scottp

Thanks so much bojote. Here are my results as measured using FPSMark. I isolated each tweak on my system that you had suggested using your autotweaker tool to see its affect Overall I picked up 7.7 FPS! My baseline FPSMark score was 15.6After all of the fixes below my FPSMark score is now 23.3 AffinityMask not present. Adding for optimal results with your hardware.[JOBSCHEDULER]AffinityMask=8416.5 (+0.9) You don't have a BufferPools section!, adding optimized settings.[bufferPools]RejectThreshold=13107215.9 (+0.3) ALLOW_SHADER_30=1 has been added to make FSX Shader 3 compatible[GRAPHICS]ALLOW_SHADER_30=115.6 - no change Adjusting TextureMaxLoad to 12 based on your hardware.[Display]TextureMaxLoad=1215.6 - no change SWAP_WAIT_TIMEOUT has been added to your configuration to prevent blurries.[TERRAIN]SWAP_WAIT_TIMEOUT=215.6 - no change SmallPartRejectRadius have been added to your config for performance[sCENERY]SmallPartRejectRadius=4.016.7 (+1.1) Adding MAX_ASYNC_BATCHING_JOBS for improved autogen batching performance.[sCENERY]MAX_ASYNC_BATCHING_JOBS=315.6 - no change Adjusting SOUND_LOD for improved performance when using AI sounds.[sOUND]SOUND_LOD=115.7 (+0.1) We suggest to try setting FPS to unlimted inside FSX, and use an EXTERNAL FPS limiter. You'll experience a massive performance increase. Download the limiter here20.7 (+5.1)

Thanks for the LOD_RADIUS tip. Here are my FSMark numbers:4.5 - 22.11 FPS (default)6.0 - 21.437.5 - 19.599.0 - 19.3412.0 - 18.98For me I took a FPS hit going from 4.5-7.5 so I think I will keep it at 6.0. It did look better though as objects and textures are drawn out further. Then from 7.5 to 12.0, I didn't see a big frame rate drop but the load time of the flight was much, much longer. I am sure my memory usage was way up as well. It looks fantastic though to see so much autogen.Scott

In case anyone is interested, here is what is done at each water setting from Phil Taylor (I apologize if this is old info):1.x has no reflections: The lowest 1.x level does the basic shader. The mid 1.x level adds an animated detail texture. The highest 1.x level adds some specular effects2.x adds a 2nd pass to get reflections and this can be a killer: 2.low reflects only the clouds and has a little more complicated shader

Thanks for the feedback. I took your advice on the water setting. Here are my numbers using FSMark:Max 2x - 14.41 FPSHigh 2x - 15.97 FPSMid 2x - 16.80 FPSLow 2x - 21.52 FPSThat is quite a huge savings to go down to Low 2x. I think I will keep that. Regarding my traffic settings, they are set to Ultra High and I use MyTraffic. I did another quick experiment and turned all traffic off with Low 2x. My number using FSMark is:Low 2x with no traffic (instead of Ultra High) - 26.91 FPSI like traffic though so I will keep at Ultra High.Thanks,Scott

Given the suggestions, I ran tests with the TERRAIN tweak. Here are the results in FPS (1st # is TREES, 2nd # is BUILDINGS):Sparse - no tweak - 18.89Dense - no tweak - 18.05Dense - 1500/1000 - 17.83Dense - 4500/3000 - 17.59Very Dense - no tweak - 17.39Very Dense - 450/300 - 17.56Very Dense - 1500/1000 - 17.00Extremely Dense - no tweak - 15.38Extremely Dense - 450/300 - 17.65Extremely Dense - 1500/1000 - 17.18I agree that 1500/1000 seems a good spot if you want to run at Extremely Dense. Notice that I lost 0.83 FPS going from Dense to Very Dense with this tweak. But going from Very Dense to Extremely Dense, I actually gained back 0.18 FPS. So definitely at the Extremely Dense level we are hitting a ceiling regarding maximum objects placed (remember I am flying South through Seattle).Interestingly I lost 0.39 FPS when adding this tweak at the Very Dense level but I gained 0.17 FPS when I went down to 450/300 at this level. This makes sense because less objects are placed.So given this info, I will use Extremely Dense at 1500/1000.Thanks for the feedback,Scott

