I can't comment on the DX9 side of things ... I've done NO testing on that side.

 

I'm just trying to gather data on the Bloom aspect with DX10 and DX10 Fixer.  Jeroen's results are what I would "expect" ... however in my case DX10 + Bloom has resulted in about 25-33% FPS increase.

 

I need to test more aircraft and more situations (scenery areas, Default vs. GEX vs. photo).

 

But in my case VAS usage has not changed at all with or without Bloom, fps seems to be the only benefit.  BUT, if I don't have control over Bloom "reflective" (aka gleam) on solid (non-transparent) surfaces then I'm not sure I want use Bloom (even with the fps benefits) ... bloom it's way over saturated.  Hoping Steve can chime in on why "reflective" settings seem to have no impact on non-transparent surfaces?

 

Rob

Meant only that you don't have to run FSX to either choose DX9 by unclicking the box and rebooting. You can change up via the Fix, before you run your flight, and before you fire up FSX.   I do not mean that you can switch between the two while FSX is in fact, running.

 

Ok, thanks for clearing that up. Well, congrats on the performance you get but take note that not everyone will have similar improvements!

I have a test scenario set up at UK2000 EGLL clear skies,same time, runway etc with GEX and UTX in the NGX.

 

I sit on the runway for 2min and start timing as soon as the NGX countdown has finished to let everything settle. Suffice to say, throttle and takeoff are all done at set points (I'm anal when testing lol) I then just let it fly for 2 min. End the flight switch off FSX and check the avg. FPS in the FSUIPC log file.

 

I've not tested in DX9 but was was only interested in DX10 with and without bloom. I had both sliders at 60%. In my opinion each test should be run a minimum of 10 times and then discard the smallest and largest result and average the rest. Also, shut down FSX between each test cycle.

 

Didn't have time for that so just did one of each. To be honest, nothing conclusive on my system. Well nothing to make me want to do a full test run.

 

So why you are seeing such a difference is a mystery. Well it's either, difference in how each scenery interacts with other scenery, with itself or with different planes. Or, my electrons just weren't playing ball! The test was certainly done outside of peak electricity usage times but then with switch mode power supply's this shouldn't be a factor.

 

If I were to go into this test without knowing the results, my expected result would have been that there should be no difference between the two. So it's odd, and another PC/FSX Higgs Boson

FSX and PMDG 737 NGX (Bloom off) - leave libraries installed switch to DX9 and I get better fps as opposed to DX10, what's interesting is that the big difference in performance happens looking forward in vc (pilot view), all other views including exterior DX10 has better fps.

 

 

If I were to go into this test without knowing the results, my expected result would have been that there should be no difference between the two.

 

Actually, when I first turned Bloom on, my expectation was a slight drop in FPS as a post-pipeline render needs to happen to make the Bloom work.

 

I need to do more testing ... I'm seeing less of an FPS increase using the PMDG 777 in an FSDT airport with MSE scenery vs. using RealAir Duke in default airport and GEX ... where Bloom is the variable between both test sets.

 

Data Sets for far:

 

PMDG 777 + FSDT + REX + MSE + Bloom = 25 fps internal - 43 fps external

PMDG 777 + FSDT + REX + MSE = 21 fps internal - 33 fps external

Duke B60 + FS Airport (KMSP) + GEX + REX + Bloom = 42 fps internal - 59 fps external

Duke B60 + FS Airport (KMSP) + GEX + REX = 33 fps internal - 45 fps external

 

More testing to come ... maybe ... like others, this hit and miss approach can be frustrating leading to eventual dismissal.

Actually, when I first turned Bloom on, my expectation was a slight drop in FPS as a post-pipeline render needs to happen to make the Bloom work.

Actually, your correct. I should have qualified that with 'on my system, under normal conditions I would expect it to impair performance.' The reason is that for any given situation, on my PC I can turn things like light flare and aircraft shadows on and not notice any impact in DX10. Other variables like AI have a huge impact.

 

In addition, this got me thinking. I had SweetFX enabled at the time so don't know if that played a part?

 

 

I had SweetFX enabled at the time so don't know if that played a part?

 

I don't have SweetFX, so I can't test.  I did spend about 8 hours yesterday going over my FSX.CFG and NVidia Inspector settings and none of the "traditional" settings seem to work well for me.  I discovered that my 10 fps performance boost from Bloom appears to be ONLY in NONE photo-real areas (in my case Minnesota is the only MSE state I don't have) ... I have GEX with AutoGen density = 5 (which I believe is max?).  I did install Steve's latest released DX10 Fixer version 1.0 build 19.

