Hi Dave, sure, we have almost identical specs. Well as far as tweaking is concerned, I haven't, the only thing i have in the cfg is AF=15. It may be a drag and please don't bother if it is, but if you could share your slider settings and also your graphics settings I'd be extremely grateful.

 

Hey Howard, try this setting: AffinityMask=242

 

I found that to work much better on my system. I tried it after Beau at LM posted the following in this LINK:

 

 

 

A few other observations we have made in our testing:

- For some processors core 0 is clocked a bit higher. There isn't actually a dedicated windows core. Windows and other windows applications may schedule work on any core at any time. The OS does however try to load balance work, so if you leave a core open, it is much more likely to schedule work on that core. I don't think it considers a core open unless both HT cores are open. This is why masking off an entire core tends to help with stutters. We haven't done a lot of testing to back this up, but some initial findings suggest that you can leave open whichever core you choose. It might be better to use the first core and leave open a different core:

eg for a quad core with hyper threading try: 241/242 ( 11110001 / 11110010 ) or 61/62( 00111101 / 00111110 ).

We also noticed that the AMD Phenom processor doesn't like having any cores masked off as anything but using all cores caused odd stutters which seemed to be related to hardware and OS rather than simply having one less core to play with.

Beau Hollis

Rendering System Lead - Prepar3D® Team

 The Majestic Q, I find isn't doesn't' exactly pan around very fast as it is, but it's quite smooth and more than adequate. I do have cloud shadows on and while it can hit the frames pretty hard, it's still very flyable and heads and shoulders above FSX.

 I did a flight last night at dusk from SFO to KJAC with ASN weather. The initial climbout did have lower numbers fps (high teens- mid 20's) wise but still very smooth and enjoyable.

OK, that's what I see.  The difference is that one person's 'very smooth & enjoyable' is another person's 'not quite good enough'.   There's 'smooth enough', then there's 'absolutely floating buttery smooth!'  

 

This is why we need a realtime utility that lets the user prioritize what's important to them.   I posted about this on the P3D forum under suggestions for new product features.   Since LM understands the relationships involved they are arguably positioned to do the best job at something like this.  The logic would be something along these lines:  When GPU utilization (might be better than FPS, or not) goes above X, lower cloud draw distance by 1 click, etc.   Lower cloud draw distance before lowering autogen density, etc etc.   We can all agree that a frame rate below a certain value becomes a real problem.  Smoothness is not the issue when frames go below a certain value for obvious reasons. 

 

Something like this as a realtime modulator could go a long ways towards giving all users the best possible experience given the hardware and visual priorities they might have.  I have no idea what features could be modulated in realtime, but I'm guessing there are quite a few that can potential impact 'performance', which we might agree to define as:

 

1.  Freedom from stutters (or smoothness)

2.  FPS

3.  Texture update rate/visual clarity

 

There have been optimizers I know for FSX, but I don't think they worked like this, dynamically, changing the sim during runtime.  I think LM could have a very vested interest in this too, for their commercial users who want to be able to count on something.  I have generally excellent performance, but again, given the wrong situation performance can tank as with all users.  I like looking at the GPU utilization because there does appear to be an issue once the GPU gets up to 100% utilization (GPU-z). 

 

BTW, no options to add images now here?

There have been optimizers I know for FSX, but I don't think they worked like this, dynamically, changing the sim during runtime.

 

 

Its a good idea, but I wonder what sort of overhead this would add, plus I can imagine the constant juggling of assets or effects could lead to stuttering... but, I could be very wrong. I know the Unreal engine had a Min Desired Frame Rate setting in its config, that was supposed to dynamically cull certain elements if it fell below the prescribed limit, but I'm not sure it ever worked; I could certainly see no change in visuals or frame rate with it.

 

I've been testing some scenarios in both FSX and P3D and have concluded that there is only a marginal lead in performance in FSX.. but this is with P3D's new features enabled, so Cloud Shadows, Volumetric Fog, Pop Free AGN ect, which I'm not going too lose... I'm sure turning them off would allow me to match FSX but that would be pointless, as its what defines this platform from the old. Maybe V2.2 scales badly with resolution, so going from 1080 to 1600 makes a far greater impact than FSX... but I see no major disparities, and have a similar spec to Howard's... weaker in the GPU.

