Flyinside + VR + P3D v3.3 experiences and questions and cfg tweaks

[glider1 90]

 

 

 Still waiting on the leap, although using the mouse isnt so bad.

 

PTA is a shader tweak you can use to adjust things like texture and shadow brightness. It's not worth stuffing around with on VR until everything else is sorted. The perception of brightness is different for everyone. For me, Rift is far from ideal but just barely good enough to stick with VR permanently and pack away the monitor for good.

 

It will be interesting to decide whether to go leap motion or wait for the rift touch controllers. Leap motion sounds ideal but I worry about the CPU consumption.

 

Cheers

[Skywolf 992]

Biggest issue is LM needs to do rapid cleaning of VAS or move to 64bit all together.  I have done extensive (and I really mean extensive) tests with flyinside beta builds with p3d 3.2 and 3.3.5.  I am literraly left with hardly any free VAS.  I have to pick which addons to install and which birds to fly.

 

VR is pushing the old 32bit beyond its capabilities (there really isn't any more gas left in the sim  - aka "Jim, She can't take it any longer").  Flyinside is doing excellent work but the sim has to change in order to full load with complex addons/sceneries and what not.

[theskyisthelimit 48]

Well i have some results of the comparison from 980ti SC+ to 1080 FTW, frankly, i'm at a loss here.

Short story.. saw VERY Marginal difference in the VR tests, maybe slightly smoother panning, but overall, im actually seeing near identical if not 1-2 fps WORSE performance with the 1080..

 

I thought at first it must be a driver issue.. so i ran the installer doing the "clean" install option/rebooted etc..

No change..

 

Really puzzled here.. anyone else made the swap and ran into this?

 

IE: my fraps 120sec test for fsmark07 in ksea area was about 24.8 fps prior to the swap (with shadows and loads of things enabled), after, it became 23.80 or so.. same thing in vr, sitting on runway kors in bonanza was around 39 fps in the goggles.. now its the SAME.... same drop over the water to 28 (maybe up from 24 there).. still nowhere as smooth as aerofly fs2 with everything i have enabled too.. and affinity mask of 1365.

 

I also took off from kors in the realair lancair.. flew straight ahead for maybe 8 mins.. then ran out of memory.. lod was 5.5 during that test..

Frustrations here.

 

edit: my 3dmark firestrike ultra score is 5511, whereas the 980ti was 4169, i'd say this seems about right here at least.

 

I also recreated the prepar3d.cfg with no change

[Bert Pieke 7,585]

The 1080 does not have better specs on all counts, compared to the 980ti..

 

For example, memory bus is 384 bit for the 980ti,  256 bit for the 1080.. :excl:  

[Skywolf 992]

The 1080 does not have better specs on all counts, compared to the 980ti..

 

For example, memory bus is 384 bit for the 980ti,  256 bit for the 1080.. :excl:  

 

 

IE: my fraps 120sec test for fsmark07 in ksea area was about 24.8 fps prior to the swap (with shadows and loads of things enabled), after, it became 23.80 or so.. same thing in vr, sitting on runway kors in bonanza was around 39 fps in the goggles.. now its the SAME.... same drop over the water to 28 (maybe up from 24 there).. still nowhere as smooth as aerofly fs2 with everything i have enabled too.. and affinity mask of 1365.

 

I also took off from kors in the realair lancair.. flew straight ahead for maybe 8 mins.. then ran out of memory.. lod was 5.5 during that test..

Frustrations here.

 

The upcoming patch for Flyinside will fix that memory issue.  The current 1.6 and all the previous builds have higher VAS issue and bug with realair aircrafts.  It is potentially fixed and patch will be coming soon - all the beta testers are testing heavily and helping Dan to get the release ready

 

For Flyinside it is recommend to have LOD at 3.5.  From VC there really isn't any different to see 4.5 and have to reduce VAS hits.

