influence adjustments on VR using FCAT-VR

[Dutch2]

 

 

WARNING, WHAT IS RUNNING GREAT ON MY SETUP COULD BE A DISASTER ON YOURS, ALWAYS REMEMBER YOUR LAST SETTINGS. DO TAKE NOTE THAT THE INGAME FPS COUNTER WILL TELL YOU NOTHING.  

BTW I’m not a specialist in benchmarking or wathever, any remark or improvement is very welcome. But please keep in your mind I’m trying to verify the influence of the adjustments on the gameperformance and not testing the CPU/GPU. 

My system:

i7-7700k, GTX1080, 16Gb RAM, Only for BoX 500Gb M2, Win10pro, Samsung O+ 

Thurstmaster T160000m, Simped, Saitek trottle quadrant. 

 

First try out of FCAT-VR software. Just a few free single flight nothing special but as you see big performance problems here. 

Edited March 18, 2019 by Dutch2

[Dutch2]

Overclocked my RAM 3000->3200 and CPU from stock->4,8Ghz.  OC=yellow. purple=stock

Edit: the scaling for the frametime has been changed so it looks now that it is now more peaking if compare this to my first post.  But that is only to let see the difference better. Here you see very clear OC does have a very positive effect on the deviation, the improvements in frametimes is only small. I’m sure you will never see that in your onscreen FPS counter. But for smoothness this is a very important aspect. 

 

Edited March 17, 2019 by Dutch2
scaling explained

[JonRedcorn]

Dang those are huge gains no? Need to oc my ram, have 3000mhz myself. Maybe time to buy some 4000mhz. I've tried to in the past, but I am not that good at ram oc'ing and with a high cpu oc its difficult sometimes to get it stable.

Edited March 16, 2019 by JonRedcorn

[5280magpie]

Having chucked a computer out of a window in frustration,  I'm dubious.

Vr seems to slog along only marginally no matter what I try.

 

[Dutch2]

New results starting from the bottom so first into the nVidia setting, I have set everything to application settings, off or to (max) performance, one thing did intrigue me  the Virtual pre-rendered frames in the global settings.  Originally it was set to 4 so I did a run on this setting starting in the air and  landing on a airstrip . After that the same only now at 1.  

First is free flight, blue is Virtual rendered frames set to 4 and the purple set to 1, verdict purple is more smaller but also very peaking and does have an higher frametime, while the blue one is wider but more straight. 

Second plot, is approaching and landing airfield,  blue is Virtual rendered frames set to 4 and the purple set to 1 Verdict, think here the blue is the winner.

 

Next 2 sessions, before going to the in-game settings itself, would be:

Session 1: To solve the resolution myth, while some see improvements in lowering to 700x400 or whatever and some not. I will start at 2560x1440 my max monitor resolution and go for the lowest that my mouse can handle as the sec. graphic.  

Session 2: From the Nvidia settings the "threaded optimisation"   On or Off. 

 Any other suggestions would be welcome, only keep in your mind what will run as an optimum on my system and Samsung O+, does maybe not count for yours.  

 

edit: For determine the quality, I watching the wires, wooden strut and leather trim on the Camel. 

 

  

 

Edited March 17, 2019 by Dutch2

[JonRedcorn]

37 minutes ago, Dutch2 said:

New results starting from the bottom so first into the nVidia setting, I have set everything to application settings, off or to (max) performance, one thing did intrigue me  the Virtual pre-rendered frames in the global settings.  Originally it was set to 4 so I did a run on this setting starting in the air and  landing on a airstrip . After that the same only now at 1.  

First is free flight, blue is Virtual rendered frames set to 4 and the purple set to 1, verdict purple is more smaller but also very peaking and does have an higher frametime, while the blue one is wider but more straight. 

Second plot, is approaching and landing airfield,  blue is Virtual rendered frames set to 4 and the purple set to 1 Verdict, think here the blue is the winner.

 

Next 2 sessions, before going to the in-game settings itself, would be:

Session 1: To solve the resolution myth, while some see improvements in lowering to 700x400 or whatever and some not. I will start at 2560x1440 my max monitor resolution and go for the lowest that my mouse can handle as the sec. graphic.  

Session 2: From the Nvidia settings the "threaded optimisation"   On or Off. 