 

These are my settings.

 

 

[GRAPHICS]

TEXTURE_MAX_LOAD=1024

NUM_LIGHTS=4

MIPMAP_VC_PANELS=0

TESSELLATION=1

COCKPIT_HIGH_LOD=1

AIRCRAFT_REFLECTIONS=1

LANDING_LIGHTS=0

DAWN_DUSK_SMOOTHING=1

LENSFLARE=1

HDR=1

BLOOM_LEVEL=2

SHADOW_QUALITY=2

TERRAIN_SHADOW_CAST_DISTANCE=2

CLOUD_SHADOW_CAST_DISTANCE=2

OBJECT_SHADOW_CAST_DISTANCE=2

CONTENT_NO_SHADOW=0

INTERIOR_SHADOWS_CAST=1

EXTERIOR_SHADOWS_CAST=1

SIMOBJECT_SHADOWS_CAST=0

TERRAIN_SHADOWS_CAST=1

VEGETATION_SHADOWS_CAST=0

BUILDING_SHADOWS_CAST=1

CLOUD_SHADOWS_CAST=1

INTERIOR_SHADOWS_RECEIVE=1

EXTERIOR_SHADOWS_RECEIVE=1

SIMOBJECT_SHADOWS_RECEIVE=0

TERRAIN_SHADOWS_RECEIVE=1

VEGETATION_SHADOWS_RECEIVE=1

BUILDING_SHADOWS_RECEIVE=1

SHADOW_NUM_CASCADES_LOW=4

SHADOW_NUM_CASCADES_MID=5

SHADOW_NUM_CASCADES_HIGH=7

SHADOW_NUM_CASCADES_ULTRA=12

SHADOW_LOG_PARTITION_LOW=0.850000

SHADOW_LOG_PARTITION_MID=0.850000

SHADOW_LOG_PARTITION_HIGH=0.850000

SHADOW_LOG_PARTITION_ULTRA=0.850000

EFFECTS_QUALITY=1

EFFECTS_DISTANCE=1

[DISPLAY]

SKINNED_ANIMATIONS=1

TEXTURE_BANDWIDTH_MULT=30

UPPER_FRAMERATE_LIMIT=30

TEXTURE_FILTERING=16

MSAA=4

SSAA=0

FXAA=0

VSYNC=1

TRIPLE_BUFFER=1

WideViewAspect=0

FullScreenExclusive=1

FullScreenAutoFill=1

[PANELS]

PANEL_OPACITY=0

[sCENERY]

IMAGE_COMPLEXITY=5

[TERRAIN]

LOD_RADIUS=6.500000

TESSELLATION_FACTOR=85

MESH_RESOLUTION=23

TEXTURE_RESOLUTION=28

AUTOGEN_VEGETATION_DENSITY=4

AUTOGEN_BUILDING_DENSITY=3

DETAIL_TEXTURE=1

WATER_REFLECT_CLOUDS=0

WATER_REFLECT_USERAIRCRAFT=1

WATER_REFLECT_SIMOBJECTS=1

WATER_REFLECT_TERRAIN=1

WATER_REFLECT_AUTOGEN_VEGETATION=1

WATER_REFLECT_AUTOGEN_BUILDINGS=1

WATER_REFRACT_CLOUDS=0

WATER_REFRACT_USERAIRCRAFT=0

WATER_REFRACT_SIMOBJECTS=0

WATER_REFRACT_TERRAIN=0

WATER_REFRACT_AUTOGEN_VEGETATION=0

WATER_REFRACT_AUTOGEN_BUILDINGS=0

ENABLE_BATHYMETRY=0

WATER_CLARITY=30

WATER_DETAIL=3

[WEATHER]

CLOUD_DRAW_DISTANCE=5

DETAILED_CLOUDS=1

OPTIMIZE_DENSE_CLOUDS=0

CLOUD_COVERAGE_DENSITY=8

THERMAL_VISUALS=0

DownloadWindsAloft=0

DisableTurbulence=0

DynamicWeather=2

VolumetricFog=1

[TrafficManager]

