Prepar3D v2 at high res (5760x1080), micro stutters, some other tweaks discovered, fps impact on settings

[theskyisthelimit 48]

I took the plunge into Prepar3D and so far i'm rather pleased.  I've done extensive tweaking, changing one setting after another which has taken hours to fully complete at this point, with some more to go.  So I thought I'd share my experiences here.  Take a look at my DX10 thread link at the bottom of my sig for a comparable series of tests I did with FSX.  Incidentally the Estonia Migration tool is a fantastic piece of software.. makes installing most FSX addons a breeze.

 

Short story:

 

Since this is a long read.. here are the key points of what i found:

 

-fibertimeframefraction 0.1 being added made a world of a difference, relatively.. 5 fps boost

-bufferpools usepools 0 may or may not have added 0.5 fps

-autogen (both) on normal is about 4 fps better than the next level up

-Mesh changes.. not much of a hit and can maximize

-tessellation makes everything look THAT much better and with it off i saw a LOSS of 2 fps

-road traffic.. going from 0 to 7% LOSS 6 fps or so.. settled on 4% for now

-LOD radius from 6.5 to 5.5 only gains about 0.5 fps

-adding AffinityMask resulted in LESS stutters (bad ones anyway).. used 254 

-4xSGSS results in an average of about 9 fps LOSS.. sticking with "Multisample/off" for now, the improvement wasnt THAT great to begin with with SGSS (on my system/this particular test)

-4096 textures seemed to give slightly BETTER results (for me) than 2048 fps wise (unclear why, retest needed)

-REX runway textures in Rex4 dont look as "nice" as they did in Rex3 (side note)

-If i turn off FTX hybrid with FTX installed..i'm around 37 FPS over 120 sec (vs 32.50 with it on)

 

Not listed below but P3D is crashing randomly on the following dlls (usually on close):

Program Files (x86)\Lockheed Martin\Prepar3D v2\atc.dll (dated 11/21/2013)

Program Files (x86)\Lockheed Martin\Prepar3D v2\g3d.dll (0xc0000005) (dated 11/21/2013)

 

 

 

Longer story:

Keep in mind my results are for a very high resolution 5760x1080, at lower resolutions, the results below may be less obvious or dramatic.

 

The Hardware:

i7-4930k 12 logical cores HTon running at 4.4 ghz 1.32volts (max cpu temp with prepar3D around 74C with GPU at 82-84C) with the Noctua NH-U14S cooler.

Ram.. Gskill 2400mhz Cas9 16GB (4 chips)

Video.. Evga 780ti SC 3GB (single card)

LCD/Resolution:  3 x 24" Asus VG248QE at 144hz 5760x1080x32

OS: Sata III 64MB cache 6 Gb/sec drive

P3D: split partition on a separate drive with FSX on the other partition (Sata III 64 MB 2TB 6 Gb/sec)

 

The Software:

Nvidia Driver 332.21 (latest as of 1/17/2014)

Evga Precision Profile with fan curve (slightly worse if not running and fan maxed out made no difference in gpu performance)

Windows 8.1 x64

Prepar3D 2.0

Addons so far: various aircraft imported over, Rex4 (4096/max textures installed) Ftxg, ftx pnw (yes there are glitches in the textures, where you have to toggle the texture resolution back and forth to fix.. actually in the one gif image i have below you can see this)

FPS testing:  Fraps and the old Fsmark07 test and the jet from the test, imported overtop the AI only model in Prepar3D (seems to work).. ran for 40 seconds on a host of tests, then towards then end to get more accurate results 120 seconds on each test as noted.  The fsmark07 flight can be found here.