[theskyisthelimit 48]

The upcoming patch for Flyinside will fix that memory issue.  The current 1.6 and all the previous builds have higher VAS issue and bug with realair aircrafts.  It is potentially fixed and patch will be coming soon - all the beta testers are testing heavily and helping Dan to get the release ready

 

For Flyinside it is recommend to have LOD at 3.5.  From VC there really isn't any different to see 4.5 and have to reduce VAS hits.

 

Thanks for the info and tip, ill give that a shot for VR.

 

Still no real progress in getting the 1080 speeds that it should produce.  I put that topic in a separate thread here (made a few tweaks in NI, etc, no change, still the same).  Debating if this performance doesnt improve, if two 980ti's would make more sense, or maybe there would be no perceivable difference there as well (for VR).

[glider1 90]

 

 

Still no real progress in getting the 1080 speeds that it should produce

 

Are you distinguishing between the headset FPS and the simulator FPS? In your posts you seem to be talking more about simulator FPS. With the 1080 all that will happen is that it will lock to a headset FPS of 90-93 almost all the time. The simulator FPS is only marginally improved because it is largely CPU limited.

 

The 1080 might allow you to turn on more GPU intensive settings compared to the 980ti.

 

At the end of the day, the 1080 is there for maintaining full headset FPS lock. A 980ti will probably not hold the headset FPS lock as often as the 1080 does. It is just a matter of degrees and what CPU load you have on the simulator.

[theskyisthelimit 48]

Are you distinguishing between the headset FPS and the simulator FPS? In your posts you seem to be talking more about simulator FPS. With the 1080 all that will happen is that it will lock to a headset FPS of 90-93 almost all the time. The simulator FPS is only marginally improved because it is largely CPU limited.

 

The 1080 might allow you to turn on more GPU intensive settings compared to the 980ti.

 

At the end of the day, the 1080 is there for maintaining full headset FPS lock. A 980ti will probably not hold the headset FPS lock as often as the 1080 does. It is just a matter of degrees and what CPU load you have on the simulator.

 

Yeah i was mainly referring to simulator fps, which basically didnt change.  I really didnt notice a huge change/increase for the goggles either (all things held equal), i guess since i had alot of things already enabled in the comparison test, there isnt much left to turn on. i tried cranking more shadows and other things with the 2d test, it lowered things by around 10fps.

 

GPU max load seems to now get to 79%, with a gpu temp of 80c, fan speed 71%, 3152 memory use (2d test).. i may try a clean version of p3d/OS just to rule things out, plus the sim keeps crashing randomly.

 

Last crash was during 2d, this time terrain.dll caused a fault.

[theskyisthelimit 48]

Ran some more tests.. this time i took a bare bones v3.3.5 copy and cleaned out the local user files back to that clean state, just for fun.

 

flyinside reinstalled (not uninstalled).. i tested the KORS flight from the runway and also the default flight in VA.. in both cases i still get that darn studdering when panning view, absolutely stellar fps (set to medium basically, the defaults that flyinside uses).. still on 3840 res in flyinside though.. tried various monitor settings in flyinside/p3d.. 25 fps in sim, 30 in flyinside, 60 etc, unlimited..

 

doesnt matter what i do, i get the view studdering.. on the default flight its pretty subtle.. 

 

I tried an NYC test, everything was smooth (in the bare bones test).. tried a local western pa test with rain and full clouds.. smooth as ice.

 

Im at a loss, even on a bare bones install, why it seems certain areas cause pan/view studder and others dont.

 

Also.. i'm having moments where the screen throws up "static" for a half second.. almost like the hdmi/video signal is cutting out.. this only happened since switching to the 1080.. maybe a usb or hdmi port issue, unsure.

 

So my question is if this studdering on view panning exists for everyone else in VR at say KORS or the default flight (key is once you are in the air, not on the ground.. up in the air it is noticeable).. for a sanity check here.

[glider1 90]

 

 

monitor settings in flyinside/p3d.. 25 fps in sim, 30 in flyinside, 60 etc, unlimited

 

I'm not at the PC now so can't test.

 

The monitor settings you want as low as practical since your GPU is having to drive both the headset and the monitor and the headset FPS is far more important than the monitor FPS.

 

You could try reducing the FOV field of view right down to see if that cures your stutters.