Thanks for the feedback. I did not try the autogen tweak. From what I understand, this just places fewer autogen objects. Thus I believe the same result can be achieved by just lowering the slider. The only difference is that with the tweak, you can control the ratio of trees to buildings.Scott

According to NickN from this link regarding why FPS Limiter works http://www.simforums.com/forums/forum_posts.asp?TID=29685This may be one of the all-time best tweaks to use however at this point testing is limited to a small number of usersThis fix is working perfectly here. What I can see at this point is people on lesser hardware, or, those who are not tuning FSX correctly, or, those who have very high overhead such as i7 clocked on 285 cards, may not get any advantage with this because a. A slow system lacks the overhead higher end systems have and with the higher resources better systems run they are not being properly governed by the application where the slower system never displayed the issue to begin with.b. A very high end i7 clocked may or may not see advantage with this. I am seeing some positive and some neutral results. Because i7 clocked places a huge overhead on the system, moreso that DDR3 QX, using the frame lock in FSX may accomplish the same goal as this app. I am seeing scenery areas where the app does help with i7 and others where there is no difference.c. A system which is not set up right for settings as outlined will be so out of balance with the hardware this will produce no result, or, a worse result.This fix changes definitely the priority to autogen and terrain and appears it may throttle some threads through the Windows service csrss.exe. Those on faster systems that have the overhead and are getting thread collisions.. this fix eliminates those. Those on slower system that do not have the overhead to begin with and their stutters are related to perf based on hardware limits, this will probably have none or little effect.Regarding the other questions: - I run on Win7 64 bit. I did not compare with any other system - I have OS and FSX on separate drives to always keep FSX defraggedScott

I should have titled the post differently because actually I was concerned with three things: - smoothness of flight - high visual quality - maximizing FPSI believe all should be considered when tweaking settings/installing add ons. Obviously if you are getting 5 FPS, no matter what you won't have a smooth flight. I found the settings that acheived the best result for me. In some cases I did get a higher FPS but threw the result out because the visual quality or smoothness decreased. In other cases, I got a lower FPS but accepted the result.Regarding FPS Limiter, this is a way to make your flying smoother (and increase frame rates for me at least). It reduces the FPS fluctuation. You set unlimited in FSX but tell FPS Limiter to limit to 30 FPS. Here is the link to NickN's excellent installation guide (very extensive):http://www.simforums.com/forums/forum_posts.asp?TID=29041Here is the link to the discussion on FPS Limiter (there are lots of posts actually):http://forums1.avsim.net/index.php?showtop...&hl=limiterHere is a link to download FPS Limiter:http://www.filecrop.com/24317415/index.htmlHere is how I use FPS Limiter in a bat file:start D:\Data\FSX\FPSLimiter\FPS_Limiter.exe /r:D3D9 /f:30 /x:OFF /l:OFF D:\FSX\fsx.exeScott

I just posted my findings on tweaking a few things and then running a recorded flight over Seattle to measure FPS. I did test BufferPools. I found the following:1. Set to 1000000 = 17.40 FPS2. Set to 4000000 = 17.59 FPS3. Set to 6000000 = 18.76 FPS4. Set to 10000000 = 18.38 FPSSo there appears to be a spot where you get best performace and is probably different for each user. For me I am sticking with 6M.Scott