 

**Note.. i'm using fsmark07 for most of the results below.. its an antique i know, maybe not true to all situations, but the purpose was to see how each change would affect fps and it worked.. and of course there are some glitches with PNW, with the texture glitch and missing taxiways at KSEA, everything is still relative (and run in hybrid mode with vector checked)

 

Prepar3D settings:

The basics are that I have 4096 textures installed.. autogen is on "normal", scenery complexity at 3/4 slider, weather 70 miles, max coverage, traffic at 4% on cars only and minimal airplanes at airports  (in fsx i was at 7% cars.. 4% seems less of a hit.. going from 0% to 7% for instance, i lost 6 fps at my resolution)

 

Here are the screen shots and i've attached my Prepar3D graphic settings file I've ended up sticking with (so far).

workingfine.CFG.txt

 

And now for my resultant tweaks/settings that correspond from the images:

 

 

ADDED/Tweaked:  usepools=0  **adding this may have boosted fps by 0.5-0.7 fps, placebo maybe, default is (no entry for buffer pools by default)?

ADDED/TWEAKED:  TEXTURE_BANDWIDTH_MULT=120  (default of 30), adding this may have boosted fps by 0.6 fps yeah placebo

Per the image:  TEXTURE_FILTERING=16

Per the image:  MSAA=4  **changing to a 2 was horrid on shimmering but fps went up**

SSAA=0  **entry has no matching graphic entry nor does it seem to affect anything**

Per the image:  VSYNC=1

FXAA=0 **changing this made no difference in shimmering**

Per the image:  TRIPLE_BUFFER=1

Per the image:  UPPER_FRAMERATE_LIMIT=0

Per the image: TEXTURE_MAX_LOAD=4096

Per the checkbox TESSELLATION=1

Per the slider TESSELLATION_FACTOR=90 (max)

ADDED AffinityMask=254

Per the slider LOD_RADIUS=6.500000

Per the slider WATER_DETAIL=2

ADDED:  FIBER_FRAME_TIME_FRACTION=0.1  **This was a bigger booster.. 0.05 lowered fps, but 0.1 caused them to raise relatively by 5 fps***

per the slider CLOUD_DRAW_DISTANCE=4 (in FSX this same position was a 3 for value)

Per the image:  VolumetricFog=0  **setting this to ON resulted in some loss of fps in heavy clouds**

 

Some notes on what i found above..

 

Affinity mask really did help, by adding it here are some of the ones I tested and I tried many values here based on charts online:

Example on the 40 second tests:

AM none: 39.0 (used 5 cores from what i could see in Performance Monitor) **but stuttering was BAD at around 45 seconds

AM 1364: 38.55 (smoother, but still these micro stutters, cant get rid of them though)

AM 1365: 37.32

AM 254:  37.47 fps and smooth with minor micro stutters

AM 255: 31.8

AM 4092: 31.7

 

I settled on using 254 for now.. on longer duration tests, it was slightly better where there was previously more stuttering.  This combined with the fiber setting made a world of a difference.

 

Even with my settings as I have them now, my GPU still hits 99% as you can see below.. this is NOT using 4XSGSS as well.. its using the 4X Bias option for AA and multisample set to off instead of 4SGSS (9 fps hit with this).. 2xSGSS looks bad, better to leave off in my case.. someone in another thread suggested the 4X Bias option and it looks decent.. yes there is some "distant" shimmering, but no real "flickering" at least by cars or airplanes on the ground.. 

 

 

Inspector Settings (final ones i'm using for now):

 

So now I can dig into the nitty gritty of what I found, a little more than already mentioned.. again 4xSGSS was too much of a hit on my system at this resolution, so for now I'm using the bias 4x setting above.. i took a c90 stuck it in a severe thunderstorm, heavy clouds situation above KSEA and it holds its own, fluctuating from 18-24 fps with the full on rain  (which the rain graphic still has a cone effect not filling the widescreen).. so this made me content.. this being one of the hardest hit areas.. I also tried NYCx.. i can fly by the new tower building in a p92 (light vc basically) and maintain 19 fps.. in FSX it was holding around 13.. again, happy here.

 

More particulars..

 

My full 120 second test with all above settings..

