Found a dyno test of a CF Moto 650 in the latest AMCN and the chart bears out my previous statements. However, with the stock filter it makes 39.30kw, a foam type 39.83kw, a fabric type 40.34kw and no filter 41.47kw. The numbers are not huge, so, unless you are chasing absolutely every last pony (as I do for the sake of my ego); go with the most convenient filter.
Pretty much as I remember the relative performance. While difference gets bigger with higher power engines (basically due to higher airflow) the trade off is always filtering/service cycle.
In my experience foam, if properly prepared (and this is actually harder than for oiled cotton) is fine, but has a shorter service cycle than oiled cotton and loses flow more quickly with dust build up. I used to work on a lot of rally cars back in the day fitted with foam cleaners and there was always fine dust/grit in the barrel of the carbs on those. I consider it a last resort when nothing else will fit the available space.
More recent testing on VFR Honda's (one of my "other" bikes - slightly modified VFR 800) has shown the OEM paper filter flows as well as any aftermarket (effective) filter. BUT, It lacks the "cool" factor, AND costs about the same in terms of "whole of service life" for those of us that don't turnover a bike every other year (mine has 260,000K on it of which I have done 230,000+k over 13 years), thus to an extent, de-buncking the "cheaper than OEM" replacement argument so often used when the objective evidence is presented.
I spent a lot of time with John Bennett (originally of Head Mod fame, which spawned the Crankshaft rebuilders in Kew and HM headers, and....,a number of other now well known busineses from the ashes when he broke up the business in the 70's) during the 80's watching his analytical and objective testing of everything performance automotive. It was an eye opener to see what actually made a difference and what didnt (like far too many of the speed components on the market). He also introduced me to flow benches, stress plate cylinder honing and careful, step validated blueprint engine building with dynomometer validation of cylinder flow testing.
We used a leak down test of each cylinder to verify perfect ring sealing, before, final assembly. If it wasn't, it got pulled down and rings (and valves if required) re lapped. His race engines were performance characterised on the engine dyn (under controlled conditions) and then re tested in the car on a rolling road. If the installed engine didn't make the power achieved on the engine dyn, the installation was methodically reviewed and modified until it did. This analytical approach and attention to detail was unheard of at the time (but allowed/maybe together with a holistic engineering approach of the car (arguably)' Moffat to become a legend in the Coke, later Brut 33 Mustang.
A fitter and turner by training (but clearly with a deeper and more intellectual engineering insight), John is a bit of an unsung and (for some) forgotten hero of the Australian performance and racing scene from the 60's on. This gives you some idea:
http://www.ozlaverda.com/australia-s-quickest-chair-dpa-laverda-t59.html
http://www.ozlaverda.com/john-bennett-head-mod-engine-development-t97.html
At the time he was pretty much the only guy running air filters on race engines. The prevailing philosophy of the day was filters cost horsepower, John contends that the dust ingested while racing had a far bigger impact on performance over a race (and proved it with many before and after engine testing on a controlled enviro dyno). His engines not only finished as strong as they started, they also went a whole season with only oil changes between races on class leading engines. At the time he also was the only guy in town that regularly calibrated his dyno (and ran them under controlled conditions) so he could compare performance over time on a standardised basis.
Oh and check out the 10,000rpm from a ford 144ci engine (yes the one std in the 1963/4 XL/XM falcons) John built in that second link. The old 144 was an interesting design, light (hollow crank and cam) and compact (2.4L 6 cyl). John machined off the built in inlet manifold, fabricated one to take triple weber carbs, modified the head flow, balanced and blueprinted it designed and built a work of art exhaust manifold. From memory it made around 250 hp, on modified std components . Didn't just make the hairs on your back stand up' as jump off! Went into a fairly stock early Corolla van where it became rather notorious in the Northern suburbs of Melbourne, and beyond!!!
I have lost track of John over the years but after recounting the stories to my son recently have begun chasing up his whereabouts.
Have only come up with this so far:
http://www.bennettclayton.com.au/
If there are any old petrol heads that are in contact with John still, I'd appreciate some contact details (by PM).