 

For the purposes of testing, lock your internal frame rate at 30. ATW should work very well in the  gap between 90 fps headset and 30 fps sim.

 

For the purposes of testing, fly the exact same scenario on your 2D monitor with VR off and compare performance.

 

It is possible that some sceneries cause the sim to stutter, which is easily confused with headset stutter when it is actually sim stutter.

 

 

 

Also.. i'm having moments where the screen throws up "static" for a half second.. almost like the hdmi/video signal is cutting out.. this only happened since switching to the 1080.. maybe a usb or hdmi port issue, unsure.

 

I'm a bit concerned about how hot your 1080 is running that could be a factor. You mentioned it is at 100% load and 80 degs C? Mine is 90-100% load at 54 degrees C. If it over heats it could cause all kinds of throttling issues that end up as stutter. Try running flyinside without upsampling to see whether the half second screen lock goes away when the GPU is under less load.

[theskyisthelimit 48]

I'm not at the PC now so can't test.

 

The monitor settings you want as low as practical since your GPU is having to drive both the headset and the monitor and the headset FPS is far more important than the monitor FPS.

 

You could try reducing the FOV field of view right down to see if that cures your stutters.

 

For the purposes of testing, lock your internal frame rate at 30. ATW should work very well in the gap between 90 fps headset and 30 fps sim.

 

For the purposes of testing, fly the exact same scenario on your 2D monitor with VR off and compare performance.

 

It is possible that some sceneries cause the sim to stutter, which is easily confused with headset stutter when it is actually sim stutter.

 

 

 

 

I'm a bit concerned about how hot your 1080 is running that could be a factor. You mentioned it is at 100% load and 80 degs C? Mine is 90-100% load at 54 degrees C. If it over heats it could cause all kinds of throttling issues that end up as stutter. Try running flyinside without upsampling to see whether the half second screen lock goes away when the GPU is under less load.

Are you suggesting setting the monitor fps in flyinside to 30 and the the frame lock at 30 in the sim? I did try that.. i think i tried 30 in flyinside and unlimited in the sim too..

 

My gpu seems to be about 64c when its idle.. it heats up to around 83C and now i tweaked the fan to go as high as it needs and it seems to keep it there, even at 99% gpu load.. Didnt you say yours was watercooled though? Mine is the standard air cooler for the FTW 1080.

 

So the update here, is that i did another test, going a little lower on the flyinside resolution/supersample.. i dropped down to the near native, 2368x1464.. tried those same situations that had view stuttering (and i did verify its view not scenery thats stuttering).. its gone.. so i guess even the 1080 couldnt handle the highest super sample with medium p3d quality settings.

 

What is the supersample res that you run at? Also. Maybe it's some correlation on the stutter when flying over water plus resolution.

[glider1 90]

 

 

Are you suggesting setting the monitor fps in flyinside to 30 and the the frame lock at 30 in the sim? I did try that.. i think i tried 30 in flyinside and unlimited in the sim too.

 

The monitor fps setting in flyinside is just for the external monitor. It does not have any other function on the sim or the headset. I had my external monitor set to 1 FPS but found that 15FPS didn't load my GPU down too much. We don't need a very high monitor FPS because we aren't using the monitor for simming, just for setting up the simulator before flying. If you put the monitor on unlimited, that means the GPU is driving the external monitor really hard as well as the headset. That would be too much.

 

 

 

What is the supersample res that you run at? Also. Maybe it's some correlation on the stutter when flying over water plus resolution

 

I run at 4K supersample and almost all the important shadows as well as ultra water including receive shadows for simobjects. All those settings are GPU intensive settings. Doing that makes the GPU run at 100% but temps on water keep it at less than 60c on a 2.1GHZ overclock and there is no headset stutter even after many hours.

 

Your idle temp of 64c is too high in my opinion. Boot up your PC from cold and watch the GPU temps when it is running the desktop only. There is no way it should be on 65c for doing that. If it is, there is something wrong with it in my humble opinion. It should be 35c tops under no load.

