Ok, so lot's of stuff going on in this topic and way too much to try and post my thoughts on / input all at once - I'm verbose as it is without trying to tackle all the things that have come up in one go 
I'm also going to have to try and limit myself still for time here on the forums and my general VP time as I've been having a lot of issues with pain from positions and vision aspects working on my computer and seemingly resulting in chronic neck and head ache issues, which has contributed to me stepping back from VP and these forums a fair bit.
Now, for a quick start and more working backwards in things discussed lately in this growing thread. The first thing I think that is skewing some of the most recent discussion (today's) and detracting from honing in on a more consistent understanding and perception of where lag is and is not coming from is references and comparisons to VP9 and PM5 vs. VPX. vogliadicane and kernel have referred in their posts to, all other things being equal on their systems, that VPX is laggier than VP9 / PM5. This is almost certainly due simply to the vast difference in performance / FPS that those earlier 2 versions could attain vs. VPX (VPX lower of course for the benefit of nice features such as 4x AA and the new built-in lighting additions), but with a caveat as to why it would differ from one person to the next depending on configuration / preferences.
It is most likely that in these situations where people on the exact same setup "feel" VPX lag worse than another table in VP9 / PM5 are experiencing them while running in an open / non-vsynced configuration for all versions of the VP engines and thus FPS would be wildly different - in turn hugely affecting input lag and diminishing it. The older versions would be producing much higher FPS. This depends on course vsync being enable or not in VP but at least as far as VP9, vsync would not have been on because it broke the physics as the physics back then was still tied to the screen frame rate (wild flipper inaccuracies and couldn't be used).
This seems plausible / suspect as from kernel's previous post at least he also refers to the performance / FPS reaching 240 non-vsynced for some VPX games which seems to support this being the setting. It's not the case as much for some / the best performing VPX tables reaching a decent 240 in that configuration but VP9 and PM5 are probably then reaching 600-1000 FPS and improving lag aspects greatly from that element with the rest of system and lag aspects indeed held constant. Vsync setting / choice being a factor only because if it's on for PM5 and VPX then it's more apples to apples as it will be at (should be at if other things configured right) your monitor refresh. But anyone running non-vsynced / frame capped, will indeed notice a big difference simply because of the differences between the engines. It's a little indirect to compare the versions in general as well since one table from PM5 can't be immediately loaded (again holding as much equal as possible) into VPX and played. However, I can relay that on my Firewpower PM5 / BMPR hybrid and it's VPX counterpart with the table's kept very close in composition and physics settings (a seemingly forever WIP), on PM5 it runs at between about 900 and 1100 FPS, whereas on VPX (using 4x AA, reflections, and other eye candy lighting) it runs at about 150. This here is the largest culprit / difference in input lag between the versions and this pretty much alone.
VPX is more demanding but it's also not just for the hell of it. We just have more options, features, and lighting enhancements than before and most of use would like to use the things that are still options but make it look great even at the higher "cost" (4x AA is tough to go back from using once one has seen it). With those same Firepower table examples (flipper settings essentially the same but one PM5 and one VPX) with 4x AA off I get about 350 FPS in VPX but I also have lots of light objects, ball reflections, and table reflections.
But the point of all this isn't the options or what goes into eating FPS, it's that if you're not vsyncing or capping (and it seems there's enough out there that don't) then comparing a table on VP9 / PM5 to another table on VPX (whether the same title or not) is still comparing apples to oranges. If would be prudent to have more information, from people discussing the comparisons between engines for input lag variation, and should accompany this general type of feedback and have actual game time frame rates as well as VP preference screen captures or accurate listings. Plus for a proper comparison of the tables in question, we would need to be as identical as possible at least as far as flipper settings (physics values and position parameters). Also and more general and potentially important than just the engine version comparison aspects, all discussion about variations from end uses experience with input lag as far as VPX goes should be qualified with whether they run vsynced or not (and can confirm it's actually working F11 or FRAPS as expected) as well as if they're on a T.V. or computer monitor.
It seems like these aspects and extra detail in general is being overlooked and potentially steering things down an incorrect logic path and confusing the issue more than clarifying it. I'm also not trying to come down on the people who have posted this engine version comparison information, just that it's likely depicting the scene described above and therefor needs to be in the right context and / or accompanied with more details about configuration and FPS which would likely show how / why it isn't really a factor or a mystery as to why the older and newer VP engines - when FPS / vsync is not limited and performing at the max possible for that versions / table - are different for input lag.
As said, there's so much more to discuss and get onto especially regarding the direct fired flippers / bypassing pinmame (ideas and existing options for that) but that will have to wait for the next post (hopefully in the next day or so). This may not be the ideal "first post" back from me after an absence / in a while and especially regarding this overall topic category, but we're seemingly getting close to being off track here with several things and this aspect was one I felt needed raising and emphasizing.
Still keep ideas flowing but feedback needs to be more detailed and encompassing to draw more accurate conclusions from other people's experiences considering the wide array of configurations and options that people run.
Edited by jimmyfingers, 20 October 2016 - 01:28 AM.
Top
Contributor
are all trademarks of VPFORUMS.