this problem. Either this was introduced in one of the recent patches, or it's some weird interaction with the AMD 6900XT.


Edited by Aries144
  • Like 1
Link to comment
Share on other sites

  • Aries144 changed the title to Severe input lag with F-14 when force feedback on with MS Force Feedback 2 stick in use
3 hours ago, fat creason said:

I use the same stick to develop and test the FFB model and have never seen this. It's most likely a local hardware issue or possibly a DCS control input/FFB handling issue, which we have no control over.

This happens only with the F-14. I have most of the modules and this doesn't happen with anything but the F-14.

This is a recent issue for me. New GPU (6900XT), new VR headset (Reverb G2) is all that's changed on my system. Never had this problem before with the F-14 and my force feedback stick.

Could someone with a MS Force Feedback 2 joystick run the F-14 Dogfight Instant Action mission briefly and confirm that I'm the only one experiencing this issue? It'll take about 5 minutes.


Edited by Aries144
  • Like 1
Link to comment
Share on other sites

I have amended the original post to include pressing the esc key repeatedly in the "How to Reproduce" section. This makes the issue obvious, whereas it might otherwise briefly escape notice. Note that the delay here can be as much as two seconds.

I can confirm this does not occur with any other module, including others that make use of FFB, including the F-5E and Spitfire.

It can also be reproduced in the Free Flight instant action mission, though it seems to take some maneuvering for it to show up. Again, repeatedly pressing esc key makes this apparent. As soon as lag is seen pressing esc key, it can also be witnessed in the response of the aircraft, firing of the gun, and delayed movement of the virtual joystick in the game relative to the user's real flight controller.

  • Thanks 1
Link to comment
Share on other sites

As an owner of a MSFFB2 (the silver trigger/green power light version of the MSFFB2, if that matters) this input delay issue has plagued me since I got the module over a year ago and has severely impacted my ability to use it. I haven't taken the poor gal out of the hanger for more than a handful of flights. I can also confirm it only exists on the F-14 module and none of my other modules, some of which have also developed their own FFB effects.

For reference, my DCS is the steam version and this input lag issue has been apparent on both my old system, and my completely new system. Neither of my systems have shared any hardware or DCS configs. Everything is new and different, from monitors to keyboard and mouse. The only similarity has been the MSFFB2 joystick. Also, not that it matters all that much, but my old system was an Intel i5-4670K & Nvidia 2070 Super, and my new system is an AMD 3900X & Nvidia 3080.


Anyway, Aries, I may have a semi-fix you could try out: In your autoexec.cfg, paste this and let me know if it makes any difference to the input lag:

input = { threaded_ffb = false }


While the above removed the severe input lag on the F-14, I unfortunately still couldn't stomach the insane FPS hit I'd get when any force feedback was in effect for this module. For example: I could be in a free flight with no other entities but the F-14, and I could be getting a stable 120fps, but the moment I pulled enough G's to trigger the rumble force feedback effect for example, I'd drop to ~50fps or so for as long as that effect was active. If I were to turn off force feedback in the settings and do the same test, my frames wouldn't budge from 120fps during the high G effect.

While on a monitor I'd find this acceptable-ish, I unfortunately only really fly in VR these days and with DCS being a tad unoptimised in VR, the performance hit is unplayable and pretty much headache inducing.


Edited by SgtNathan
Link to comment
Share on other sites

  • 2 months later...

I can confirm this removed the input lag! 

I still need to try and measure differences in FPS/frame time to confirm what you report there.

Thank you SgtNathan! I consider this to be an improvement.

I wonder what is going on that's giving us these issues, yet Fat Creeson can't duplicate them?

  • Thanks 1
Link to comment
Share on other sites

I could never reproduce this issue; this apparent solution further confirms this is not F-14 specific and is more likely a DCS or hardware problem. All we do is read control inputs coming in via DCS's control input system and then tell the FFB stick how much force and rumble to apply via a very simple API provided by ED. This fix via an option you added in autoexec.cfg suggests the issue lies somewhere within the DCS control layer and/or FFB interface, neither of which we have any control over. Glad you could find a fix!