32.64 fps (unlimited fps with above settings)

37.00 fps (unlimited fps with FTXg hybrid UNCHECKED) with PNW installed

29.75 fps (set fsx fps to 36 fps and NI to 0.25 for 144hz matching rate)

 

My 40 second tests:

39.05 fps (unlimited fps)

 

Test results showing what changes caused increase/decrease in FPS:

 

Turning Tessellation off:  LOSS 1.85 fps (interesting result)

Going from Tessellation low to max:  INCREASE 0.34 fps  

 

And turning tessellation to off resulted in dark water, shimmering water and overall darkness.. as seen in this gif image animation below:

 

Putting fibertimefraction to 0.1 from no setting GAINED 5.5 FPS

 

Setting FXAA to on in FSX resulted in blurry textures and only a 0.10 fps Gain (leaving off for sure).

 

Lowering resolution to 2048, strangely resulted in a drop in FPS in my system.. i may need to retest this one.. past tests i've done comparing screen shots shows that runway and scenery has a "blur" when going from one extreme to sharp 4096, with in my case minimal impact.

 

Going to 4SGSS as mentioned, drops the 40 second test from 39 ish to 29 or so.. (9+ fps hit).. setting it to 2SGSS (ugly) was about a 5 FPS drop

 

Switching water from medium to max resulted in around a 4 FPS LOSS (off the best 120 second test of around 32.5 fps above).

 

I tried 4xMultisample in NI as well as other options there, none really seemed to offer the performance + quality of 4x bias, but maybe i was just seeing things on quality.

 

Mesh resolutions and complexities had very little impact on FPS.. going from autogen at 2 to 3 seemed to lose 4-6 fps, so i've left it on 2 for now.  Of course terrain vegetation shadows will kill fps as well.

 

I've got a few issues to work out but its looking like a solid future for me to roll to P3D completly here.. certainly awaiting the real Orbx PNW installer for P3D v2, which will spice things up in that region.

 

One issue i have is that on most flights after i start them (press P key to start), the rudder goes full left.. not sure why.. i've calibrated my ch flight yoke and pedals in the ch control manager and everything seems fine.

 

One test to compare here again.. is the Word Not Allowed AA Clouds test (dx10 in thread).. (its 5 layers of cumulus clouds that truly shows how limited the GPU can be with higher AA) I ran that in my thread over 60 seconds and in my test in FSX with 36 fps internal limiter and 0.25 fps sync and had results that were something to effect of 7.30 with a 670 gtx, 9.21 or so with the 780 and 12.05 on average with the 780ti, but this was with 4SGSS.. when i ran the same test in FSX with supersampling instead, it was around 28.7..

 

Here with p3d and the 4x bias with multisample being OFF (for now) and 36 internal limit.. i'm at 31.58 fps (32.50 unlimited).. an improvement but hard to tell with new prepar3d features.. If i turn on 4SGSS this becomes 16.21 (15.98 if unlimited) comparable to the 12.05 or so I got with FSX and 4SGSS.  (i will run with unlimited frames ordinarily to prevent stutters but keeping all tests equal i ran this at 36 fps locked).

 

If your interested in running this saved flight (AA clouds test).. the file can be found here.

 

While a good multilayer real world cloud test for KSEA can be found here.  It has a thick cloud layer from 4k up to 10k or more.  what i've done for this one, set throttle to max.. load it up, let it fly for 60 seconds.. record results.. with fsx i may be around 18 fps on average, again ftxg+pnw in both cases.. and with my tweaks above i'm now around 30 fps with multisample / off.  This is just the default cessna (which i copied over)..

 

Like I said before I used a carenado c90 around this same airport, switched to a less drastic weather/thunderstorm weather scenario with rain and now i'm 18-24 fps due to the rain i guess and a higher res cockpit)

 

Lastly, the cpu utilization of the prepar3d process even at 4.4ghz, is only around 35% max (this at AM of 254) with an overall max of around 45% for the whole cpu and 7 cores in use for AM of 254.  Which corresponds to tests I've done before at least with fsx, showing that going from 3.9 ghz turbo to 4.4ghz made little improvement in fps sadly.  

 

The GPU hits 82F at 73% fan speed using Evga Precision.. 82F being the throttle point.

 

I cant seem to get rid of micro stutters in my test flight at least.. they are very very small, but you can certainly feel them there and even see them.. LM is working on 2.1 and according to their forum and many on here this is one of many bugs that may (finally) be fixed going forward.

