How to benchmark FS2004

[Jimmy Angel 19]

As we try to tweak FS2004 and find our own individual "sweet spots", we are getting a lot of suggestions on the forums about various settings and tricks to get some more performance out of the sim.Lots of people try these tweaks, of course, and a lot of times they report some success, but how often is this percieved improvement just wishful thinking .... and how many times will these tweaks actually end up worsening performance or reducing the visual quality for no good purpose?I've found it very helpul to set-up a benchmarking process that allows me to make a tweak, and then view the result in a purely objective fashion.Here's how to do it...* Download and install fraps from www.fraps.com. This is a very popular benchmarking utility that is a bit more powerful than FS2004's built-in fps display* Create some flights that you can run "hands off". A good flight, for example, would be on final approach onto runway 34L at KSEA (Sea-tac), with a few clouds. This will give you a good combination of airport scenery, air traffic, autogen, and the distant buildings of Seattle to the north. If you really want to make this a "worse case scenario" benchmark, you can add some heavy overcast or thunderstorms. Set your plane on autopilot heading for the runway. You can also create other flights for specific situations ... mountain flying, heavy cities, high altitude, etc. * The only requirment is that each flight can be run via autopilot without touching the controls. You should also make sure that each flight is paused, and using a standard zoom level. The eyepoint should reset when you load it, but hit the space bar just to make sure.* Run FRAPS. This will put a fps display onto your window.* Now, set up FS2004 to your default settings. These can either be the "out of the box settings", or what you are currently using.* Load your test flight. Let it sit for a second paused, just so it can load some textures, and then hit the P key to start it running. Keep your hands and feet off the controls. Wait 2 seconds (count to yourself ... one thousand one ... one thousand two), so that the sim can get past the early chug-chugs, and then hit the F11 key to engage FRAPS benchmark mode. You will see the FRAPS display turn green for a second, and disappear. FRAPS is now keeping track of your FPS, and will wait until you hit F11 again to show you the results.* Let the flight continue for a predetermined time. You can use some sort of visual cue (when a particular building passes by), an event (the wheels touch the runway), or you can just use a stopwatch and wait for a a specific amount of time to pass. You don't want the length of the flight to be too long or to short, since that will either skew or avergage your results too much. About 20 seconds to a minute is probably good.* When you're ready to stop the benchmarking, hit F11 again. FRAPS will quickly flash the average FPS on-sreen, and it will also log the results to a text file in the FRAPS directory along with the low and high FPS.* Now that you have your baseline, start tweaking. Move a slider or make a video card adjustment, and then reset the flight and see what happens. Try changing screen sizes, detail settings, AA and AF settings (via your video card, etc.)* You might want to close and re-start FS2004 every so often when you are benchmarking (I've found that the fps tends to reduce after multiple loads of a flight). You might also want to think about an occssinal reboot just to give your system an enema. * When you've made a few tweaks that you are happy with, keep track of this as your new baseline that you will benchmark against.This process can be done quickly and will leave you feeling confident about the result of any tweaks you are making to your system. You may be surprised at some of the results.----- As an example of this process in action, someone recently uploaded new textures for FS2004 that scaled the autogen textures down by about 50%. I eagerly downloaded and installed these, hoping that I might be able to run autogen above "normal". What was the result? Absolutely no change in FPS with autogen in view. Turning my autogen from "normal" to "dense" still gives me the 3fps penalty in my standard benchmark. I went back to the original textures, glad that I wasn't sacrificing quality for some unsubstantiated hope of improved performance. If you are runing a standardized benchmark you can do the same test with these textures--I expect they might work for some people (esp. anyone with 64MB or less on their video card).Caveat: There are some tweaks that will effect the performance of the sim in ways other than FPS. Texture blurring and pop-up, for example, might be independent of FPS but there are some tweaks to improve this. If nothing else, however, you can make sure that any tweaks you perform to improve things like texture blurring doesn't have an adverse impact on smoothness.