666 replies to this topic

[htamas]

and another rev852 this one introduces a new video option 'Vertex Buffers in VRAM'. If you enable this VP will store all 3d model data on the memory of the graphics card. It can improve the performance on some tables/systems but that depends on your system, graphics card and driver of the graphics card. The default setting is disabled like in previous versions.

 

I ran some tests using rev.849 and 855, with the latest Starship Troopers as the test table.

 

The "Vertex buffers in VRAM" option actually causes a drastic loss in FPS on my system, going from average 400 in 849 to about average 220 in 855, and even shows microstutter in 855 while 849 is perfectly smooth. So my rig is obviously not one of them where there is a performance improvement (the video card is a GTX 560 with 1 GB GDDR5 memory, and I run the good ole' XP with the ancient 270.61 driver)

 

Interestingly, I also lose some FPS between 849 and 855 with this option disabled (from 400 it goes down to about 360) but the effect is by far not as bad as with the "vertex buffers in VRAM" enabled.

Still, there is a noticable drop just by going to 855. I wonder what would cause that...

Edited by htamas, 09 January 2014 - 09:35 PM.

[toxie]

Maybe its the change of the ball rendering/disabled lighting..

As there is a problem with the ball color anyhow, i think i'll revert my changes, too, and think of a new variant.. :/

 

As for VRAM: I have roughly the same setup, newer drivers though, and i also do not benefit, but thats why thiis thing is selectable..

[fuzzel]

 

and another rev852 this one introduces a new video option 'Vertex Buffers in VRAM'. If you enable this VP will store all 3d model data on the memory of the graphics card. It can improve the performance on some tables/systems but that depends on your system, graphics card and driver of the graphics card. The default setting is disabled like in previous versions.

 

I ran some tests using rev.849 and 855, with the latest Starship Troopers as the test table.

 

The "Vertex buffers in VRAM" option actually causes a drastic loss in FPS on my system, going from average 400 in 849 to about average 220 in 855, and even shows microstutter in 855 while 849 is perfectly smooth. So my rig is obviously not one of them where there is a performance improvement (the video card is a GTX 560 with 1 GB GDDR5 memory, and I run the good ole' XP with the ancient 270.61 driver)

 

Interestingly, I also lose some FPS between 849 and 855 with this option disabled (from 400 it goes down to about 360) but the effect is by far not as bad as with the "vertex buffers in VRAM" enabled.

Still, there is a noticable drop just by going to 855. I wonder what would cause that...

 

Thanks for all the infos. All the latest versions are basicly tests to see how the compatibility mode of todays gfx cards behave. I don't have a clue why this behaves so strange. On my system (I'm using a AMD HD6570/7770) I don't see such a dropdown in performance. I lost just a few fps (if any) with the dynamic flippers and using the VRAM option it is equal or sometimes better than disabled... I think we oversee something in the code which cases all this because rendering some flippers on the table is really a joke for todays gfx cards (one flipper has only 270 polygons) and if you compare it with all the meshes used in newer table releases it's really hard to believe that we can't switch to dynamic objects.

[toxie]

IMHO its simply the legacy modes of DX and on top of that the legacy modes of the drivers..

Otherwise it wouldn't be fast on some systems and super-slow on others..

[toxie]

The ActiveBall.Color method is not working in rev 851.

I'm not sure, but I think using a custom Ball.Image defeats the default lighting.

 

 

Actually the Color method is broken since a looong time, i think even since the balls are rendered with vertex buffers..

 

But i now introduced 'DisableLighting' to the script (rev 859), too, so with this one can make 'Color' work again.

 

So just set Ball.DisableLighting = true and Ball.Color = RGB(255,255,255) to have the ball always render with its original image, for example.

Edited by toxie, 09 January 2014 - 11:50 PM.

[toxie]

btw: is the touch screen input now working?

[sleepy]

I don't have a touch screen, and I noticed that mouse clicks activate the Flippers in rev 851 when the Debug > Throw Balls in Player box is Unchecked. I'm guessing this is deliberate?

 

When the Throw Balls box is checked, I can accidentally throw balls outside of the playfield where they get lost and can screw up the ball enumerations(?). For instance, if I toss a Ball off of the playfield so it gets lost and then click in the Plunger lane to create a new Ball without tossing it/without moving the mouse, and then Right-Click the new Ball in the Plunger lane > choose "Set as ActiveBall", the new Ball then vanishes. Enumeration miscount?

 

Is it possible to bind any Balls to the Outer Wall of the playfield, or simply set the Ball's position .X .Y .Z not to exceed the playfield's base parimeters? Something like If IBall.X > Table1.X Then ...