 

As with FSX when going to high 3 screen resolutions, only a few choices can be had for getting decent fps with heavy clouds and AA.. one is to of course NOT use 4SGSS (or dont fly in heavy areas like KSEA or NYC, as my fps in FSX with 4SGSS elsewhere even with moderate weather were still 25-30 fps not 18-24 as noted above), another is to choose the next lower res.. 4800x900, but its fuzzy looking since its not native.. or another to replace screens with 22" screens (i'm sold on 24" not happening) or maybe one day Oculus rift

[Waterman981 2]

Interesting observations. While my system isn't the same, I'll try playing with some of the things you set. I know I carried over my FTFF of .2 from FSX (which I initially set with my old video card), I'll have to try a lower setting.

[theskyisthelimit 48]

Interesting observations. While my system isn't the same, I'll try playing with some of the things you set. I know I carried over my FTFF of .2 from FSX (which I initially set with my old video card), I'll have to try a lower setting.

 

yeah with my setup anyway.. going from 0.2 down to 0.1 resulted in a gain of about 1 fps (not much but still some) and not having the entry alone, as mentioned loss of 5 fps.

 

Also added a note that i get about 5 fps higher if i uncheck hybrid with ftxg.  Makes me wonder with the real PNW install that is forthcoming, if the fps will jump back up with hybrid checked again.

 

One thing i've never been sure on, was the "blurries".. whether or not i'm really having blurries or just scenery in a distance that is outside the LOD (in this case 6.5).. though as you can see in this image there are two "blurry" spots highlighted in yellow boxes.. my question is whether this is the norm or not.. i'm assuming it is, as no settings changes have ever gotten rid of this type of appearance in either fsx or p3d (disregard the broken texture via ftx.. even in non hybrid i think they still are an issue unless you toggle say 4096 to 2048 and back up):

 

 

[denali 74]

Awesome Info, thanks.   

 

I am also running 5760 x 1080 on a 780 acx and 3077 at 4.4 with 32GB of 2400 RAM.   I can get my board up to a stable 4.8, but I lost my turnings on a carelessly done bios update; that little bit makes a difference.  I have yet to OC the card.

 

I wish I had more time, but I have my first son on the way, and I've decided to wait until .05 to start tinkering again.  My issues are around high res global meshes and .33 meter texture, downloaded with FSEarthTiles.   I don't care for autogen that much.  After about 1200 agl satellite textures are the way to go.

 

If anyone wants a quick and dirty tutorial on FSEarthTiles, let me know.  I've got it down pretty nice.

 

But my frame-rates are lousy where I want them to be nice, and I think it's the multiple texture issue LM is aware of.  I run my tests in the san francisco area or at lake tahoe, and my rates are stuttery below 15fps.  Elsewhere I get over 20 smooth, but strangely if I remove the mesh it doesn't improve.  I am I may have broken something, cause I don't think it started out that way.  I'll be doing a complete re install once .05 is released.  But I wish I could collaborate more.

 

Fun thing is to see how many of us have these uber high end systems and it's blaze in this forum; no other PC hobby like it

[Yuji Y 0]

Thanks a lot for the very interesting report!!

 

Being inspired by your report, I juggled the setting again and I could reduced stutters significantly!

I had been suffering stuttering, which, for example, had been making using DHM in Opus impossible.

My system is i7 4770k OC to 4Ghz, 760GTX 2G OC, 3 moniters with 5210x1050. I fly with Q400, REX, FTXG, Opus. I moniter with Process Explorere.

 

## Here what I did. (Note that this is only my non-rigorous observation).

 

1. I used to set MASSx2 SGSSx2 in NI. Runways looked awful and heavy clouds and (many trees?) easily made my GPU usage maxed out causing stuttering. I understood that SGSS was too heavy for my humble video card, so I set instead BioShock option=On, AA=AA_MODE...4X_BIAS, SGSS=Off. Then, I got runway-notches significantly reduced and I could fly heavy clouds with almost without stuttering.

 

2. I had not toughted Prepar3d.cfg at all because I do not believe so much cfg tweaking. Since I got P3Dv2, I had been concerned about FPS spike/drop (about 10FPS) occuring nearly every 5 second. When I put AM=15, then this spike seems to happen much less. I have absolutely no idea why.

