it would be cool if someone did that but compared:
== already in video ==
- terp 64, pft 0
- terp 0, pft 0
== for future science ==
- terp 64, pft as 1.40 has it set
- terp 0, pft as 1.40 has it set
Recording in question (youtube link supplied on prev page -
YouTube - terp = 0 vs 64) is of a player
with ~28 ping (cannot emphasise this enough) who was using 1.4 (though this fact should be immaterial right?).
Filming & merging to compare player positions under different combinations of 1.3/1.4/pft/terp (fraps/v.low timescale/manual merging -- this was done several months ago so some of the details are lost to the sands of flaky memory) gives us
four cases to compare:
#1 - 1.11 stock (64,0) vs.1.4 stock (64,?) - effectively no difference
#2 - 1.11 (0,0) vs. 1.4 (0,0) - effectively no difference
#3 - 1.4 stock vs. 1.4 (0,0) - what you see in the youtube link above
#4 - 1.4 (0,0) vs. 1.3 (64,64) - very little difference other than lack of stutter in (64,64).
#4 is imo the most interesting. It seems to imply that a terp value >0 lags the drawing of other players etc by that value (in milliseconds). Setting pft = terp seems to compensate as exactly as possible for that lag, albeit by using a less accurate predicted value. At low pings, this is not presumably much of an issue.
The above effects hold largely true for combinations of terp/pft tested with netset on 1.4 up to 256/256. Higher PFT values push the player further and further forward, while higher TERP values 'lag' the player in an almost exactly equivalent manner. To reiterate, setting terp = pft counteracts the effect for all values 0-->256, simulating (with decreasing levels of accuracy) the case where terp=pft=0.
HAY LOOK A FUCKING PICTURE
In summary, 1.4 (stock) appears to offer no particular advantage over 1.3. If you believe terp = 0 in 1.11 offers you an advantage over terp = 64, then there is little reason to believe terp = 0 will not offer the same advantage in 1.4 - namely, a slightly less delayed player position.
I am aware this 'analysis' is valid for a few seconds of a single recording. If you want to extrapolate, go for it. I was tempted to include a section on how nofix/opsayo appear to be spewing shit when they state 1.4 stock>*, but I know can't really do that. If you want verification, get somebody else to recreate it. If you want a proper test, do it yourself. After all, it sounds very much as though there's a 1/3 chance netset is already on your hard drive.
I don't pretend to understand properly any of the theory involved. I don't even claim to be the one who carried out this testing. This terp/pft 'discussion', however, could do with a little less theory and a little more evidence - perhaps this can be a start. Also fuck you all, maybe I do give a shit.