AirlineDensity=15

GADensity=25

FreewayDensity=10

ShipsAndFerriesDensity=15

LeisureBoatsDensity=20

IFROnly=0

AIRPORT_SCENERY_DENSITY=1

[AContain]

ShowLabels=1

ShowUserLabel=0

ShowLabelManufacturer=1

ShowLabelModel=1

ShowLabelTailNumber=0

ShowLabelDistance=1

ShowLabelAltitude=1

ShowLabelAirline=0

ShowLabelAirlineAndFlightNumber=0

ShowLabelFlightPlan=0

ShowLabelContainerId=0

ShowLabelAirspeed=0

ShowLabelHeading=0

LabelDelay=1000

LabelColor=FFFF0000

 

 

 

 

 

absolutely floating buttery smooth!

 

I like that term :lol: ... I think Tom needs to adopt that as AVSIM's motto.

 

 

 

Q400 must be very frames friendly.

 

It is very frame friendly ... the only issue I have with it are "jitters" (not stutters) in external view and it slews thru the P3D world in a very uneven manner.  MJC Q400 goes outside the "normal" bounds of the flight model ... I think someone says it "slews" thru time and space, don't know how accurate that is, but it's exactly how it feels to me.  The MJC Q400 supports threading so I'm almost certain this could be why I'm seeing inconsistent flight behavior under 2.2 ... under P3D v2.0 the MJC Q400 had no issues (other than turning off mipmap cockpit).  So change in threading in P2.2 doesn't seem to meld well with how the MJC Q400 works.  I've followed the many suggestions but I'm still unable to get a good consistent flight experience with the MJC Q400 - shame because it's a wonderful aircraft ... fingers crossed for the P3D version soon.

 

Cheers, Rob.

I like that term :lol: ... I think Tom needs to adopt that as AVSIM's motto.

 

 

 

 

It is very frame friendly ... the only issue I have with it are "jitters" (not stutters) in external view and it slews thru the P3D world in a very uneven manner.  MJC Q400 goes outside the "normal" bounds of the flight model ... I think someone says it "slews" thru time and space, don't know how accurate that is, but it's exactly how it feels to me.  The MJC Q400 supports threading so I'm almost certain this could be why I'm seeing inconsistent flight behavior under 2.2 ... under P3D v2.0 the MJC Q400 had no issues (other than turning off mipmap cockpit).  So change in threading in P2.2 doesn't seem to meld well with how the MJC Q400 works.  I've followed the many suggestions but I'm still unable to get a good consistent flight experience with the MJC Q400 - shame because it's a wonderful aircraft ... fingers crossed for the P3D version soon.

 

Cheers, Rob.

 

Hi Rob, thanks to everyone for their help with my issue, it's appreciated. I've just managed to follow some settings which were suggested to me by Dave_YVR and I'm pleased to say that things have improved far more than I could have wished. However, I am at a dead end with the Vsync issue. I simply cannot get it to work, no matter what I try. I've read that Aero needs to be started in Win 7, but for the life of me I can't get to the bottom of things. Can you suggest anything? Cheers

 

 

Maybe V2.2 scales badly with resolution, so going from 1080 to 1600 makes a far greater impact than FSX

 

If you think about what a shadow is and how it impacts pixel information, then it makes sense to see higher resolutions having more impact (more pixels to calculate appropriate color information - I think this is what Beau was referring to in his LM post).  Shadows are just calculations on pixel color information.  What I thought I saw the other night was cloud shadows being cast on another layer of clouds (cloud tops) ... I need to revisit that scenario because if P3D is calculating cloud shadows relative to EVERYTHING (including other clouds) in the scene/frame then it's easy to see computational requirements sky rocketing.

 

See my video here -- fast forward to 12:38 ... those look like cirrus clouds casting shadows on cumulus clouds to me ... if that is accurate then the calculations are just mind boggling.

 

 

Cheers, Rob.