 Any other suggestions would be welcome, only keep in your mind what will run as an optimum on my system and Samsung O+, does maybe not count for yours.   

 

  

 

You are a stud! Keep it up this is the real deal info we need. Keeping those frame times below 11ms is everything. Thank you so much! Be interesting to see which performs better, 4xaa or higher SS. Could do 200% no AA and 100% 4xAA.

 

What settings are you running to get those 8ms frame times? 

Edited March 17, 2019 by JonRedcorn

[Dutch2]

Here another Myth to be busted: the screen resolution. I did a test in; free flight, airstart 1500m, winter map, clear sky, all on autopilot and no OC.

Because I could not do any judgment I did plot them first separate and then as an total.   I did use the 2560x1440 (=purple) as being the resolution from my monitor and the 1176x664 (=blue) as is the lowest resolution that does have the same screen ratio. Lower then 1176x664 does always bring in a mouse freeze or an pointer disappearing at the selection fields.  On the blue one you see a bump in Frame time and this could be the sun.  

   

Check for your self what is the best? 

 

 

 

[lightbulbjim]

Cool, I won’t bother with the low resolution any more. Thanks! ?

[Alonzo]

This is interesting stuff. @Dutch2 Can you standardize the test track, to make it easier to compare runs? Chili has a replay file from the benchmarking thread, maybe try that.

 

For both pre-rendered frames test and resolution test, can you try again where you are getting < 90 FPS average? I don't think it's telling us very much when the GPU is easily achieving the 90 FPS framerate. I'm more interested in situations where you're < 90 FPS -- does pre-rendered frames or window resolution make a difference?

[Dutch2]

1 hour ago, Alonzo said:

This is interesting stuff. @Dutch2 Can you standardize the test track, to make it easier to compare runs? Chili has a replay file from the benchmarking thread, maybe try that.

 

For both pre-rendered frames test and resolution test, can you try again where you are getting < 90 FPS average? I don't think it's telling us very much when the GPU is easily achieving the 90 FPS framerate. I'm more interested in situations where you're < 90 FPS -- does pre-rendered frames or window resolution make a difference?

 

The whole problem is you need two exactly the same flights that is the most important , so to keep that I have to skip every activity, coulds, AI action, clear weather etc. You will not believe it, but even if the flight direction is not the same, the results are even at 1500mtr different because of that. BTW, for an intensive benchmark mission you will get more a less the same results as in my first posting. 

As long if 1C does not publish any game benchmark this is the only thing I could do.

 

An recorded flight, a video, does not bring the right benchmark I believe, as was already been answered by @SharpeXB when I did ask voor a benchmark advice. 

 

But I could try an PW-CG mission, that does have lots of La5 planes in the sky and an ME-110e mission that does have lots of smoke as been made by @NETSCAPE  

Only I do not know if this are the right benchmarks, I still think a benchmark contains an phase with high CPU/low GPU and  a phase that contains low CPU/high GPU load, who knows somebody can design that.  A random mission without that, gives in my point of view more an indication of the performance of your specific hardware and I would like to log the impact of the adjustments versus the game performance. 

 

The constrain FPS is btw the FPS the system could reach without the 90fps border. 

 

 

Edited March 18, 2019 by Dutch2

[SCG_Fenris_Wolf]

You don't need to call it "resolution myth". It was an actual thing in the starting days in VR, before all these HUD magic patches. The image is processed for VR first and dumbed on the monitor anyway as far as I know. So meh.

 

I haven't tested an 3440x1440 on an ultra-widescreen monitor yet, on a very low VR-SS level. That might show some effect, but tbh, who even does that?

 

That said, the myth will never die, it's futile to try killing it ?

[Dutch2]

19 hours ago, SCG_Fenris_Wolf said:

You don't need to call it "resolution myth". It was an actual thing in the starting days in VR, before all these HUD magic patches. The image is processed for VR first and dumbed on the monitor anyway as far as I know. So meh.

 

I haven't tested an 3440x1440 on an ultra-widescreen monitor yet, on a very low VR-SS level. That might show some effect, but tbh, who even does that?

 

That said, the myth will never die, it's futile to try killing it ?

 

Another testing is always welcome, I only trying to imitate on how the VR sites are test the Headsets. 

Edited March 19, 2019 by Dutch2

[dburne]

Two threads on testing with FCAT VR?