Hi,Apologies if this has been covered already, but while casually browsing (stutter hunting) the newly created FSX.cfg post SP1 I happened to notice a couple of variables I had not seen before: ProcSpeed=4500PerfBucket=6Anyone know what they represent?Oh, and I too have the 'stutters' in Spot view. More like brief surges every few seconds and it does rather spoil the experience. Oddly enough, I don't see this while flying over ground tiles downloaded by TileProxy. Fortunately, this is currently my main use of FSX otherwise I would be fretting rather more about this issue. FS9 still remains my main sim for the moment.MikeASRock 939Dual-SATA2, AMD Athlon 64X2 4800+ (2400MHz)(Toledo), 2GB Crucial PC3200 DDR400 Ram 3-3-3-8 (2T)(Dual Channel), (PCI-E)Sapphire ATI Radeon X1950 Pro 512MB (Catalyst 7.1 WHQL), Hyundai ImageQuest Q17 17" TFT LCD 20ms Digital Monitor (1280x1024x32), SB Audigy2 ZS Platinum (Drivers version 5.12.0001.1196 WHQL), WD Caviar SE16 250GB (WD2500KS) SATAII + Hitachi Deskstar T7K500 ATA133 UDMA-6 (320 GB), ASUS DRW-1608P2S Optical Drive, Antec P150 case with NeoHE 430W PSU, CH Products USB Yoke, Pedals and Fighterstick, Plantronics Headset and Mike combo, TrackIR Pro ver. 4.1 (build 29), Logitech MX1000 Laser Mouse, Windows XP Home Edition (SP2), DirectX 9.0c (April 2007 Update)

Hi Mike,Funny you should post this, I also just noticed these new variables and was thinking about posting here. I have no idea what they are but I can add some info.I have run FS with specialaized config files for quite some time and still do with FSX. So I have one FSX.cfg for single-engined prop planes, a different config file for turboprops, and yet another for jets. I use FSAutoStart to load FSX and select the appropriate config file.I first noticed these variable in the file used for the C172. The ProcSpeed was set to 38nn and the PerfBucket was 5. Next I looked at my Leajet config and found ProcSpeed=18nn and PerfBucket=2.Why? Do I need to change on or the other or both? I have not yet experimented with changing the values and probably won't have time until next week but if you find any info I'd sure like to know about it.In the meantime, this will act as a bump for your thread.R-

Hi Ron,Strange there has been no other responses to this query.I'm guessing that these are new parameters determined by your system's capabilities CPU-wise (single, dual, quad, grunt or whatever). We know that the pre-SP1 CPU detection routines were inaccurate when it came to dual core CPUs, so the initial slider settings following installation and initial FSX.cfg build were somewhat conservative. It may be that other variable values in FSX.cfg will now be adjusted more appropriately on referencing these new parameter settings.I wonder whether they can be altered and also whether they would have any bearing on the rather disappointing stuttering that has been introduced on many systems by SP1, mine included. Have you tried replaying a flight? On my system it's a joke and it is barely watchable now.Mike

Just an educated guess, but the PerfBucket value seems to be the smallest "quantum" of CPU horsepower or CPU time that the internal task (or thread) scheduling of FSX uses to distribute workload to the cores.It may be that PerfBucket is measured in the same units as the ProcSpeed value.

For me the PerfBucket became the 'magic' setting, set to 2. I had the best performance with none of the sliders maxed out, and besides the Max_autogen no other tweaking of the cfg file. On 2 cores with frames set to unlimited frames ranged between 10 and 50, the best setting on my system but lots of stutters too. Like your more technical explanation i too had the idea that a lower setting would balance things, processing cycle wise. And it did! I think its the smoothest ride i've ever had in FS. Even with only 10 frames at a big airport its very smooth. Regards, Gerrit

I found that lowering ProcSpeed gives me a smoother ride. Maybe ACES could elaborate on the potential tweak uses of these two parameters. Pleeease. :)==Marten:Weber==P-D920 oc3.64MHz 2GB RAM WinXP ATI X1900+1024MB 2xSATA250 2xVIEWSONIC 19"Best AI available today: www.world-of-ai.com