 

3. I did not get much improvement with FFTF. When I set 0.1, I got stuttering. With 0.2, FPS became higher by ca 5fps but my GPU usage went up about by 15% nealy up to 95% and I get some stuttering in cloulds. So I stick to the default. I do not why FFTF is affecting GPU usage. Is FFTF something to do with balancing CPU and GPU usage?

 

/YY          

[spirit_66 2]

Well, thanks for sharing your observations. I'm still reading and not finished yet.

 

Just stumbled over your Affinity mask setting and I think that for your CPU with 6/12 cores and HT on the right AM would be 1364. I use it with success.

 

Now I'll go on reading.

Spirit

[theskyisthelimit 48]

Another update..

 

Previously i had the wrong figure for running without hybrid enabled.. i had some process bringing the fps down on that value..

 

The corrected ones should read:

 

Hybrid mode off (pnw installed) with vector lighting:  

43.02 fps

 

Hybrid enabled with vector:

32.50 or so..

 

North America mode:

32.50 

 

So the difference is actually a bit more.. 11 fps.. enough to justify tinkering more with 4SGSS perhaps, though i didnt see a huge elimination in shimmers with it.. or maybe upping some of my sliders and options

 

I just hope that when the real P3D installer is out for PNW that the difference in fps isnt this drastic.. I'm going to retest FSX with each and see how it was and report back another time.

Well, thanks for sharing your observations. I'm still reading and not finished yet.

 

Just stumbled over your Affinity mask setting and I think that for your CPU with 6/12 cores and HT on the right AM would be 1364. I use it with success.

 

Now I'll go on reading.

Spirit

No problem.. just hoping the info is useful to some.  Sorry it was so long lol

 

Yeah deeper down in my writeup, i had tossed back and forth between AM 1364 and 254... both work fairly well though for some reason a little less stutters (micro) with 254.

Thanks a lot for the very interesting report!!

 

Being inspired by your report, I juggled the setting again and I could reduced stutters significantly!

I had been suffering stuttering, which, for example, had been making using DHM in Opus impossible.

My system is i7 4770k OC to 4Ghz, 760GTX 2G OC, 3 moniters with 5210x1050. I fly with Q400, REX, FTXG, Opus. I moniter with Process Explorere.

 

## Here what I did. (Note that this is only my non-rigorous observation).

 

1. I used to set MASSx2 SGSSx2 in NI. Runways looked awful and heavy clouds and (many trees?) easily made my GPU usage maxed out causing stuttering. I understood that SGSS was too heavy for my humble video card, so I set instead BioShock option=On, AA=AA_MODE...4X_BIAS, SGSS=Off. Then, I got runway-notches significantly reduced and I could fly heavy clouds with almost without stuttering.

 

2. I had not toughted Prepar3d.cfg at all because I do not believe so much cfg tweaking. Since I got P3Dv2, I had been concerned about FPS spike/drop (about 10FPS) occuring nearly every 5 second. When I put AM=15, then this spike seems to happen much less. I have absolutely no idea why.

 

3. I did not get much improvement with FFTF. When I set 0.1, I got stuttering. With 0.2, FPS became higher by ca 5fps but my GPU usage went up about by 15% nealy up to 95% and I get some stuttering in cloulds. So I stick to the default. I do not why FFTF is affecting GPU usage. Is FFTF something to do with balancing CPU and GPU usage?

 

/YY          

 

Thanks

Ill have to try the bioshock option in NI.. do you happen to have a screen shot of your NI settings?

 

Yeah the AM tweaking takes some tests.. and they have to be consistent each time.. thats why i like fsmark07 because it allows for consistent testing.. i went through about 10 values before I found the best for both fps AND stuttering

 

Curious.. you dont seem to be maxing out your GPU like I do with my 780ti.. i wonder why this may be..