 

 

I've read that Aero needs to be started in Win 7

 

Oh, do you have Aero enabled?  FYI, if you run the Windows 7 Performance Index and your system is deemed fast enough, Win7 will enable Aero.  Could you let us know what your Window 7 Performance Index is?

 

To get the Performance Index to run ... 

1. Click start icon in task bar

2. In the "search program and files" box enter "Performance Index"

3. Select "Check the Windows Experience Index"

4. Select the "Refresh Now" button or select the "Re-run the assessment"

 

This will give you 5 values with an overall score.  If your score is good enough it will enable Aero (or at least give an option to enable).  Can you share those 5 values with us?  Might help us identify an issue.  For example my work PC (which is an i7 3820 with a GTX 570):

 

Processor: 7.7

Memory: 7.9

Graphics: 7.9

Gaming Graphics: 7.9

Primary Hard Disk: 7.9

 

This doesn't tell us A LOT of information, but it will identify quickly and obvious performance issues.

 

Cheers, Rob.

I think it's important to remember or read again the Prepar3D Rendering and Performance FAQ - http://www.prepar3d.com/forum-5/?mingleforumaction=viewtopic&t=3920 as many functions have been reprogrammed to the GPU.  For instance:

 

FSX and/or Prepar3D 1.4 ran great on my rig. Why doesn’t v2?
Most likely you don’t have a powerful enough GPU to run our new features. We’ve offloaded tons of work to the GPU such as the terrain mesh generation and particle simulation. We’ve also added many new features such as shadows and 3d water that stress the GPU. You will get poor performance on low end GPUs unless the settings are tuned carefully. We’ve optimized for high end hardware because we want Prepar3D to be a viable platform for years to come. Hardware will keep getting better and cheaper.

 

I’m running on multiple monitors or at ultra high resolutions. Why is my framerate so low?
Back in 1.4/FSX, the system was CPU bound most of the time so resolution changes didn’t impact performance very much. Now that we are heavily tasking the GPU, you will likely see your framerate scale with your window resolution. If you want to stretch a view across multiple 1080p displays, you’ll need to dial back the settings or lower your screen resolution.

 

Best regards,

If you think about what a shadow is and how it impacts pixel information, then it makes sense to see higher resolutions having more impact (more pixels to calculate appropriate color information - I think this is what Beau was referring to in his LM post).  Shadows are just calculations on pixel color information.  What I thought I saw the other night was cloud shadows being cast on another layer of clouds (cloud tops) ... I need to revisit that scenario because if P3D is calculating cloud shadows relative to EVERYTHING (including other clouds) in the scene/frame then it's easy to see computational requirements sky rocketing.

 

See my video here -- fast forward to 12:38 ... those look like cirrus clouds casting shadows on cumulus clouds to me ... if that is accurate then the calculations are just mind boggling.

 

 

Cheers, Rob.

 

 

 

Oh, do you have Aero enabled?  FYI, if you run the Windows 7 Performance Index and your system is deemed fast enough, Win7 will enable Aero.  Could you let us know what your Window 7 Performance Index is?

 

To get the Performance Index to run ... 

1. Click start icon in task bar

2. In the "search program and files" box enter "Performance Index"

3. Select "Check the Windows Experience Index"

4. Select the "Refresh Now" button or select the "Re-run the assessment"

 

This will give you 5 values with an overall score.  If your score is good enough it will enable Aero (or at least give an option to enable).  Can you share those 5 values with us?  Might help us identify an issue.  For example my work PC (which is an i7 3820 with a GTX 570):

 

Processor: 7.7

Memory: 7.9

Graphics: 7.9

Gaming Graphics: 7.9

Primary Hard Disk: 7.9

 

This doesn't tell us A LOT of information, but it will identify quickly and obvious performance issues.

 

Cheers, Rob.

 

Hi Rob, sure...

 

Processor: 7.7

 

Memory: 7.8

 

Graphics: 7.9

 

Gaming graphics: 7.9

 

Primary hard disk: 7.9

So no obvious hardware issues ... good news.  Sounds like you'll just have to experiment with graphics settings and/or AffinityMask values.  

