Today I ran 12 minute dynamic FSX benchmark I have made on FSX RTM then SP1 straight after. I used FRAPS to record an FPS measurement each second. I used exactly the same setting for both tests, which are the default global high settings detected for my system, except I turned AA and AF on, set the FPS limiter to unlimited and ran windowed on a single screen @ 1680 x 1050. System specs are as below.The benchmark is a 12 minute long CRJ700 flight that starts just as the aircraft has taken off from KPAE, climbs to 1500ft, 250kts towards the Seattle city center, overflies KSEA then heads directly for Mount Rainier. The idea is to get a good mix of rural / suburban / cityscape, clouds, water, and AI aircraft / boats / cars. To give you a better idea, here is a screen shot of the flight plan:http://forums.avsim.net/user_files/172397.jpgThe results are quite impressive for SP1, as you can see by the attached graph of the results:http://forums.avsim.net/user_files/172398.jpgOf note is the following:- at around the 1:30 mark the city buildings of Seattle start appearing and the FPS drops off accordingly- between 4:30 and 5:00 the aircraft is directly overflying the city and banking such that the buildings are mostly in full view- the dip in FPS at 7:30 is just as the aircraft overflies KSEA and gets an eyeful of all the ground activity as it banks left towards Mt Rainier.- RTM closes in on SP1 as the BUDAT intersection is passed at around 10:30, then it is pretty much even from there on.- I actually cut the benchmark short at 12 mins because about a minute later the CRJ700 becomes a fireball on the side of the mountain as 1500ft ground elevation is crossed :-vuur Summary results of the two runs are: Min Max AvgRTM 4 31 14.746SP1 9 35 23.604% Gain 125% 13% 60%In terms of smoothness SP1 had the edge on RTM. Blurries were not an issue with either run. The 60% improvement in average FPS over such a complex and varied run is an outstanding achievement ACES! Hope you guys found this interesting. :) Gary

Nicely done....I agree that I too saw an increase in FPS, but with post SP1, I have more stutters and blurries...

Very interesting! Could you post instructions on how to create a benchmark for FSX?

That's the best analysis I have seen yet. Nicely done. I am experiencing much the same with SP1. Nice to see it qualified there.

Gary, great job!! This is a great analysis.. very interesting.Thank you.

For a dynamic benchmark such as this, I first created a flight plan of what I thought would be a good path to capture all the different FSX features I wanted to exercise. Then I created a flight with the CRJ700 taking off rwy 16R at KPAE with the subject flight plan loaded up. I configured the autopilot for GPS nav, 1500 ft alt and 250 kts speed and set the view to how I wanted it. Just after I took off, I raised the gear and flaps, engaged the autopilot then paused the sim and saved it. I also save my settings.I ran the scenario a few times and made some tweaks to the flight plan to have the flight bank exactly when I wanted it to, then saved the final flight plan.I set up FRAPS (freeware version) to use F1 as the benchmark hotkey, to stop benchmarking after 720 seconds (12 mins) and to save minmaxavg and FPS detailed benchmark statistics.To run the benchmark, I simply run FRAPS first, start FSX, load the settings, and load the benchmark flight (which also loads the flight plan automatically). Then in quick sequence I hit P to unpause the sim and F1 to start FRAPS measure. The FRAPS FPS counter disapears from whatever corner it was showing in, indicating that it is now collecting data. Sit back and watch the flight for the next 12 mins. When the FRAPS FPS appears again, the benchmark is finished (although the flight will continue to its fiery death if you let it). Then you go to your FRAPS benchmark directory, open the relevant CSV files it has created and get your result.The main thing you are after is repeatability, so I do a couple of runs of it to make sure it is giving the same result. I also check it out on my sons computer to make sure that it translates the same across to other computers.Gary