 

Note: this was done on my 680 4GB, not my Titan 6GB.

 

nVidia Inspector v1.9.7.2  (drivers 327.23)

 

Antialiasing compatibility = 0x00000000

Antialiasing setting = 4x4 Supersampling (D3D only)

Antialiasing transparency Supersampling = 4x Spare Grid Supersampling

Maximum pre-rendered frames = Use the 3D application settings (any other value I used here would produce drastic reduction in FPS)

 

FSX.CFG

 

[DISPLAY.Device.NVIDIA GeForce GTX 680.0]

MipBias=8

Mode=2560x1600x32

Anisotropic=1

AntiAlias=1

 

[GRAPHICS]

MultiSampleQuality=4

MultiSamplesPerPixel=16

NIGHT_THRESHOLD=4096

DAY_THRESHOLD=32768

SHADER_CACHE_PRIMED_10=1693500672

STALE_BUFFER_THRESHOLD=2147483647

HIGHMEMFIX=1

ForceWindowedVSync=1

SHADER_CACHE_PRIMED=1693500672

TEXTURE_MAX_LOAD=2048

NUM_LIGHTS=8

AIRCRAFT_SHADOWS=1

AIRCRAFT_REFLECTIONS=1

COCKPIT_HIGH_LOD=1

LANDING_LIGHTS=1

AC_SELF_SHADOW=1

EFFECTS_QUALITY=2

GROUND_SHADOWS=0

TEXTURE_QUALITY=3

IMAGE_QUALITY=0

See_Self=1

Text_Scroll=1

D3D10=1

 

[TERRAIN]

SWAP_WAIT_TIMEOUT=2

IMAGE_PIXELS_FOR_AUTOGEN_POLYGONS=2048

BATCH_AUTOGEN2=1

LOD_RADIUS=9.500000

MESH_COMPLEXITY=100

MESH_RESOLUTION=25

TEXTURE_RESOLUTION=29

AUTOGEN_DENSITY=5

DETAIL_TEXTURE=1

WATER_EFFECTS=4

 

[sCENERY]

ObjectsToBatchPerFrame=50

MAX_ASYNC_BATCHING_JOBS=4

SmallPartRejectRadius=0

LENSFLARE=1

DAWN_DUSK_SMOOTHING=1

IMAGE_COMPLEXITY=5

 

[Weather]

CLOUD_DRAW_DISTANCE=12

DETAILED_CLOUDS=1

CLOUD_COVERAGE_DENSITY=8

 

[Display]

BLOOM_EFFECTS=1

SKINNED_ANIMATIONS=1

TEXTURE_BANDWIDTH_MULT=80

UPPER_FRAMERATE_LIMIT=0

WideViewAspect=False

Hi Rob, out of curiosity I copied over some of your cfg settings (some which I hadn't used before) - instant improvement in autogen performance and loading :blink: Where did you pick these up?!

 

They were

IMAGE_PIXELS_FOR_AUTOGEN_POLYGONS=2048

ObjectstoBatchPerFrame=50

MAX_ASYNC_BATCHING_JOBS=4

 

Thanks for letting me plagiarize! :lol: 

I am so going to give this a try tonight when I am done with my session of Shadowrun with the boyz!   If this is true I will explode with delight.   I already finally managed to got flawless smooth operation on the ground with UT2 at Payware airports by dialing back AF to 8x from 16x in Nvidia Inspector.    16x was killing my performance.  8x, smooth as glass with very little difference in visual quality.

 

This bloom thing will be the icing on the cake discovery!

 

 

Hi Rob, out of curiosity I copied over some of your cfg settings (some which I hadn't used before) - instant improvement in autogen performance and loading  Where did you pick these up?!

 

Probably the same place most of us do, trial and error, reading, re-reading, reading some more, searching, file monitor, etc..  Just be aware that it's often a combination of entries along with graphics cards settings, inspector settings (and more) -- possible number of unique combinations is very large and probably why so many have very different results.

 

 

 

This bloom thing will be the icing on the cake discovery!

 

May the force be with you.

Hi Rob, out of curiosity I copied over some of your cfg settings (some which I hadn't used before) - instant improvement in autogen performance and loading :blink: Where did you pick these up?!

 

They were

IMAGE_PIXELS_FOR_AUTOGEN_POLYGONS=2048

ObjectstoBatchPerFrame=50

MAX_ASYNC_BATCHING_JOBS=4

 

Thanks for letting me plagiarize! :lol:

 

Under which section in the fsx.cfg do those get listed ?