I did a complete reinstall of my FSX system but this time I took a more systematic approach. For each change, I played back the exact same flight meausured for exactly 300 second by FRAPs to get the average FPS. I verified that I got back the exact same avg FPS each time I started FSX and played back the flight given that nothing changed.Main system components:Core 2 Duo E8400 @ 3.83 GHz4 MB of DDR2 @ 800 MHzHD 5870 Video CardWin 7 64 bitI downloaded FSMark from flightsim by Gary Dunne. He has a 5 minute flight over Seattle that hits water, city, airport and clouds flying a CRJ700 @ 250 KIAS at 1500 feet. He has instructions how to set up FRAPS to measure the flight. Each test is cumulative so once I keep an addition/tweak, it carries over to the next test as well. This way I only ever change one thing at a time.Test 1Just FSX + SP2 installed w/default settings. Result = 16.26 FPSTest 2Kill all unneeded services and processes. I was able to get Windows 7 down to 680 MB. Result = 16.42 FPS. The 0.16 FPS gain for me was not worth the trouble.Test 3Use FPS Limiter w/target frame rate = unlimited. Result = 23.54 FPS. Huge 7.1 FPS increase. This utility is endorsed by NickN as well. If you don't have it get it! Test 4Set Mesh Complexity from 75 to 100, and Mesh Resolution from 38 to 10. Result = 21.85 FPS. This 1.7 FPS drop is too big a hit for me so pulling back sliders.Test 5Set Mesh Complexity back to 75 but keep Mesh Resolution at 10. Result = 22.91 FPS. This is better. So I am keeping MC=75 and MR=10.Test 6Set AutoGen from Dense to Very Dense. Result = 19.53 FPS. Drop of 3.4 FPS. Too big a hit so not keeping it. Keep default setting of Dense.Test 7Install FSGenesis mesh. Result = 21.77 FPS. Drop of 1.2 FPS but well worth it.Test 8Install UTX USA. Result = 19.12 FPS. Drop of 2.6 FPS. Keeping product but need to make back up FPS somewhere.Test 9Reduce Water from High 2x to Mid 2x. Result = 22.90 FPS. Gain of 3.8 FPS so switch to Mid 2x. Got my frames back!Test 10I am really annoyed by the shimmering water. For NVidia you can use clamp and MipBias to reduce the effect. For ATI, you use Super Sampling (SS) AA mode. Works like a charm. Result = 19.60 FPS. Drop of 3.3 FPS. Big hit but worth it though to get rid of shimmering.Test 11Install Flight Environment with Super High Definition 4096kb cloud textures. Result = 18.14 FPS. That is a 1.5 FPS hit. Too rich for me.Test 12Install Flight Environment with 2048kb cloud textures. Result = 18.76 FPS. Drop of 0.8 FPS. Worth it for the great clouds. Tried 1024kb and got about the same FPS so sticking with 2048kb.Test 13Bumped up regular AA, AF and MipMap detail. Result = 12.27 FPS. Way too big a hit. Not keeping these changes.Test 14Set TEXTURE_BANDWIDTH_MULT from 200 to 40. Result = 18.92 FPS. Gain of 0.2 FPS. Sticking with 200 because only slight gain and I think textures fill in better at 200.Test 15Set TEXTURE_BANDWIDTH_MULT from 200 to 400. Result = 17.26 FPS. Drop of 1.5 FPS. Sticking with 200.Test 16Set PoolSize from 6000000 (6M) to 1M. Result = 17.40 FPS. Drop of 1.4 FPS. Keeping it at 6M.Test 17Set Pool Size from 6M to 4M. Result = 17.59 FPS. Drop of 1.2 FPS. Keeping it at 6M.Test 18Set Pool Size from 6M to 10M. Result = 18.38 FPS. Drop of 0.4 FPS. Keeping it at 6M. Seems to be the sweet spot for me.Test 19Set FIBER_FRAME_TIME_FRACTION from 0.33 to 0.10. Result = 19.70 FPS. Nice 0.9 FPS pickup but textures not given enough time to be rendered. Looks bad to me. Keeping it at 0.33.Test 20Set FIBER_FRAME_TIME_FRACTION from 0.33 to 0.50. Result = 17.50. Drop of 1.2 FPS. Keeping it at 0.33.So my final changes that I changed from the default settings are:Use FPS LimiterMesh Complexity = 75Mesh Resolution = 10AutoGen = DenseUse FSGenesis, UTX USA, and FEX w/ 2048kb cloud texturesAA=4X, AA Mode = Super-sample, AF=8x, Mipmap=QualityWater = Mid 2xTEXTURE_BANDWIDTH_MULT = 200PoolSize = 6000000FIBER_FRAME_TIME_FRACTION = 0.33Hopefully this information will be useful to others as well. At least now I don't need to tweak anymore and I can just fly (unless I want to test the effect of the traffic sliders).Thanks,Scott