So I’m finally done. It’s been a tough benchmark to finish. Just as I started the test I managed to drown my brand new computer in Ice Tea and had to build a new one. Basically I started fresh with a new computer and fresh installs of my hardware, software and FSX Acceleration and P3D v 1.2. I then programmed an autopiloted flight with the default King Air 350 crisscrossing New York from KJFK to KEWR. I did one set of flight with every slider maxed out, and one set with “high settings” which equals my default settings when I go flying. Anyway, here are the results:
As you can see there is no significant difference in performance between FSX and P3D on my computer. In addition I found several bugs in P3D and had two crash to desktops (CTD) during testing. I’m just as surprised as many of you are, especially since many P3D-users have reported an increase in fps and stability when using P3D. If you want to see how the test was done please read on.
COMPUTER AND SETUP
Because of my little clumsy and expensive accident I had the luxury of starting with a brand new computer and fresh installs of Windows and hardware drivers:
I ran various benchmark and test programs to ensure the outmost stability of my PC before installing FSX and P3D. You can see my PC specs in the right column of this webpage. My PC is running stock – no overclocking or BIOS tweaks other than those needed to make my hardware function properly. Windows, drivers and software are all up to date – no beta drivers are installed. I have no anti-virus software running in the background either. Here are the benchmarks of some of the tests I did before running FSX or P3D:
FSX AND PREPAR3D SETUP
After installing FSX Acceleration and P3D in the proper manner I maxed out all settings to put as much strain on the software and hardware as possible. My goal with these tests is to find out both how stable and fast the programs can run without applying custom tweaks or installing addons. I meticulously compared the configuration files of both programs to ensure equal settings. In P3D I have disabled bathimetry, refraction and clarity as these aren’t present in FSX. I did two small adjustments to the original configuration files though: Enabeling WideWiewAspect and AverageFrameRate. I run both sims in full screen mode at 1920x1200x32 resolution. NVidia control panel is set to “Let 3D application decide”.
High settings is basically my day to day setup of FSX and Prepar 3D. High water settings and bloom are the main performance killers. Since they look totally unrealistic it is no loss reducing water settings and disabeling bloom. Starting with maximum settings I’ve done the following edits:
- Disabled bloom and lens flare (Performance killers that looks artificial – so no real loss).
- Disabled cockpit tooltips and aircraft self shadowing (Reported as being buggy).
- Reduced water settings to “Low 2x” (No loss either, as more reflections look just silly. In my 20 years of landscape and aerial photography I’ve never seen reflections in the ocean like those generated by the FSX water settings).
- Autogen reduced to “very dense” and shadows disabled. I havelater increased autogen settings to extremely dense as they didn’t impact performance on my computer.
- Reduced cloud range to 90 mi (After benchmarking I’ve returned these to the maximum settings. Clouds don’t bother my computer much).
- Reduced Airline and GA traffic to 59% and airport traffic to “medium” (The default airline traffic is too sparse at larger airports even at 100%, so I will install My Traffic X at a later stage. I’ve increased airport traffic to maximum after benchmarking).
- Reduced road traffic to 30% and boats to 15%. Ships are kept at 100% (This makes traffic look realistic in my opinion. In large cities I sometimes increase road traffic to 50%, and when flying over the backcountry I reduce it to 10-15%).
As you can see from the results, an i7 2700K powered computer runs FSX and Prepar3D pretty good at “high settings”. However some annoying bugs will always be there like the “popping” autogen and clouds, the occasional stutters and black/blue texture tiles.
THE BENCHMARK FLIGHT
To make a fair comparison I programmed an autopiloted flight crisscrossing New York City from KJFK to KEWR. At each waypoint along the route I took an average fps reading. Each flight was done three times with reboots of windows in between. The results for each flight thus represents an average of three flights. The variations were small however, up to one fps at most.
The flight is done with the default King Air 350 in stormy weather at 22. September 2011 at 12:00. Realism settings are set to maximum and are exactly the same for both sims. I fly from the virtual cockpit at maximum wide angle zoom setting with the 2D GPS window open. I let the aircraft rest for 2 minutes before taking off to ensure that all the traffic, weather and scenery is properly loaded. Here’s the route to be flown at a cruise altitude of 1500 feet:
KJFK RW4R > MAHAL > OMEME > PONEY > NORAY > 6N7 > MOSME > ORFAN > CANBO > KEWR
The King Air is configured with all lights on, full flaps, throttle 100%, props 100%, gear down – keeping the girl slow so she can manage the tight turns). After takeoff I climb to 1000 feet and turn on the autopilot. Then I just sit back and take notes of the fps at the waypoints. When reaching KEWR I let the aircraft circle the airport twice on autopilot. During the last circle I pause the sim to check for blurries of the ground textures. The flight takes 15 minutes (I tried longer flights but didn’t see a decrease in performance over a 50 minutes flight circling New York, so 15 minutes is long enough for proper testing). I did three flights in both FSX and P3D, rebooting Windows between each flight.
There is little difference between FSX and P3D in terms of performance and fps at maximum settings. Occasional black texture tiles, blurries and stutters are present in both programs when you push ‘em. Both run relatively smooth when shadows, water and bloom effects are reduced or disabled. P3D has a few bugs that need sorting out and I even managed to crash it twice when fiddeling around in the menus during flight, something that never happened in FSX, although there are a few minor bugs to be found in FSX as well.
P3D isn’t FS11 – yet. Since it isn’t developed for the mass market I doubt it ever will be, but it’s a nice addition to the FS family. I use it as a scenery development platform. It works perfectly as a clean spare FSX install, so now I don’t need two different computers to develop scenery anymore. But if you think about P3D as a replacement for FSX I really see no reason to at this moment in time.