338 replies to this topic

[toxie]

Yup, of course, and it's actually a) an equivalent code transformation for the transformation matrices and b) only touching the primitives code.

And, as said, i don't see a difference on my machine for 442 vs 444, but rather for 467 and 470!  (which btw also has nothing to do at all with the ball updates/redraws)

 

That's why i think that this bug is not related at all to these specific changes, but rather was just -revealed- by these, due to different stack layout, memory locations, etc.

[jpsalas]

I just did a little more testing, and I turned on Adaptive Sync. And after a few seconds that bad movement went off and it became perfect, with a constant frame rate of 60. I tried all the revisions and they all worked perfectly. So I turned off Adaptive Sync and it still was perfect movement (?!). I even have restarted my computer, and it is still perfect on all revisions. The only option I have on both computers I have tried the table (one with an Nvidia DTX 680 and the other with an Ati 5xxx) is the anisotropic filtering, no AA, or anything else. So I don't understand anything The only difference between versions is the frame rate, I get a frame rate between 330 and 400 on all the latest versions, but I get from 450 to over 500 in the older ones, like 442 or 434. All these tests was done with the Star Wars table.

[toxie]

That just seems to prove what i was saying.. There is some evil memory corruption or something hidden somewhere, and depending on some other code, it leads to that ball vanish/missing updates or not..

That will be fun to debug.. :/

[RipleYYY]

as said already, tweaking, always and always... but its different for all of us as nobody got the perfect SAME system+soft as his neighbour...

 

[jpsalas]

I have been repairing computers since 1984, and I have been a system administrator since 2000, and I still found unexpected things where they should never happen.

 

Well, now I have turned off the computer, remove the power, turn it on again, and the latest 476 version is perfect (just about 100 frames slower than 442). Why it was doing that before is beyond me. I can't try the Ati card since it is in the computer at work. So I think you don't need to debug anymore and move on with the next fixes, whatever they may be

 

The slow feeling that RipleYYY is having may be of the general less frames that 476 has compared to 434 or 442. I don't know if I can trust the framerate (by pressing F11), but a framerate from 500 (revisions 434/442) or 400 (newer revisions up to 476) is not noticeable in my computer (but I have a monster computer). Those less frames may also be because of compiler options, if koadic has used other options, like removing the SSE extensions (which used to cause the ball through objects when destruk was compiling VP).

 

I must say I'm really happy you (toxie) and fuzzel are working on VP. I was going to give up to make more tables for VP, since after nearly 8 years doing tables I started to get tired. But by introducing the primitives and the several fixes it was like a  new fresh breeze of air So expect more tables with primitives from me

Edited by jpsalas, 16 March 2013 - 01:48 PM.

[Slydog43]

wow I got so scared that you JP might stop making tables.  So glad the new items in VP has kept your interest.  I have been working with computer also my whole life since I was 13 (I'm now 45).  They drive me crazy since they are suppost to act the same every time, but with windows, that simple is not the case.

[toxie]

@JP: Nevertheless i'll continue searching for this..

There is definetly some bug somewhere with the ball drawing, that's for sure.. :/

[ViriiGuy]

I too have been experiencing an extreme slowdown with pretty much every version I have tried. However the slowdown is only on a few tables so far. Big Bang bar and Al's garage band being two of them. It seems to me like it is only tables that use the Pi function but I may be wrong. I am still doing some testing.

But on the tables that slow down, the ball looks like it is dropping a lot of frames, as it almost warps from one place to another.

My cabinet PC is a bit of a monster and previously has not had any issues running any tables. My video cards are both nvidia 460 superclocked's.

Beyond that, great work!! And I see from testing that the Pi error was fixed for Xp 64! Well mostly anyway. Besides the slowdown.

[ViriiGuy]

Ok, well I did some more testing on my end. I even decided I would rule out my video card. So I bought a 660 GTX TI and installed it as my primary. With the 460 GTX SSC I was seeing about 45 fps on Al's Garage Band. And on the 660... 50 FPS. So definitely not a worthwhile improvement But what I did find out, is if I turn off the full screen Anti Aliasing, I get up over 200 FPS on both setups. So, for the time being, Iw ill just leave the AA off.

[ViriiGuy]

It's me again I have spent most of the weekend working on my cab and testing out these new builds. It seems all of the 9.15 builds have an issue with sound. I can MUTE the sound on my PC in windows, but the 9.15 builds still play all the table sounds, like Flippers, slings and bumpers. As far as I can tell, it seems to exist in all of the 9.15 builds.

 

My sound card is an Onboard Via.

[koadic]

Do you have the sound split up going to multiple outputs? or is it a basic setup with all sounds (music/rom/table) going to one set of speakers?

[LoadedWeapon]

yea thats what it sounds like.. Are you plugged into the front audio jack? it will split up sound also..

[ViriiGuy]

No I do not. As a matter if fact I even disabled my HDMI audio device in trying to determine what happened to my sound. I am only using the one output which goes to my Boss 400w amp.

Edited by ViriiGuy, 18 March 2013 - 02:22 AM.

[ViriiGuy]

Ok, after more testing.. This seems to have been an OS issue. When I loaded up XP 32, the problem is gone.

[blur]

having sound or not having sound is always os issue - if you set sound level on a output device to mute - there is no way you can get sound from any application - only if there is some other device that is not muted

[ViriiGuy]

Good to know!

[jpsalas]

@JP: Nevertheless i'll continue searching for this..

There is definetly some bug somewhere with the ball drawing, that's for sure.. :/

 

Yesterday the slowlyness started again. It is like the ball is redrawing wrong with very strong tearing, something that makes the ball redraw in steps and sometimes only displaying a small part of the ball, I don't know how to explain it better, but he movemnt is not fluid at all, even the flippers show a delay. But I start to think it is Windows 7 playing tricks on me, but the strange thing is that versions up to 442 plays fine, and the newer ones don't (maybe a change in a compiler setting?). And the question is why it works sometimes?

 

I tried all the versions, from 434 to 467, with the work computer with an Ati 5xxx and all the versions play the same, I can't see any change at all. This computer has always been slow. So it plays just as expected, since it is slow playing any VP table (vpm or original).

 

I hope this helps in tracking down this error.

Edited by jpsalas, 19 March 2013 - 07:15 AM.

[toxie]

The main problem is that there wasn't really any stunning changes directly after 442 that could cause this..

(as said, one change just fiddles with the transformation matrices (tiny change, just to unify/simplify code), one with the primitive update code (exclusively), so if there are no primitives, this should never be triggered at all)

As for compile flags, i guess koadic should comment on this (btw, @koadic: maybe you could also make a build with 443 so that JP could test that, too? Just to nail it down more to a specific checkin that causes this for JP)

[ringorian]

Is it possible that its a thing oft the gfx card or CCC ? Profiles?

[koadic]

My compiles have been using the same settings from the beginning... fast + sse, nothing has changed in that aspect.