2087 replies to this topic

[Spektre]

If you would like to pass along pointers to the various routines that affect ball physics, I wouldn't mind lending a hand.

 

I find VP code quite difficult to follow at times.

[ClarkKent]

Didn't had much time to test today but I found the same strange behavior with dB2S and table at TSPP. Heavy ball stutter when a lot is going on on the backglass but high frame rate. And Homers head has a black square behind it.

[BananaBoat]

Is this stutter direct from VP or from within hyperpin?

Sent from my HTC_PN071 using Tapatalk

[LoadedWeapon]

just set the color on the playfield and plastic images to pure black not gray... takes just a min..

But that dont fix the head from turning into a giant flasher lol...

 Untitled.png   1.35MB   4 downloads

just tried ripley believe or not night mod

 

http://www.vpforums....s&showfile=8625

 

lights are off on dx9 at start of the table when they are lit in dx7

 

9.2 Sans titre.png

dx9 Sans titre.png

 

Edited by LoadedWeapon, 03 March 2014 - 10:33 PM.

[tipoto]

 

As for the lines on T2 chrome (T2_CE_Tipoto_1.07_FS).  I had them prior to changing my NVIDIA settings as here: https://www.youtube....h?v=pCk3E7hUSDU under v9.2.1xxx.  Under DX9 V7 I have a few back again.

 

I will take a look at this, I didn't try VP_DX9 yet.

In the current version of the table (1.07) there are a few offsets in some textures, because I had to compensate graphic issues we have with the DX7 version of VP. Maybe the DX9 version doesn't have these problems anymore and I can finally fix all my textures and make them consistent and logical.

Edited by tipoto, 04 March 2014 - 12:56 AM.

[arngrim]

clartkent, seems your mailbox is full, i want to sent you a pm

[ClarkKent]

clartkent, seems your mailbox is full, i want to sent you a pm

Sorry, I'll immediately delete some...

[ClarkKent]

Is this stutter direct from VP or from within hyperpin?

Sent from my HTC_PN071 using Tapatalk

There is no difference. Happens when starting it directly via VP and HyperPin.

 

If I deactivate dB2S and use VPinMAME Controller the stutter is totally gone. With dB2S the average framerate is about 400 but the ball stutters despite that. It's interesting that the average framerate without dB2S is about 500. But with such a high framerate the ball should never stutter!!!

 

So I think the dB2S has some influence on performance. Maybe I should try to revert back to server version 1.2.5 (instead of the latest 1.3.x) - there is a thread here in the forum of someone experiences the same problems and he reverted for good.

 

Maybe the server 1.3.x should be inspected?

[gtxjoe]

Is there function that can be called to read system time or a free running timer/counter that increments when the table is running?   I want to record some timestamps in the table script when certain sub routines are called

 

I guess I could simply create my own timer, but if system clock exists already, I would rather poll that

Edited by gtxjoe, 04 March 2014 - 01:35 PM.

[Argo]

Even Congo, a table which used to stutter a lot, now plays very well in VP dx9

[koadic]

Is there function that can be called to read system time or a free running timer/counter that increments when the table is running?   I want to record some timestamps in the table script when certain sub routines are called

 

I guess I could simply create my own timer, but if system clock exists already, I would rather poll that

 

The system variable 'Timer' will give you the system time in seconds down to hundredths of a second... for example 8:18am exactly would be 29880 (8 hours for 28800 seconds plus 18 minutes for 1080 seconds totaling 29880)

[DJRobX]

 

If I deactivate dB2S and use VPinMAME Controller the stutter is totally gone. With dB2S the average framerate is about 400 but the ball stutters despite that. It's interesting that the average framerate without dB2S is about 500. But with such a high framerate the ball should never stutter!!!

 

 

I would like to see maximum inter-frame delay (in ms) added to the F11 display, perhaps resetting every 10 seconds.  

 

For smooth 60fps play, every frame needs to be rendered under 16ms.    It sounds like the physics engine needs it to be even less (10ms?) for it to perform optimally.      500fps doesn't mean good performance if you're rendering 499 frames in 100ms but the last one is taking 900ms.   