"It may be that other variable values in FSX.cfg will now be adjusted more appropriately on referencing these new parameter settings."Hmm! That theory appears to be way off the mark and, indeed, the CPU detection routine seems to be no better than it was with FSX-RTM.I renamed my current FSX.cfg and allowed a fresh rebuild. The values without my having done anything other than run FSX and then exiting are as follows:ProcSpeed=1885PerfBucket=2The in-sim settings noted are as shown in these shots:http://forums.avsim.net/user_files/172445.jpghttp://forums.avsim.net/user_files/172446.jpghttp://forums.avsim.net/user_files/172447.jpghttp://forums.avsim.net/user_files/172448.jpghttp://forums.avsim.net/user_files/172449.jpghttp://forums.avsim.net/user_files/172450.jpgEven allowing for the more demanding nature of this sim, these settings do not even remotely reflect the capabilities of my system - not top of the range, I grant you, but certainly no slouch.The ProcSpeed and PerfBucket values in my original post presumably changed following my more appropriate in-sim adjustments.Mike

"The ProcSpeed and PerfBucket values in my original post presumably changed following my more appropriate in-sim adjustments.Nope, this appears to be another red herring!All my testing was done at Friday Harbor and I started with the base line settings immediately following the creation of a fresh FSX.cfgThe first thing I noted was NO STUTTERING (!) at these base settings.I then proceeded to up my settings one by one and was careful to check for stuttering and performance changes each time a setting was changed.I achieved the following settings with performance (fps) maintained with little variation at 25 and noted only occasional micro stuttering in spot views, which is a massive improvement over what I was experiencing before (replay performance also showed a marked improvement):GRAPHICSTarget Frame rate: 25Full Screen Resolution: 1280x1024x32Filtering: TrilinearGlobal Texture Resolution: Very HighSCENERYLevel of Detail Radius: MediumTexture Resolution: 1m (worth noting that ground detail around the a/c became progressively less blurry as I moved from 10-->5-->2-->1m. At 1m the textures appear impressively sharp and clear.Mesh Complexity: 100Mesh Resolution: 38mScenery Complexity: NormalLand Detail Texture: box checkedWater Effects: Low 2xSpecial Effects Detail: HighAutogen: SparseWEATHERCloud Detail: DeatiledCoverage Density: HighAfter satisfying myself that performance was being maintained at this situation (Friday Harbor) I then exited the sim and checked the ProcSpeed and PerfBucket values. Surprise, surprise they remained unchanged (!?) at '1885' and '2' respectively. This I was not expecting.Anyway, for me at any rate, it seems the best way to rid myself of the excessive stuttering was to allow the sim to rebuild FSX.cfg from scratch. From that base line (which, it has to be said again, bore no relationship whatsoever to my system's capabilities) I then upped the settings one by one with careful checks being made at each and every stage.What is puzzling is the fact that I previously deleted FSX.cfg immediately after installing SP1 and the end results both in terms of the presence of major stuttering (more like brief regular surges every second or so) and the ProcSpeed and PerfBucket values were very different to what I'm seeing now.I have no doubt that there will certainly be situations when I will have to back off some of the settings to ensure decent performance. As to the stuttering, I can only guess that somehow the recreation of FSX.cfg had something to do with it. At this stage I have not reintroduced any of the pevious well known tweaks, nor have I tried any of Phil's new tweaks as recorded in his Blog.Time will tell how all this pans out, but at least the future for FSX on my HD is starting to look a lot brighter. For that I am happy and would like to thank ACES and the testing team, once again, for all their sterling work over the past few months.Mike