 

If the GPU get's too hot, it will cause stuttering and even crashes. I would expect an aircooled 1080 to comfortably run default settings but upsampled to 4K res. Beginning to suspect something with your card.

[theskyisthelimit 48]

The monitor fps setting in flyinside is just for the external monitor. It does not have any other function on the sim or the headset. I had my external monitor set to 1 FPS but found that 15FPS didn't load my GPU down too much. We don't need a very high monitor FPS because we aren't using the monitor for simming, just for setting up the simulator before flying. If you put the monitor on unlimited, that means the GPU is driving the external monitor really hard as well as the headset. That would be too much.

 

 

 

Your idle temp of 64c is too high in my opinion. Boot up your PC from cold and watch the GPU temps when it is running the desktop only. There is no way it should be on 65c for doing that. If it is, there is something wrong with it in my humble opinion. It should be 35c tops under no load.

 

If the GPU get's too hot, it will cause stuttering and even crashes. I would expect an aircooled 1080 to comfortably run default settings but upsampled to 4K res. Beginning to suspect something with your card.

 

Ah this makes sense on the monitor fps.. i misread that as something else.. so this value being higher than say 15, could be whats causing my stutter.

 

On the temps.. my video card read 45C when it booed from cold.. after about 12 minutes it was reading 63 and very low fan speed (just web browsing).. still a bit toasty.  My pc sits up above my desk in the corner of the room on a shelf.. maybe 5.5 feet high or so at the point where it sits, so that doesnt help, though its not that hot there at all.  I have one giant fan on the CPU and just the fan on the video card pulling air out.. i used to have one of those side big case fans, but it never helped much with the cpu temps at least.. i also used to run front fans on my drives pushing the air backwards.. but now 2 of my 3 drives are SSD and dont run hot, so i no longer have those active.  Not sure if this is a dud 1080 or just bad cooling perhaps.

edit: i did see others reporting idle temps 50-52 and load 82C without fan curves set (but even others said that too seemed high).. so maybe i'm just a tad warm due to the arrangement, not sure what the quick fix here is.  I should also check to see if the board is downclocking when idle.

 

edit: quick check after an hour, i show its idling at 49C with really little to no fan speed action and is downclocking to 253/405 (gpu/mem).. if i turn the fan on and let it sit at 30%, the GPU becomes 44-46C.

[theskyisthelimit 48]

I tried setting the fan curve .. idle.. i set it to 60% if 50C etc.. temps dropped to 40-44C idle.. (unsure how much idle below 55C truly matters, but tried it).. then when i run a 2D fsmark07 test again.. i hit 99% gpu, i guess thats good.. but the temp hits 82/83C on like 90% fan speed (i set the curve around here for that temp of 80+)..

 

I checked the core speeds.. showing 1800 on avg with a spike max to 2012. (boost is really 1860 i think).. so maybe its actually doing ok core speed wise.

 

Temps are still toasty.  My cpu is scary hot, 85C on one core.. in the past this never exceeded 75C.. the ivy bridge 4930k is still at 4.4 and is growing older, unsure if it just needs more cooling or a refresh on thermal maybe.. or replaced altogether frankly.. maybe with an i7-6700 or wait for Kaby Lake i7-7700k (4.2-4.5ghz turbos).

[glider1 90]

 

 

Temps are still toasty.  My cpu is scary hot, 85C on one core.. in the past this never exceeded 75C

 

I had similar CPU temps on my i2600k. I think you will get stuttering from that as well.

 

What makes it worse is that the GPU is hot and pumping heat into the same area that the hot CPU is sitting.

 

The hot CPU could be because the thermal paste has dried up. You could replace the paste or get a better cooler for the CPU.

 

In my humble opinion, one of either the CPU or GPU needs to be water cooled. If both are on air, there is too much heat in the case.

 

I once had a CPU on air running at 85C and a GPU on air running at 80C and there was stuttering.

Then I went to a i6700k CPU on water at 60C with the same air cooled 80C GPU and no stuttering.

Then I went to a water cooled i6700k on 60C with a water cooled GPU on 60C and there are no issues with stuttering.