868 replies to this topic

[toxie]

 

 

There is something wrong with the rendering. It looks like there is something on the right that pushes the table to the left. It is like for example you move a ramp to the right side and when you run the table the whole table is rendered to the left to fit the ramp. I attach a picture of the backdrop since it shows clearly the table is moved to the left.
 
 
drawing_error_2495.png
 
Edit: I have found that this happens if the table has no ramps on its sides (side rails), like for example my latest WOW Monopoly. This has been working fine until this revision 2495.
http://www.vpforums....&showfile=11612

Oh I guess that's the result of the transformation optimization Toxie did lately.

 

 

Indeed. But why??? :/

I'll fix this..

[toxie]

Damn.. I think i made a similar mistake already like a dozen times in my life now (bounding box * strange transformation != sane bounding box)..

[toxie]

..and i also finally submitted the change that one can use flashers with the DMD image now, which makes the CV display on the table itself possible.

[fuzzel]

rev2498 is up:

 

- introduce new checkboxes for textbox and flasher for script DMD image connection
- fix invalid optimization
 

[jpsalas]

Thanks fuzzel & toxie, now it works fine again

[Slydog43]

I think I have noticed some input lag when hitting the flippers in some VPX tables recently and started playing around with a few settings.  Could this be caused by having too many timers with update intervals of 1?  If this could cause this I'm wondering about core.vbs as it has PinMameInterval set to 1.  in the table scripts many timers are set to PinMameInterval. If I change this to 5 for vpx how do I have it change to 1 for vp9.x?  Maybe this shouldn't be changed.  Thoughts?

[toxie]

good catch.. i did not know that the core.vbs resets the interval on its own..

i'll add code that makes the interval larger if VPX is using it..

(but the pinmametimer should also be rather harmless performance-wise)

[toxie]

hmmm.. then again, i'm confused now, as there is this line:

Private Const conTimerPulse = 40 ' Timer runs at 25Hz

and this tells me that it was actually expected that timers run at 1000Hz (like its the case with VPX now), and is has been like this forever in the script it seems, same goes for the nudgecounter that has a 1000 in it.

[jpsalas]

hmmm.. then again, i'm confused now, as there is this line:

Private Const conTimerPulse = 40 ' Timer runs at 25Hz

and this tells me that it was actually expected that timers run at 1000Hz (like its the case with VPX now), and is has been like this forever in the script it seems, same goes for the nudgecounter that has a 1000 in it.

There are 2 timers in vpinmame:

 

PinMAMETimer that runs at 1, and

PulseTimer that runs at 40

[toxie]

yes, i know, but i just wanted to say that the pulse one seems to have expected to run with a 1000Hz timer already since a veeeery long time.

then again, this old comment could also be totally wrong, and it was always running fine with the 100Hz timer. 

 

so if one of you script wizards could do some experiments, if its worthwile to change the pinmametimer interval to 10 (or 5 or something) in core.vbs or if we could just leave it like is, then this would be awesome!

[gtxjoe]

Here is the overview from Destruk's Visual PinMame Scripting 101 tutorial (http://www.vpforums....?showtopic=4261)

 

The PinMAMETimer on the table runs a polling routine for your game that checks the light states, general illumination 

states, solenoid states, and playfield motors. The PulseTimer, as far as I know, closes and reopens the switch when 

the time interval has expired. Core.vbs lists conTimerPulse=40 so it runs at 25Hz, 25 times/second. 1000/40=25

 

These timing intervals are important. The PinMAMETimer interval is usually reset to 1, to get improved 

synchronization between Visual Pinball and Visual PinMAME. However, be careful about resetting or changing the 

PulseTimer value, as it might not work as expected if switch pulses are missed or lost (faster), or slower than the 

expected interval. Switch events are queued to prevent misses up to 5 in a row for the same switch number.

 

 

Since it is part of core.vbs, are there even any tables that have been tested/released with a PinMameTimer interval other than 1 msec?

[toxie]

Its all internally done in the core.vbs.

 

But i wonder where the 1000Hz come from. Maybe in the early VP9.0.X days it was still running at 1000Hz? I can't remember.. Or it was even before i joined the team..

 

 

EDIT: Oh.. Just forget about all the timer discussions.. Actually timers always ran (ideally) with 1000Hz, and it also was not changed with PM5 or VPX..  :/

Only the physics update rate was changed to also run at 1000Hz instead of 100Hz..

 

So sorry for all the noise!!! (me = stupid)

Edited by toxie, 08 February 2016 - 10:50 PM.

[jimmyfingers]

There really was an issue though with VP9 where setting a timer in a table script to less than 10 did not yield the expected result.  If you set two timers, one at 1 and one at 10, and had them display a counter increasing once per cycle, you can see from the output and a stopwatch that they were not milliseconds but 1/100ths (and most of all that they ran at the same counter increase rate).  That's, for me at least, where the initial talk of timers not working actually below 10ms came from.  I would try and dig up the test table or recreate a new one but I'm having to take a reasonable decrease in time from VP (and other computer aspects) for some health issues.  It's easy enough to reproduce though like I say with a couple text boxes, some counters and two timers set at 1 and 10.

[toxie]

Hmmm.. Correct.. Then simply forget my previous post edit..

 

And coming back to the comment from slydog: Could somebody please do some testing if setting the pinmametimerinterval to 10 instead of 1 in the core.vbs changes the behavior in a bad way?

[toxie]

the next build will scramble the UI components for the table around a bit. the main reason is that i wanted to combine all settings that the user might want to change for a specific table (like AA,FXAA,volume,night->day,etc) into only one panel, so that a newbie isn't scared of breaking things by changing things.

then i also changed some of the other panels slightly.

[gtxjoe]

Alright, I modified my core.vbs to have PinMameInterval = 10

Played 4 VPX tables for about 30 minutes and did not notice any issues.  I will leave it at 10 and see how it goes

[tttttwii]

I changed pinmameinterval to 5 and had no issues (tables AFM, MM, Dirty Harry, Robocop)

[jpsalas]

I also change it to 10 and it seems to work fine both for vpx and vp9 tables. As JF said the timers 1 to 9 worked as 10 in VP9. So I guess it should be safe to change it to 10. And if a table isn't triggering some events then it can be changed in the tables script to a lower value.

[toxie]

But does it also improve performance (you can look at the new counters for 'scripts' when you press F11)? Otherwise we can also just leave it like is.

[JMG]

i will try this tonight if i get to it and will check for performance improvement... i just upgraded to a gtx960 from a gtx650ti and, while everything looks a little nicer now and generally more fluid and smooth, i still get weird stutter-hiccups at certain times, both in vp9PM5 but more noticeably in VPX. Usually these occur quite randomly and for around 0.5-1.5s, so just enough time to notice it, if the ball is moving fast.

i also have the feeling that it happens when a lot of scripting is triggered simultaneously, also cpu load is very inconsistent, some tables vpx is at 10-30%, others >70% on all cores... (other than the graphics, i run a 3.1 ghz i5 3350P and 4 gb ram on win7 and ssd)