 

If you run other products in the background like TrackIR, GoFlight interface tools, ASN, EFB, video recording tools (i.e. FRAPS) this will interact with your AffinityMask setting.   What I mean by this, I run 4092 ... good for P3D improving hi-res texture loading BUT it can make TrackIR a little jittery and when I start recording with FRAPS I can see the texture system "slow down" ... as soon as I stop recording texture loading is back to normal.

 

So I can play around with AffinityMask settings to trade off some texture loading performance but gain some performance when recording with FRAPS.  So when setting AffinityMask you might also want to consider what else might be running that needs the a CPU core or two.

 

Cheers, Rob.

Hi Howard,

 

Couple of other suggestions which I have found to be helpful after sessions of fiddling around and consequent negative impact on frame rates:

 

Run: %LOCALAPPDATA%\Lockheed Martin\Prepar3D v2

And delete the 'Shaders' folder.

 

Run: %APPDATA%\Lockheed Martin\Prepar3D v2

And delete 'Prepar3D.CFG'

 

Run the sim and give it time to rebuild 'Prepar3D.CFG' and recompile the 'Shaders' folder.

 

[JOBSCHEDULER]   - added at head of Prepar3D.CFG
AffinityMask=14

 

NO OTHER CHANGES MADE TO THIS FILE

 

My System Specs: Alienware M18x (R1) with i7 2960XM, GeForce GTX 580M 2x2GB DDR5 in SLI, 20GB DDR3-1600, Monitor: 18.4" 1920x1080x32

Windows 7 Professional 64bit SP1

Component Details Subscore Base score
Processor Intel® Core i7-2960XM CPU @ 2.70GHz (Turbos to 3.3-3.7GHz)

Determined by lowest subscore
Calculations per second................................7.6
Memory (RAM) 20.0 GB............................... 7.7
Graphics NVIDIA GeForce GTX 580M......... 7.9
Gaming graphics 4096 MB Total available graphics memory.... 7.9
Primary hard disk 172GB Free (452GB Total)........................... 5.9
 

Affinity Mask - still experimenting, but 14 appears to be best for me (HT enabled, but only 3 of the 4 physical cores are mainly used with this value in Prepar3D).

 

Prepar3D v2.2.10438.0 (Settings):

 

FXAA - OFF

MSAA - 4 samples

Anisotropic - 4x (try it, you may be pleasantly surprised - I know I was)

Texture Resolution - 2048x2048

VSync - ON

Triple Buffering - Checked

Target frame rate - Unlimited

 

Level of detail radius - High

Tesselation Factor - Ultra

Mesh resolution - 5m

Texture resolution - 15cm

 

Scenery complexity - Dense

Autogen vegetation density - Normal

Autogen building density - Dense

Water Detail - Medium

Reflections - User Vehicle

Special effects detail and distance - High

 

Landing lights and Lens flare - checked

Shadow Map Count - Medium

Enable Terrain to Receive Shadows - checked

Terrain and Shadow Cast Distance - 0m

Cloud Shadow Cast Distance - 0m (worth trying at 10,000m but does impact on frame rates)

Object Shadow Cast Distance - 6,000m

Internal, External vehicles, Buildings - Cast and Receive

SimObjects - Cast

No Shadow Flagged Content - unchecked

 

Cloud draw distance - 80 miles

Volumetric Fog - checked

Detailed clouds

Cloud coverage density - medium

 

Airline traffic - 20%

General Aviation - 20%

Airport vehicle - low

Road vehicles - 12%

Ships and ferries - 50%

Leisure boats - 20%

 

Performance is very good and smooth without blurriness and barely percaptible, intermittent stuttering only noted during tight turns in the F22-Raptor, whereas in slower a/c like the A2A C172 not really noticed at all. When I first ran with this configuration I was frankly astonished at the quality of the visuals - everything was sharp as a tack and ground texture updating was/is extremely fast. Also, with full SLI optimisations on the horizon things can only get better

 

TrackIR works fine too.

 

So, if my system can achieve this level of performance yours should certainly be able to match it and better with much higher frame rates.