"What is puzzling is the fact that I previously deleted FSX.cfg immediately after installing SP1 and the end results both in terms of the presence of major stuttering (more like brief regular surges every second or so) and the ProcSpeed and PerfBucket values were very different to what I'm seeing now."I think I may have found the answer to this one.I continued testing by deleting FSX.cfg repeatedly and each and every time I was unable to reproduce the original values for ProcSpeed and PerfBucket (1885 and 6 respectively):ProcSpeed: 4425, 4566, 4630, 4627, 4503, 4628, 4505, 4626, 4500 (Note the variations)PerfBucket: 6, 6, 6, 6, 6, 6, 6, 6, 6Then I had an idea! I fired up the AMD Dashboard monitoring utility and watched what was happening as FSX loaded. What immediately became clear was the periodic fluctuations of the CPU Speed Meter needle from Max to Min as FSX loaded, with corresponding parallel matching swings of the Power Savings and CPU Voltage meters. This appears to be the clue as to what has been happening and why I could not reproduce the original values for ProcSpeed and PerfBucket. After an initial load to the 'Free Flight' screen (my preference), I had continued the testing cycle by exiting FSX, deleting FSX.cfg (after noting the values) and then restarting FSX.This, therefore, was not strictly a fresh restart of FSX as we all know how much quicker the loading phase is subsequent to that first run of FSX.I thought back and realised that I had installed SP1 before having run FSX for the first time that day and so the initial loading phase lasted much longer.Not knowing when the ProcSpeed and PerfBucket values are set during the loading phase of the sim, currently I am speculating that the delay during that first cold run allowed misdetection of the CPU speed to occur. This misdetection is reflected by the low settings as seen in the images in my original post.I have now confirmed that I can recreate values close to the originals by simply rebooting, deleting FSX.cfg and then starting the sim:ProcSpeed=1882PerfBucket=2Thereafter, the deletion of FSX.cfg and subsequent reloading of FSX produces values and settings which more closely match your CPU's capabilities.If you simply delete the ProcSpeed and PerfBucket lines in FSX.cfg they will reappear at the next run of the sim. In my case the following values appeared:ProcSpeed=4443PerfBucket=6So, if all this is correct, the message appears to be:If you have an AMD CPU and you are using the "Cool 'n' Quiet" power saving technology, then following the installation of SP1 you shoud delete FSX.cfg immediately after the first run of the sim and then, without rebooting your PC, restart FSX to allow a fresh rebuild of FSX.cfg. This should allow the more accurate detection of your CPU speed.PHIL TAYLORTo be honest, without Phil's input on this one I really don't know how important these new parameters/variables are to FSX other than to set up a rough initial configuration for the user. The more experienced of us can simply override these settings by winding them all back and then rebuild step-by-step. For me, what was disappointing was the discovery that the stutters had returned (very apparent in spot view and replay) if I accepted/changed the ProcSpeed and PerfBucket values to those produced by warm boots of FSX.I would be very interested to hear anyone else's views/observations concerning this interesting matter.Also, I wonder if anyone can offer an explanation as to why I was able to rid myself of the severe stuttering by rebuilding the settings bit by bit from those set by ProcSpeed and PerfBucket from a cold run of FSX, whereas values produced by warm runs of FSX reintroduce the stutters. This despite the fact that the in-sim settings end up the same in either case (!?).MikeASRock 939Dual-SATA2, AMD Athlon 64X2 4800+ (2400MHz)(Toledo), 2GB Crucial PC3200 DDR400 Ram 3-3-3-8 (2T)(Dual Channel), (PCI-E)Sapphire ATI Radeon X1950 Pro 512MB (Catalyst 7.1 WHQL), SB Audigy2 ZS Platinum (Drivers version 5.12.0001.1196 WHQL), Windows XP Home Edition (SP2), DirectX 9.0c (with 04/2007 update)

ProcSpeed & ProcBucket are multipliers. When fsx.cfg is created, it takes your processor speed, quite simple. It also runs a quick test on system overall bandwidth... how fast your machine can do a task based on what ram you have installed. The result of this small test assigns your PC into one of several 'buckets'. Each bucket represents a range functionality and capabilities that FSX will use when rendering a scene.The bucket thing was outlined in an earlier post by Phil I think, or on the FSInsider website somewhere.Basically, these two parameters offer FSX a way of determining just how much extra functionality can run each frames. An autogen setting of Ultra High will look different on machines with a different bucket class. The higher the bucket, the more FSX will try to render for any particular setting. While bucket 5 may give 3000 trees per cell, bucket 6 may give 3500, or 3000 trees with higher texturing.This is why an AMD2500+ can seemingly outperform a newer AMD4500/64 with the sliders matched between both machines. Look at the outpuut, the 4500 will definetely be doing more to present a prettier world... even at the same settings. The 4500 running at a higher bucket setting (simply because it crunches numbers faster) may run better, more extensive cloud code to display more realistic clouds. Even though the setting is the same, the clouds look different.By the way, another tip... there are three autogen files... one low, one mid, and one full. Take the low one and copy it over the other two. Makes a world of difference for me as my machine seems to be in the bucket that uses the higher autogen detail. Switching them to always use the low one seems to REALLY help my FPS.