 

When you run around busy airports like KSEA, KFSO or NYC and throw in thick layered clouds and rain and a decent plane VC (c90 or something more elaborate).. are you able to maintain decent fps like my last round that was 18-24 fps with rain/clouds KSEA.. i expect this value to jump way up now that i found that hybrid mode unchecked results in an 11 fps boost on fsmark07

Awesome Info, thanks.   

 

I am also running 5760 x 1080 on a 780 acx and 3077 at 4.4 with 32GB of 2400 RAM.   I can get my board up to a stable 4.8, but I lost my turnings on a carelessly done bios update; that little bit makes a difference.  I have yet to OC the card.

 

I wish I had more time, but I have my first son on the way, and I've decided to wait until .05 to start tinkering again.  My issues are around high res global meshes and .33 meter texture, downloaded with FSEarthTiles.   I don't care for autogen that much.  After about 1200 agl satellite textures are the way to go.

 

If anyone wants a quick and dirty tutorial on FSEarthTiles, let me know.  I've got it down pretty nice.

 

But my frame-rates are lousy where I want them to be nice, and I think it's the multiple texture issue LM is aware of.  I run my tests in the san francisco area or at lake tahoe, and my rates are stuttery below 15fps.  Elsewhere I get over 20 smooth, but strangely if I remove the mesh it doesn't improve.  I am I may have broken something, cause I don't think it started out that way.  I'll be doing a complete re install once .05 is released.  But I wish I could collaborate more.

 

Fun thing is to see how many of us have these uber high end systems and it's blaze in this forum; no other PC hobby like it

 

Congrats on your first on the way.. i have two in elementary school.. good fun times

 

I have FSEarthTiles.. i had started to download some regions with it, but it seemed like for a decent coverage of one area.. like say the SW corner of PA even, it was a tremendous amount in GB.. unless I wasn't setting the settings correct.. but this was awhile back.. i havent had the time yet to retry fsearthtiles but certainly would, because for now all i do at least with fsx (and i hope this still works in p3d) is use tileproxy and love it (for 2500 and up flying).

 

yeah there is definitely nothing like a flight sim hobby.. especially one that requires crazy systems like we build.. enthusiast hobby at its best.

[theskyisthelimit 48]

I was able to run the FSX comparison between hybrid on and off.. interesting results:

 

Hybrid Off:

32.70

 

Hybrid On

34.27

 

So a GAIN of 2 fps.

 

So with p3d it was a LOSS of 2 fps by having it on (43 vs 32), again likely due to incompatibilities i would hope.. rather encouraging results... although it seems as if estonia or some reinstalls of addons may have partially hosed my FSX at first.. Managed to restore things ok.. i think for future installs with estonia running i will see if i can remove the drive letter for the old fsx install.. and strangely my old NI profile for FSX seems to bring the fps way down.. i had to remove it to run these tests (it was 15 fps otherwise, no matter what i did).. doesnt matter much, as I really dont want to look back if possible.

 

Also i reran checks on the fsx process and cpu use in fsx with AM at 1364.. fsx like p3d doesnt go above 30% or so.. though the overall cpu use said 51% but i cant pinpoint which processes lead to that..

 

FSX GPU use.. (after fixing NI by reimporting).. i set NI to 4SGSS even with this GPU load was 82%, memory used 1458, temp 80C, without SGSS the load was even lower, around 73%.. this all compared to prepar3D where the GPU use WITHOUT 4SGSS is at 99%.. so its definitely using more GPU power, so I guess this is a good sign.

[theskyisthelimit 48]

I also retested with hybrid mode off with 4SGSS..

 

Hybrid off (multisample/off):

43.30 fps

 

Hybrid off (4SGSS)

28.22 fps

 

Thats a 15 fps hit to use 4SGSS.. taking a look at how it performed visually, it was an improvement, less "fringe shimmering" but runways still flicker.. i'd call it a 50% reduction from "mutli-sample/off" or any of the other modes that arent supersample or sgss (i saw no difference with aa_mode 4x, etc, tried several of these).. so you get a slight reduction from that baseline but at a premium, in my case 15 fps difference for this test.