 

Keep at it, I'm sure you'll get there.

 

Cheers!

 

Mike

Beau posted on the LM forums that 2.3 will likely run on all cores:

http://www.prepar3d.com/forum-5/?mingleforumaction=viewtopic&t=6380.4

 

We are seeing similar results and will be setting the default AM to use all cores for 2.3. We're still trying to hunt down why the paging can seem to get log-jammed at times with fewer cores, but it does seem like throwing more job threads in the mix speeds things back up on our test machines.

 

Beau Hollis

Rendering System Lead - Prepar3D® Team

So no obvious hardware issues ... good news.  Sounds like you'll just have to experiment with graphics settings and/or AffinityMask values.  

 

If you run other products in the background like TrackIR, GoFlight interface tools, ASN, EFB, video recording tools (i.e. FRAPS) this will interact with your AffinityMask setting.   What I mean by this, I run 4092 ... good for P3D improving hi-res texture loading BUT it can make TrackIR a little jittery and when I start recording with FRAPS I can see the texture system "slow down" ... as soon as I stop recording texture loading is back to normal.

 

So I can play around with AffinityMask settings to trade off some texture loading performance but gain some performance when recording with FRAPS.  So when setting AffinityMask you might also want to consider what else might be running that needs the a CPU core or two.

 

Cheers, Rob.

 

Thanks Rob, I have now got the resolution and blurries sorted! It seems I was running MSAA and Anistropic at too high a level, also not having NI set up correctly, throw into this a locked frame rate that wasn't doing me any favours and hey... I guess old FSX habits die hard :wink:

 

 

Hi Howard,

 

Couple of other suggestions which I have found to be helpful after sessions of fiddling around and consequent negative impact on frame rates:

 

Run: %LOCALAPPDATA%\Lockheed Martin\Prepar3D v2

And delete the 'Shaders' folder.

 

Run: %APPDATA%\Lockheed Martin\Prepar3D v2

And delete 'Prepar3D.CFG'

 

Run the sim and give it time to rebuild 'Prepar3D.CFG' and recompile the 'Shaders' folder.

 

[JOBSCHEDULER]   - added at head of Prepar3D.CFG

AffinityMask=14

 

NO OTHER CHANGES MADE TO THIS FILE

 

My System Specs: Alienware M18x (R1) with i7 2960XM, GeForce GTX 580M 2x2GB DDR5 in SLI, 20GB DDR3-1600, Monitor: 18.4" 1920x1080x32

Windows 7 Professional 64bit SP1

Component Details Subscore Base score

Processor Intel® Core i7-2960XM CPU @ 2.70GHz (Turbos to 3.3-3.7GHz)

Determined by lowest subscore

Calculations per second................................7.6

Memory (RAM) 20.0 GB............................... 7.7

Graphics NVIDIA GeForce GTX 580M......... 7.9

Gaming graphics 4096 MB Total available graphics memory.... 7.9

Primary hard disk 172GB Free (452GB Total)........................... 5.9

 

Affinity Mask - still experimenting, but 14 appears to be best for me (HT enabled, but only 3 of the 4 physical cores are mainly used with this value in Prepar3D).

 

Prepar3D v2.2.10438.0 (Settings):

 

FXAA - OFF

MSAA - 4 samples

Anisotropic - 4x (try it, you may be pleasantly surprised - I know I was)

Texture Resolution - 2048x2048

VSync - ON

Triple Buffering - Checked

Target frame rate - Unlimited

 

Level of detail radius - High

Tesselation Factor - Ultra

Mesh resolution - 5m

Texture resolution - 15cm

 

Scenery complexity - Dense

Autogen vegetation density - Normal

Autogen building density - Dense

Water Detail - Medium

Reflections - User Vehicle

Special effects detail and distance - High

 

Landing lights and Lens flare - checked

Shadow Map Count - Medium

Enable Terrain to Receive Shadows - checked

Terrain and Shadow Cast Distance - 0m

Cloud Shadow Cast Distance - 0m (worth trying at 10,000m but does impact on frame rates)

