NOTE TO MODERATORS: I've reposted this topic correctly formatted. Can you please delete the previous postingsI've posted this here rather than in the Hardware Forum because I felt it might be of general interest.I have a Dell Dimension 2350 with 2MHz processor, 512MG memory, 60GB 7200rpm disk, and Intel Extreme Graphics built into the mother board, with a 19" Nokia 445Xav monitor running under XP Home Edition with SP2 and FS9.1.Extreme Graphics isn't extreme and I'm thinking of getting a separate video card. Before doing so, I decided to try to run some tests using System Monitor. (Start/Run then type "perfmon.msc" without the quotes.) The two values I monitored were %Processor Time and %Disk Time for the physical disk.I set up a flight using the stock Cessna 172 with UK VFR Photographic Scenery and VFK Terrain (19m grid) at 2500ft QNH and 100kt IAS. Weather was scatterd clouds, 20 mile visibility, with no precipitation. I had no AI aircraft.I'd changed fs9.cgf using the utility that came with VFR Photgraphic Scenery. I normally get 16-18 fps under these conditions with a few excursions outside this range. I ran 6 cases as shown below where:- Visible means running FS9 full screen at Alt-Tabbing to switch to System Monitor- Invisible means running FS9 in the background (Settings/General/Pause otask switch unchecked) showing System Monitor all the time- Unlimited means the display is set to unlimited fps- Limited is with fps limited to 15 fps- No NAV means Norton Anti-virus 2003 disabled.and I collected 6 samples for each case, each sample about 2 min in lengthThe average results are: %Processor Time %Disk Time1 Visible Unlimited 99.0 8.72 Invisible Unlimited 12.5 23.03 Visible Limited to 15fps 98.9 15.34 Invisible Limited to 15fps 12.4 14.25 Visible Unlimited No NAV 98.7 7.26 Invisible Unlimited No NAV 7.7 10.9I compared a number of cases while attempting to determine if the differences were statistically significant using Student's-t distribution with the following results: Processor Disk1 & 2 S S1 & 3 S P2 & 4 N N1 & 5 S N2 & 6 S Swhere- S implies statistically significant- P implies possibly statistically significant- N implies not statistically significantTentative conclusions are:- comparing cases 1 & 2, it appears that only about 12% of the processing time is needed to "fly" the simulation with about 88% needed to drive the video. % Disk Time may be constrained by the processing load in case 1- comparing cases 1 & 3, limiting fps reduces the %Processor Time and possibly increases % Disk Time - comparing cases 2 & 4, limiting fps doesn't affect %Processor Time or %Disk Time when the video isn't being driven- comparing cases 1 & 5, disabling NAV while driving the video reduces the %Processor Time but not doesn't significantly affect %Disk Time- comparing cases 2 & 6, disabling NAV while not driving the video redfuces both the %Processor Time and the %Disk Time.I suppose I should add a "health warning" because, it appears that whenever someone proposes changed settings to improve performance someone else will say it worsens or doesn't affect performance so these results might not be reproducible. Also statistically significant doesn't mean much munless the underlying physical mechanism can be explained. I'm not sure I know enough about FS9 to do that but hope other may contribute. For example, what do changes in %Disk Time actually mean for FS9?

This is a great piece of work, but I'm fairly certain I'm too stupid to understand its' import; are we all doomed?:-hmmm

Good show! Interesting how the %Disk Time varied between Visible Unlimited vs. Visible Limited.

If these are correct (well I try to be conservative), then it also is amazing to see how much a graphic driven program can take in fact so much CPU instead of offloading it to the video card, unless, your video card hardware does a poor job at hardware accelerating things and then relies on the software driver (CPU) to do its tasks.I wonder on your system, with a modern ATI/NVIDIA video card. If similar results, this would mean THERE IS room for improvment in FS if more of its code (GFX related) would take more use of the video card than the CPU.This confirms me again that FS performance is less affected by the complexity of the add-on (read: aircraft systems) and much more affected by its graphics complexity (read: FS drawing the gauges on the screen).Thanks for sharing this!