 

Then I reran my C90 Thunderstorm with rain situation (above the airport test).. with 4SGSS it was 18-21 fps close to the airport (19 avg).. turn on volumetric fog, the drop is about 4 FPS more on top of this.. turn off wind and rain.. the average becomes 21 up close to the airport.

 

Same c90 thunderstorm with MultiSample/off.. 31 fps on average (close to the airport), with volumetric fog checked (distant shimmering visible more so, but the frame rate much higher)..

 

I have to say that even with 4SGSS it seemed "smooth" for only averaging 17-23 fps for some reason, but 31 is certainly a better overall performer.  Also, having volumetric fog made my other thick cloud test above KSEA look light night in day.. without it checked.. i just had a solid white unrealistic block of clouds from 4k on up.. turn it on.. boom.. the clouds were multi shaded, distinctly unique and much more realistic with gray shading.  

 

However.. having 4SGSS really kills the frames for this multi layer test.. 31 without.. turn on 4SGSS + volumetric fog for the realism.. its only 10 fps as you fly over the runway.. turn off fog.. it becomes about 16-17 fps.. Going to try to recreate the clouds without volumetric.. perhaps that was a glitch being a solid white mass.

 

EDIT: recreating didnt seem to change the "solid white mass" issue without volumetric turned on..

 

The original settings were:

 

Layer1:  cloud type stratus base 1430, tops 10744, coverage 8/8 ;   Layer2: cirrus tops 43132 base 41,158 and coverage 1/8..

 

I deleted both layers, recreated just layer1.. it still shows as a solid white mass (rex4 textures installed).

EDIT: seems if i hit clear weather, clear skies.. then go back in and recreate layer1, it now is NOT a solid mass.. success.. so now on to discover the merits of leaving volumetric fog on vs off.. (and yeah stratus clouds at  low altitude layer1 is silly, but it was a "stressful" test to conduct on the GPU:)

 

The other scenario.. NYCX.. flying past the new tower in the p92 eaglet (VC).. avg framerate is about 20 fps with 4SGSS.. add "major thunderstorm" it becomes 10-11.. Switch to Multisample/off.. avg framerate is 20-24 or so.. add in the storm, this becomes 17-21.. this is one add-on where you can see a difference at least with the mass amount of buildings with SGSS vs not on

[Slayer 190]

I have a similar setup and have about the same results. I've been heavily using EVGA precision to monitor GPU and watching system CPU loads. Once these get close to 100% you need to back down settings a tiny bit. Ideally you don't want to be at 100% before going into heavy airports or sceneries.

 

Just watching FPS really tells you nothing if your CPU and GPU are maxed out. You need to monitor all 3 and find a harmonious blend. There is no cut and paste settings when you run 3 screens.

 

On microstutters: I really only notice them when my GPU is running at 99% with a little headroom they almost disappear. Having to use a FFTF of .1 can result in blurries because you cut the time allocated to loading by 2/3. Yes you get more FPS but at the cost of image quality and delayed loading of textures ir textures "popping" as you are flying near. Suggest you instead turn some settings back and unload the GPU below 99% Mostly in FSX you could barely get above 50% GPU load unless you use crazy AA settings so just forget about what worked in FSX...It's not the same.

[theskyisthelimit 48]

I have a similar setup and have about the same results. I've been heavily using EVGA precision to monitor GPU and watching system CPU loads. Once these get close to 100% you need to back down settings a tiny bit. Ideally you don't want to be at 100% before going into heavy airports or sceneries.

 

Just watching FPS really tells you nothing if your CPU and GPU are maxed out. You need to monitor all 3 and find a harmonious blend. There is no cut and paste settings when you run 3 screens.

 

On microstutters: I really only notice them when my GPU is running at 99% with a little headroom they almost disappear. Having to use a FFTF of .1 can result in blurries because you cut the time allocated to loading by 2/3. Yes you get more FPS but at the cost of image quality and delayed loading of textures ir textures "popping" as you are flying near. Suggest you instead turn some settings back and unload the GPU below 99% Mostly in FSX you could barely get above 50% GPU load unless you use crazy AA settings so just forget about what worked in FSX...It's not the same.