Edited by fat creason
  • Like 1

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

Glad I could be of help Aries!


fat creason, this could be a long shot, but while you were attempting to reproduce this issue, what sort of frame rate were you getting? I have a loose theory that the reason some people have this issue and some don't is perhaps because of a difference in system performance.

To start, my test to consistently reproduce this issue is to start a flight, and begin rapidly and continously rolling the aircraft left and right, while also continously hitting the escape key. When there's no input lag, the escape menu will appear and disappear immediately(!). When there is input lag, the escape menu will take from a few miliseconds, up to ten seconds to appear.

In this test, I can remove the input lag issue by software limiting my game's FPS to <50. On the flip side I can also exacerbate the issue by putting all settings to the absolute minimum, pushing my FPS around 200. This can cause up to a 10 second delay at times.

Link to comment
Share on other sites

I've been away from the F14 and DCS for the past week or so (waiting for the VR clouds jitter fix) but I will try and reproduce this issue. I've never noticed this behavior before (input lag) in the F14, but I did notice that the F14 underperformed compared to other modules with regards to performance it just never occured to me that it might be FFB related. Will try turning FFB on and off and report back as soon as I am able. 

 

(MSFFB2 user here)


Edited by Lurker

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 have been experiencing this for months. It's generally not noticeable in general flight but as soon as FFB shake effects activate it delays the registration of both axis and button inputs; snap gunshots at 15-20 units of AoA are heavily compromised. 

 

Tried it last night in the 2.7.0.5659 Open Beta and it is still a factor.

 

I have EDs F/A-18C, A-10C, Spitfire, P-51, P-47 and this FFB input lag is not occurring in these modules.


Edited by DD_Fenrir
Link to comment
Share on other sites

Has anyone tried the above fix? This topic should really be posted on the general DCS part of the forums, if anyone is going to have enough insight into the controls layer to see what's going on, it's ED and not us. We have no control over how or when inputs are read from hardware. Since it seems to be caused while the stick is shaking maybe we could remove all shaking effects from FFB, but I have no idea why that would cause DCS to have issues reading control inputs or have an impact on FPS. The same code is running all the time, only thing changing is the magnitude of shaking we ask the stick to do. As far as we know, all commands being sent to FFB are valid. To reiterate: we have no insight into this area at all.


Edited by fat creason
  • Like 1

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

I would call it more of a bandaid than a fix, as it introduces some severe side effects (a huge hit to fps), but if people deem that acceptable enough, and it makes the module actually enjoyable for them, then I'm very glad to have helped them out.

Also, fat creason, could you please attempt to try and reproduce this issue by following my above post? Please? Thank you.

Regarding posting this in proper forums, I think the reason it's posted in your section is because this issue isn't found on any other module - this has been constantly mentioned many times by people in similar threads to this one. I, for example, can experience stick shaking FFB with the JF-17, the same effect that I would with the F-14, only I'm not hit with this input lag issue at all. This asks the questions: Why does the F-14 module do this, when others do not?

Whatever the reason, would it not be possible to implement an option for people to use a more simplified FFB model that doesn't cause input latency issues?

Link to comment
Share on other sites

On 5/6/2021 at 8:59 AM, SgtNathan said:

In this test, I can remove the input lag issue by software limiting my game's FPS to <50. On the flip side I can also exacerbate the issue by putting all settings to the absolute minimum, pushing my FPS around 200. This can cause up to a 10 second delay at times.

I think this sounds quite intriguing, and another avenue to explore towards trying to reproduce and resolve this. I'm speculating very wildly now, as I'm not familiar with this part of the code at all, and don't have FFB myself, but perhaps we are somehow sending FFB updates to the joystick every graphics frame,  and over a slow USB1 interface, and with a driver from the 1990s or something like that. I can imagine something like that potentially resulting in buffering of IO etc. If so, it should be possible to rate-limit such updates to say 25Hz or something (again, I'm making up numbers here for the sake of argument) to perhaps solve the problem.

 

Edit:  the above is probably totally wrong... it seems DCS queries our code for values of various FFB settings (e.g. pitch and roll deflection forces, shake amplitude and frequency etc.), and we simply return values for each of those (with no complicated calculations or anything that looks like it could explain some slowdown). Not sure what the issue could be then, and I don't have FFB to even try anything. Perhaps F-14 is the only module that returns shake effect values or something?


