fraps is basically not effecting anything while doing the benchmarking. its what almost everybody uses to check real in game results because it is very accurate and consistent. only if its recording video does it have an impact on performance. and actually I am testing a somewhat scripted part towards the beginning of the game for part of the bench. the other part involves just walking around the same path with little going on.
I spent hours testing the demo with different cpu and gpu speeds before I ever got the full game so I am very familiar with the performance of Batman. my cpu at 1.8 is most certainly sluggish during large fights and that is reflective during the fraps benchmarks. I can just turn on fraps and clearly see the framerate difference between the cpu speeds even without benching with fraps. the only thing I havent done is test Batman with physx off and my cpu at 1.8.
I hate to break it to you but the built in benchmark for Batman is NOT accurate and thats why I dont use it. I am a member over on Eidos site so I know this for a fact. there are people with killer machines that will still get low 20s for minimum framerate. if I personally was getting consistent results then I would use the benchmark but I am not. hell on some occasions it will show me have a max of a 120fps while on the very next run be at 80fps with no changes. when I get home today I will go dig up a thread that came out when the game did that proves this too.
EDIT: I had time to look and here is one of the threads on the Eidos forums. as you can see the built in benchmarks are very crazy and not reflective of actual gameplay for some. sometimes people with gtx285 cards and dedicated physx cards are getting in the 20s for a minimum. in fact I think only 2 people using Nvidia got in the 40s for a minimum while everyone else was in the 20s regardless of what level cpu and gpu. of course the ATI users arent running gpu physx so I just ignore their results.
http://forums.eidosgames.com/showthread.php?t=95824
even the mod on there said:
"Remember the minimum is more then likely a dip in the performance recorded during a transistion between scenes. So is a bit of a false minimum since it could be a single frame with that rate but as the system records ALL frames it reports this as the minimum."
"Yeh i know what you mean the minimum is the key figure during gameplay, but this benchmark isnt really representative of that its just a flythrough a level and shows some stuff off."
anyway so the best way to test Batman accurately and consistently is using fraps during the actual game. even if you somehow are getting fairly consistent results with the built in benchmark they are NOT accurate and certainly not reflective of in game results.