Inspired by the problems I've been having in pCARS I did a fairly extensive comparison of my racing sims and how quickly the on-screen action responds to my steering inputs. I used a Casio Exilim camera shooting at 210 fps, using a monitor with a refresh rate of 60Hz and made sure I had the FPS always over 70 in every game. I compared pCARS, rF1 and rF2, Live for Speed and netKar Pro.
The results went in the exact order I predicted. pCARS is noticeably the laggiest, rF1 and 2 are OK, but LFS and nKP just feel more instant (even though as a whole I like rF2 better due to the FFB, not to say it's bad in LFS and nKP, but I prefer rF2's implementation). I also do a lot of drifting in sims, and the same order applies: nKP is the snazziest feeling when drifting, though some cars that were better suited for drifting would be appreciated, LFS is also great, rF1 and rF2 work but aren't quite there, and pCARS is just about impossible for me.
Now to the actual results. I measured the time from when my G25 moved to when the on-screen wheel moved. I did multiple takes and (especially in the case of nKP, read on to see why) checked every clip multiple times to make sure. The biggest deviation between tests for a game/settings combination (pCARS with DX9 and DX11, with two alternative exes) was 1 frame, so I'm fairly confident of the results. I'm rounding a bit due to the fact that the camera shoots at a faster rate than the screen refreshes:
pCARS: 0.1 seconds in every mode. Instantly noticeable in feel and visible to the naked eye when looking closely. Though they're clearly working on this and I'm confident they'll get to around LFS level in a year. *EDIT: As of build 141 pCARS has improved immensely for my system and is now somewhere between 0.06 and 0.07 seconds, meaning performance equal to or slightly better than rFactor. It also transformed the game for me, it's vastly better now.*
rF 1&2: 0.07 seconds. Better, about the same difference between them and pCARS as my TV has between game mode and normal mode. FPS games are still playable in game mode even though it's slower than my monitor, but in normal mode it's almost unplayable. Well, not unplayable as such, but not any fun. *NOTE* I didn't use my TV for the test, just mentioned it as a reference.
LFS: 0.05 seconds. Much better, you can feel how much easier it usually is to straighten out slides without being late and prevent tank slappers than in your average rFactor street car mod. More instant, more natural, though I'm not a huge fan of the FFB when compared to the one in rF2.
And last but not least, the BLISTERINGLY fast nKP. I didn't even believe these numbers at first, but I had to accept them after checking them over and over again: 0.03s (actually 0.025, but 0.03 is ~2 frames at 60Hz). That's just about as low as you can go, and it really does look and feel instant in every way. But again, I do prefer the FFB of rF2, not to knock on the one in nKP either.
Such big differences which were constant between runs and especially the fact that rF1 and rF2 performed equally (within a frame at 210fps) suggests it's a property of the game engine used. If so, I really do hope that ISI will take this seriously because everything I've seen and felt in sim racing suggests to me that going from 0.07s of lag to 0.03s does more good than going from stock rF1 FFB to the much improved rF2 FFB. It is just a huge difference.
I understand that if they decide to tackle this it's going to require a lot of work, so I'm not expecting it to happen soon. But now that ISI has a steady flow of money to expect from rF2 I hope it can be achieved in the future.