 

Curious.. so you have managed to get your GPU use down below 99%?  What settings did you modify to manage this.. i havent found many that made a huge difference.. are you doing 4SGSS or just off.. Yeah.. i wasnt sure on the image i posted if that is truly a "blurry" the result of FTF being too low or just the way it is, i suppose a similar screen shot with higher FTF (or visually looking) should prove the case.  It would be nice to put more stress on the cpu, since its only 30% on the prepar3d.exe process too.

[Yuji Y 0]

 It would be nice to put more stress on the cpu, since its only 30% on the prepar3d.exe process too.

 

A short notice. Do you use Process Explorere (PE) for monitering CPU stress? I read some other threads (I do not remenber which but you can surely find one!) that the number is kind of misleading. I think it was something to do with how PE calculate CPU usage of logicla cores vs physical cores. I have 4 physical core i7 ( that has 8 logical core I think), and when I see CPU 35% at PE, Windows Taks Manager shows that 3 physical cores are nearly 100%. Having AM=15, 4 physical cores seems to be fully used only when loading somthing, then PE shows ca 45%. I used to double the number (35% -> 70%, 45 ->90), then it is kind of making sense. 

 

This was how I remenber but please double check with other threads.

 

/YY  

[theskyisthelimit 48]

A short notice. Do you use Process Explorere (PE) for monitering CPU stress? I read some other threads (I do not remenber which but you can surely find one!) that the number is kind of misleading. I think it was something to do with how PE calculate CPU usage of logicla cores vs physical cores. I have 4 physical core i7 ( that has 8 logical core I think), and when I see CPU 35% at PE, Windows Taks Manager shows that 3 physical cores are nearly 100%. Having AM=15, 4 physical cores seems to be fully used only when loading somthing, then PE shows ca 45%. I used to double the number (35% -> 70%, 45 ->90), then it is kind of making sense. 

 

This was how I remenber but please double check with other threads.

 

/YY  

I have at times used PE.. though in the last few cases, just the task manager area of 8.1.. and the monitor tool.

I was a little bummed to see that one of my airports i use.. hasnt been updated even with Prepar3D.. in terms of positioning and the correct runway numbers.. hoping I can still use the airport designer tool to adjust this, as i planned on doing with fsx (airport 9G1).

 

It should be 17/35 but they havent updated it.

[theskyisthelimit 48]

I managed to get UT2 (ultimate traffic) working in Prepar3D v2.0.. it even shows up in the menu (not using legacy mode, but fsx.exe dummy file was needed to get the main configuration app to function.. only issue is that so far, at least with sliders at 100% each in UT2 (prepar3d traffic for GA/airlines at ZERO).. i take a 9 fps hit by having it load in the exe.xml..

 

If i disable ut2 in the exe.xml, i'm back to 41.11+ fps (hybrid not checked in ftx).. i'm going to play with lowering the sliders down a tad.. but even at 100%, if say in KSEA, top down view.. zoom out a few times.. only see about 2-3 aircraft in the air at a time (Supertrafficboard showing about 200 overall in the whole sim with A:8 D:46 ish at KSEA).  Even with only a few around the fps dips more than I expected or saw with FSX.. 

 

UPDATE: changed from 100% on sliders in the UT2 interface to 63 airliners and 40 GA, the rate became 38.08.. nice happy medium here.

 

Again, this is all working under Win 8.1 x64.. later today.. updating to 2.1

[theskyisthelimit 48]

As i just posted in the 2.1 thread here...

 

Installed clean full 2.1.. with ftxg+v and rex4 textures as i did with 2.0...

 

With 2.0 i was getting 41-43 fps in fraps, with unlimited frames and the settings i have in this thread.  (using fsmark07 test)

 

I started off with no tweaks to prepar3d.cfg and only got around 30.53 ..

 

So i thought.. ok, ill add in the tweaks to cfg as i did before and got a nice jump.. NOPE..

 

Fps with cfg tweaks only rose to 31.53.. somewhat studdery..

I'm using the same NI settings as in this thread as before too..

 

Rather unhappy so far with a loss of nearly 11-12 fps all things held equal. 

 

Hopefully its something simple that needs adjusted..