"If you have an AMD CPU and you are using the "Cool 'n' Quiet" power saving technology, then following the installation of SP1 you should delete FSX.cfg immediately after the first run of the sim and then, without rebooting your PC, restart FSX to allow a fresh rebuild of FSX.cfg. This should allow the more accurate detection of your CPU speed."....or alternatively:1. Temporarily disable "Cool 'n' Quiet" in the BIOS Setup2. Save and Exit and reboot into Windows XP3. Rename FSX.cfg to FSX.cfg.old3. Run FSX and configure with your preferred settings4. Exit FSX and confirm the ProcSpeed and PerfBucket values have been set correctly.5. Reboot into BIOS Setup and re-enable "Cool 'n' Quiet"6. Save and Exit and reboot into Windows XPSeems to work fine and guarantees accurate detection of CPU speed every time.Note: When you disable "Cool 'n' Quiet" the AMD Dashboard will fail to load but your CPU will be running constantly at its rated speed and voltage.Mike

I'm home from a long weekend and look forward to experimenting with some of the info you all have provided. R-

This is all very interesting to hear, i have a thread going in which i am trying to find the link between Dual core CPUS, the AMD and XP dual core pathes and bluries/stutters. Glad you guys are getting somewhere, lets keep on this.Happy flying =)J.A.P>S.Has anyone ever found more then one .CFG File? Its happend to me twice now.

Hi Jason,"Has anyone ever found more then one .CFG File? Its happend to me twice now."What are the others called? I've noted a FSX.cfg.txt (I think, if memory serves) created as part of the file creation diagnostic process when the sim crashes. No duplicates other than those I have created myself.I continue to plug away and have been experimenting with the following tweaks:[JOBSCHEDULER]AffinityMask=3 //2 cores <---- no effect noted or rather no change of core useage behaviour as observed while running without this tweak. So, my conclusion is that my CPU's cores are being used appropriately.UPPER_FRAMERATE_LIMIT=20 <---- lowering this value from my previous 25 does seem to reduce the stuttering and less variance in frame rate - mostly noted in Spot View and during turns. Interestingly enough this was the initial value set by ProcSpeed and PerfBucket when I recreated FSX.cfg. Perhaps we should be heeding these initial settings rather more than we do ;)ProcSpeed=4629 <--- created after disabling "Cool 'n' Quiet" so I guess this is the optimum value for my system/CPUPerfBucket=6 <--- DittoTEXTURE_MAX_LOAD=1024 //Default=512 <---- not noticed any obvious changes yet in performance or display. ? whether I should leave it at 1024 or revert back to the default value.SmallPartRejectRadius=2 //Try 2 or 4 (Default=1, i.e. 1 pixel) <---- haven't tested enough to comment but see no obvious change in performance/frame rate. May have something to do with my chosen testing area - Friday Harbour and Autogen currently at 'Sparse'.[bUFFERPOOLS]PoolSize=5000000 //Default=1000000 <---- Can't convince myself that this has any effect on my system with (PCI-E)Sapphire ATI Radeon X1950 Pro 512MB (Catalyst 7.1 WHQL)I have not bothered with the FIBER_FRAME_TIME_FRACTION tweak as the message appears to be that it is no longer relevant on multi-core machines post SP1:From Adam Szofran -"On multi-core machines in SP1, we moved many fiber jobs off of the primary thread and onto secondary threads. Since FIBER_FRAME_TIME_FRACTION only affects scheduling of jobs on the primary thread, it will have less of an impact on the performance of Flight Sim on multi-core machines. In fact, we moved so many jobs off of the primary thread that there probably isn

regarding PerfBucket: everywhere I see values of 5 or 6. I hav 50!My Procspeed value is similar to yours (4768 - same CPU)on the other hand: I can't remember if I changed it myself due to a tweak suggestion?Marin