 

Thanks ... Hari

See above Hari:

 

[TERRAIN]

IMAGE_PIXELS_FOR_AUTOGEN_POLYGONS=2048

 

[sCENERY]

ObjectsToBatchPerFrame=50

MAX_ASYNC_BATCHING_JOBS=4

@Robains

Some of the fsx.cfg entries a outdated and useless for FSX Acc package, like MipBias=8. I guess you forgot to clean up those entries.

Spirit

 

 

 outdated and useless for FSX Acc package, like MipBias=8. I guess you forgot to clean up those entries.

 

I've removed MipBias from my FSX.CFG but as with all things FSX, I've read complete opposite opinions/info regarding MipBias usage for FSX SP2/Acc.

 

But I agree, MipBias didn't seem to have any impact on my system ... see my other thread (LOD_RADIUS), I posted some images of no MipBias, MipBias=6, and MipBias=10 ... I can't see any difference, but my eyes are old.

 

I don't have SweetFX, so I can't test.  I did spend about 8 hours yesterday going over my FSX.CFG and NVidia Inspector settings and none of the "traditional" settings seem to work well for me.  I discovered that my 10 fps performance boost from Bloom appears to be ONLY in NONE photo-real areas (in my case Minnesota is the only MSE state I don't have) ... I have GEX with AutoGen density = 5 (which I believe is max?).  I did install Steve's latest released DX10 Fixer version 1.0 build 19.

 

Note: this was done on my 680 4GB, not my Titan 6GB.

 

nVidia Inspector v1.9.7.2  (drivers 327.23)

 

Antialiasing compatibility = 0x00000000

Antialiasing setting = 4x4 Supersampling (D3D only)

Antialiasing transparency Supersampling = 4x Spare Grid Supersampling

Maximum pre-rendered frames = Use the 3D application settings (any other value I used here would produce drastic reduction in FPS)

 

FSX.CFG

 

[DISPLAY.Device.NVIDIA GeForce GTX 680.0]

MipBias=8

Mode=2560x1600x32

Anisotropic=1

AntiAlias=1

 

[GRAPHICS]

MultiSampleQuality=4

MultiSamplesPerPixel=16

NIGHT_THRESHOLD=4096

DAY_THRESHOLD=32768

SHADER_CACHE_PRIMED_10=1693500672

STALE_BUFFER_THRESHOLD=2147483647

HIGHMEMFIX=1

ForceWindowedVSync=1

SHADER_CACHE_PRIMED=1693500672

TEXTURE_MAX_LOAD=2048

NUM_LIGHTS=8

AIRCRAFT_SHADOWS=1

AIRCRAFT_REFLECTIONS=1

COCKPIT_HIGH_LOD=1

LANDING_LIGHTS=1

AC_SELF_SHADOW=1

EFFECTS_QUALITY=2

GROUND_SHADOWS=0

TEXTURE_QUALITY=3

IMAGE_QUALITY=0

See_Self=1

Text_Scroll=1

D3D10=1

 

[TERRAIN]

SWAP_WAIT_TIMEOUT=2

IMAGE_PIXELS_FOR_AUTOGEN_POLYGONS=2048

BATCH_AUTOGEN2=1

LOD_RADIUS=9.500000

MESH_COMPLEXITY=100

MESH_RESOLUTION=25

TEXTURE_RESOLUTION=29

AUTOGEN_DENSITY=5

DETAIL_TEXTURE=1

WATER_EFFECTS=4

 

[sCENERY]

ObjectsToBatchPerFrame=50

MAX_ASYNC_BATCHING_JOBS=4

SmallPartRejectRadius=0

LENSFLARE=1

DAWN_DUSK_SMOOTHING=1

IMAGE_COMPLEXITY=5

 

[Weather]

CLOUD_DRAW_DISTANCE=12

DETAILED_CLOUDS=1

CLOUD_COVERAGE_DENSITY=8

 

[Display]

BLOOM_EFFECTS=1

SKINNED_ANIMATIONS=1

TEXTURE_BANDWIDTH_MULT=80

UPPER_FRAMERATE_LIMIT=0

WideViewAspect=False

 

 

 

Hi, Rob

 

I have been tweaking alot these day over in the ATI camp :-) so just for fun i tried out your settings with my HD7970 Vapor-X and took my L-39 over FTX Australia with YBCS Canberra and Cityscape by ORBX and it was by surprice pretty good. 20-30fps but must of the time in the upper 20´s lower 30´s and pretty smooth. Might go back to a little lower settings but nice with some VFR flying and crispy textures

 

Michael Moe