Edited by gyrovague
edit to state that DCS actually queries the FFB values from F-14
  • Like 1

____________

Heatblur Simulations

[sIGPIC][/sIGPIC]

Link to comment
Share on other sites

Yeah, we simply have no insight regarding what could cause this problem. Even if I could reproduce it, there's nothing I could look at in the F-14 code that would tell me what the cause is, as well as no way to fix it. I would just get the same delays you're experiencing and that's about it.  We merely set a couple values for FFB devices that DCS queries whenever it wants them (again, we have no control over when they're queried or sent). We're not overloading a controls buffer or anything like that. That's why you should post this problem on the ED side of the forums, they're the only ones who have the ability to see what the cause might be.


Edited by fat creason
  • Like 2

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

Gyro, Fat, thank you both for giving this some attention. This looks like one of those glitches that's going to be a PITA to pin down.

I have a powered USB3 card (that has it's own direct connection to the PC power supply) plugged into my motherboard that I can try.

SgtNathan, I also just realized that I'm running an Intel 7700k processor with hyperthreading disabled in the BIOS. You aren't running with hyperthreading disabled as well, by any chance? I'm going to try turning hyperthreading back on in BIOS and see if that changes anything.

 


Edited by Aries144
  • Thanks 1
Link to comment
Share on other sites

On 5/11/2021 at 8:52 PM, gyrovague said:


@SgtNathan@Aries144 Have you guys tried using a powered USB hub for the MSFFB2?  (grasping at straws here...)


Gave it a try because why not, but unfortunately the problem remained the exact same.
 

On 5/12/2021 at 10:46 PM, Aries144 said:

SgtNathan, I also just realized that I'm running an Intel 7700k processor with hyperthreading disabled in the BIOS. You aren't running with hyperthreading disabled as well, by any chance? I'm going to try turning hyperthreading back on in BIOS and see if that changes anything.

 

 


On an AMD processer myself, and I have always had hyperthreading turned on. Just tried it without and no change on my end. Problem remains unfortunately.

Link to comment
Share on other sites

  • 3 months later...
On 5/11/2021 at 1:58 PM, fat creason said:

Yeah, we simply have no insight regarding what could cause this problem. Even if I could reproduce it, there's nothing I could look at in the F-14 code that would tell me what the cause is, as well as no way to fix it. I would just get the same delays you're experiencing and that's about it.  We merely set a couple values for FFB devices that DCS queries whenever it wants them (again, we have no control over when they're queried or sent). We're not overloading a controls buffer or anything like that. That's why you should post this problem on the ED side of the forums, they're the only ones who have the ability to see what the cause might be.

 

I have an update that I think completely solves the issue without using the "input = { threaded_ffb = false }" line in the autoexec.cfg file! This new solution is better because it comes without the negative performance aspect of using that autoexec entry. 

I stumbled onto this by happenstance while researching a completely unrelated problem. All credit goes to user l3VGV from the Russian language IL2 forums. The solution is simply adding the linked dinput8.dll file to the DCS World\bin folder.

https://github.com/l3VGV/FixFFBStutterH/releases/tag/v1.02

No lag, no measurable negative impact to performance. Force feedback forces appear to all work correctly. Problem solved.


Edited by Aries144
  • Thanks 2
Link to comment
Share on other sites

  • Aries144 changed the title to Severe input lag with F-14 when force feedback on with MS Force Feedback 2 stick in use EDIT: Direct Input SOLUTION!
On 9/13/2021 at 3:40 AM, Aries144 said:

I have an update that I think completely solves the issue without using the "input = { threaded_ffb = false }" line in the autoexec.cfg file! This new solution is better because it comes without the negative performance aspect of using that autoexec entry. 

I stumbled onto this by happenstance while researching a completely unrelated problem. All credit goes to user l3VGV from the Russian language IL2 forums. The solution is simply adding the linked dinput8.dll file to the DCS World\bin folder.

https://github.com/l3VGV/FixFFBStutterH/releases/tag/v1.02

No lag, no measurable negative impact to performance. Force feedback forces appear to all work correctly. Problem solved.

 


Interesting discovery, bring this to ED’s attention.

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

  • Recently Browsing   0 members

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