[mukuste]

I actually added that, but only in Debug builds, so it's kind of useless for you. I'll get that activated in all builds.

 

One thing I've also noticed is that if I have, say, Firefox with many tabs open in the background, some presumably running Flash etc, then the game stutters pretty much independently of framerate. I guess the db2s effect is similar. Anyone know how often db2s renders -- does it have a framerate limiter?

[PilzTom]

I had these stutters with db2s backglass before the dx9 port was started. Worst table was nba fastbreak. What made me really going crazy was that it worked before I tried some different graphic setup. I changed everything back (all 3 monitors attached to the connectors they had in working setup) but still bad stutters...
When I had a look at windows screen settings I saw that windows changed the enumeration of displays. So playfield monitor was not screen 1 as before but 2. Backglass was 1. Dmd 3. Order was 2-1-3, so everything was shown correct. After changing playfield screen back to 1 stutters were gone.
In more tests I got also stutters if playfield screen was not the most left one in windows display setup (screen order).
Since I follow this I have no issues...

[ClarkKent]

This can not be the reason on my system - monitors are in the right order. A friend of mine has the same components in his cab and has the same stutter with certain dB2S backglasses.

I also just tries to revert to server version 1.2.5 but this has nothing to do with the stutter.

At the moment I'm out of options other then using an other backglass which does not cause the stutter or turning dB2S off for that table.

[arngrim]

I had stutters on some tables like that, but i realized the db2s was not in exe mode, that fixed it for me

[mukuste]

I just saw that Herweh has some performance tips here: http://www.vpforums....showtopic=23364

Might be worth giving a shot.

 

Also make sure you're using the latest version.

Edited by mukuste, 04 March 2014 - 06:03 PM.

[jimmyfingers]

I'm finding some micro stutter as well but can also get it even without B2S loaded. However, one thing I noticed with B2S in the mix is that I'm seeing results that are opposite as to the previously recommended approach by allowing it to run in it's own executable. When I've switched back to not be an independent .exe for DX9 tests I'm actually getting less stutter. However, I'm mostly doing my testing with FXAA forced on via the nVidia control panel (I have seen some artifacts and other ball glitches using the built-in versions on even the default / template table and is why for now I'm still using the nVidia options). I can see a significant performance drop with FXAA enabled so the micro stutter is going to be more prevalent and potentially affected differently when combined with B2S but I wanted to mention my initial summary findings in case anyone else has experienced this phenomenon too or wants to check it out and see if this helps. One thing that's a clear indicator that this reverse approach is not the same as before is on the couple tables I've tested at least, there is no flipper sticking issues as was normally the first thing someone would notice if the DB2S backglass was not in .exe mode.

[DJRobX]

I actually added that, but only in Debug builds, so it's kind of useless for you. I'll get that activated in all builds.

 

One thing I've also noticed is that if I have, say, Firefox with many tabs open in the background, some presumably running Flash etc, then the game stutters pretty much independently of framerate. I guess the db2s effect is similar. Anyone know how often db2s renders -- does it have a framerate limiter?

 

Are you able to run B2S/VPM tables in debug mode?   When I try, the B2S.Server or VPinMame.Controller fail to initialize.   If I take them out I can still load the table (which usually lets me see whatever it is I'm going after) but it would be nice to run the whole enchilada debuggable.

[mukuste]

 

I actually added that, but only in Debug builds, so it's kind of useless for you. I'll get that activated in all builds.

 

One thing I've also noticed is that if I have, say, Firefox with many tabs open in the background, some presumably running Flash etc, then the game stutters pretty much independently of framerate. I guess the db2s effect is similar. Anyone know how often db2s renders -- does it have a framerate limiter?

 

Are you able to run B2S/VPM tables in debug mode?   When I try, the B2S.Server or VPinMame.Controller fail to initialize.   If I take them out I can still load the table (which usually lets me see whatever it is I'm going after) but it would be nice to run the whole enchilada debuggable.

 

 

I definitely do all my debugging with Debug builds. I don't have B2S but VPM makes no troubles. Not sure what could be the issue there. Make sure you're linking to the correct runtime (/MTd, multithreaded static with debugging).