Object Shadow Cast Distance - 6,000m

Internal, External vehicles, Buildings - Cast and Receive

SimObjects - Cast

No Shadow Flagged Content - unchecked

 

Cloud draw distance - 80 miles

Volumetric Fog - checked

Detailed clouds

Cloud coverage density - medium

 

Airline traffic - 20%

General Aviation - 20%

Airport vehicle - low

Road vehicles - 12%

Ships and ferries - 50%

Leisure boats - 20%

 

Performance is very good and smooth without blurriness and barely percaptible, intermittent stuttering only noted during tight turns in the F22-Raptor, whereas in slower a/c like the A2A C172 not really noticed at all. When I first ran with this configuration I was frankly astonished at the quality of the visuals - everything was sharp as a tack and ground texture updating was/is extremely fast. Also, with full SLI optimisations on the horizon things can only get better

 

TrackIR works fine too.

 

So, if my system can achieve this level of performance yours should certainly be able to match it and better with much higher frame rates.

 

Keep at it, I'm sure you'll get there.

 

Cheers!

 

Mike

 

Thanks Mike, your suggestions have been very helpful. Now I just need to try and sort out the Vsync. Cheers

Hi Howard,

 

Glad you're making progress. However, a substantial part of the credit should go to Rob Ainscough as many of my in-sim settings were taken from his helpful web page:

 

http://www.robainscough.com/Prepar3D_Settings_2.html

 

In particular, reducing AF from my usual time honoured setting of 16x to 4x was a revelation. It appears that pushing anisotropic filtering can have negative consequences visually and, rather than assuming that 16x is always best, one should be encouraged to experiment.

 

I have not used Nvidia Inspector, preferring instead to 'let the application decide'. The settings in Nvidia Control Panel have been checked to ensure that the relevant default settings are set as such.

 

Also, rather than jumping into the sim and starting out immediately, I have come to accept that it's good practice to leave sufficient time for the sim to load fully and settle down. I would be the first to acknowledge that the suggested settings may not suit everyone, but I suspect they may represent a good solid basis on which to start in most mid-range systems. Even those employing cutting edge hardware would be well advised to begin with less ambitious settings and build progressively from there.

 

Having found my 'sweet spot' it's difficult not to feel some concern when I read that the next update (2.3) may default to using all available cores. As can be seen from my Affinity Mask setting of 14, P3D is using physical cores 1,2 and 3 and core zero is 'reserved'. Hyperthreading is ON and I have no way of switching it off so the logical cores presumably are idle. I've tried all the various posted settings which force the use of the logical cores but always come back to the setting of 14 which seems to work best on my system. If LM find a way of optimising the use of all available cores then fine; If not, then I would hope that we will still be able override any default setting to fine tune the sim's performance.

 

My 2x GeForce GTX 580Ms are using the latest 337.50 BETA drivers.

 

Lastly, perhaps I should have mentioned that my Prepar3D v2 installation is not vanilla. The following are also installed: ORBX FTX Global Base, ORBX FTX Global Vector, ORBX FTX EU Scotland, FS Global 2010 FTX Compatible, AccuFeel and EZdok EZCA.

 

The imagery is often spectacular, performance is very good and both owe much to those aforementioned listed settings.

 

Mike

 

 

In particular, reducing AF from my usual time honoured setting of 16x to 4x was a revelation. It appears that pushing anisotropic filtering can have negative consequences visually and, rather than assuming that 16x is always best, one should be encouraged to experiment.

 

I have to respectfully disagree with this. There are no negative consequences using 16x in the sims settings, unless you dislike your distant terrain being sharp.. at 4x there is noticeable loss of quality, for little or no performance gain. My issue is getting 16x to stick, as it reverts to 4x every time I start the sim, despite what it says in the settings.

 

The only time you get odd things happen with filtering is forcing through NI or the GPU control panel, which makes the water look blurry.

 

 

My issue is getting 16x to stick, as it reverts to 4x every time I start the sim, despite what it says in the settings.

 

Confused, so how are you getting 16X then?  

 

Cheers, Rob.