 

From time to time, Balls also sometimes fly off of Ramps and get stuck in the corners of a set of Walls, usually when the Ball is hit hard and direct to a Ramp that is curved and functioning like an Up lane or an Orbit, or sometimes flies off and outside of the playfield where it gets lost and the game freezes. Is it possible to keep a Ball from flying off of a Ramp?

 

 

Also, the ActiveBall.Color method was working sometime around rev 866 or later.

Edited by sleepy, 10 January 2014 - 07:47 AM.

[bosvrucht]

btw: is the touch screen input now working?

 

Touch flippers are working!  They are setup for desktop tables only, however (because FS tables are rotated, without windows being rotated, the L/R division changes)

 

Another issue is that for an input to register i have to swipe the screen (just touching is not enough). This is probably a machine specific issue, because the same thing happens for gestureworks gameplay.  It could be due to the implementation.  The opensource PSP emulator PPSSPP which runs fullscreen also has this issue, but only during gameplay. In the settings screen (which is also fullscreen, at the same resolution as the gameplay), touch input registers normally.   Again, this seems specific to  my machine. Everything should probably work normally for Surface pro users.

[toxie]

Cool.. In theory fuzzel also included support for FS, but maybe there's still a bug..

[fuzzel]

I don't add bugs just surprises The reason why that doesn't work is that the screen orientation isn't rotated. Normally I divide the screen width (left/right) but if the table is rotated by 270° I divide the screen height so left is bottom half and right is upper half of the screen height.

[toxie]

@fuzzel: as for the gestures necessary: maybe handling the previously mentioned pointer messages in addition could help there?

[bosvrucht]

I don't add bugs just surprises The reason why that doesn't work is that the screen orientation isn't rotated. Normally I divide the screen width (left/right) but if the table is rotated by 270° I divide the screen height so left is bottom half and right is upper half of the screen height.

 

Humble apologies, it DOES work as expected.  I just assumed Fuzzel did not think of the FS rotation issue, and failed to test it with an actual FS table.........

 

Here is a quick/dirty video of Embryon running FS, with touch, on my 1.33 ghz intel ATOM tablet, with integrated graphics (CPU+GPU have a total TDP of 2 watt):

 

[fuzzel]

haha how cool is that

 

besides that have a look at rev861. I added a table dimensions manager. You can find in under Table -> Dimensions Manager... with this you can calculate real table dimensions into VP units and vice versa or just select on of the predefined tables. The predefined dimensions come from a list found here in the forum.

Attached Files

•  dimmanager.jpg   47.67KB   9 downloads

Edited by fuzzel, 10 January 2014 - 04:31 PM.

[toxie]

Double WTF!

 

I don't want to order a windows tablet..

But i feel eeeeeextremely tempted now..

[Slydog43]

so funny, as I'm heading over to ebay now to buy one just to play with, damm you guys

[kruge99]

haha how cool is that

 

besides that have a look at rev861. I added a table dimensions manager. You can find in under Table -> Dimensions Manager... with this you can calculate real table dimensions into VP units and vice versa or just select on of the predefined tables. The predefined dimensions come from a list found here in the forum.

 

Does it let you create the default table at the size specified from that list?  If not, it would be c00L if that could be implemented. 

 

For example, say I want to build a 1980s Bally standard body recreation, the default table is 'created' at the proper size in VP units.

 

 

Best Regards,

Todd

[toxie]

so, while we're at it, seems like all windows 8 8" tablets in question (intel z3740) currently are kinda equal (dell venue 8 pro, toshiba encore, lenovo miix2), or?

anybody has some more info? are the graphics exactly the same in all of them?

 

@bosvrucht: are the tables slowing down after a while of playing or is it kinda stable? (automatic downclock)

[bosvrucht]

so, while we're at it, seems like all windows 8 8" tablets in question (intel z3740) currently are kinda equal (dell venue 8 pro, toshiba encore, lenovo miix2), or?

anybody has some more info? are the graphics exactly the same in all of them?

 

@bosvrucht: are the tables slowing down after a while of playing or is it kinda stable? (automatic downclock)

 

Yeah, currently there are only a few smaller/affordable windows tablets, and they are all basically the same (the Toshlba has HDMI and GPS, the others are smaller/lighter but dont have hdmi/gps).   In the future you will probably see more variety.   It is only since windows 8.1 (which just came out) that windows supports smaller tablets, and official support for autorotation/portrait use.    The new bay trail atom has a great mix of price/performance/battery life.   In the coming months you may see faster tablets, but at what price?????  These are just 299 euro....  and for browsing and light gaming they are great.  Civilization 5 has full touch support and runs okay on low detail settings !!!

 

One thing that does tick me off is Microsofts stance on NAS-support (or the lack thereoff).  The microsoft metro apps, and framework do NOT support non-microsoft NAS solutions (thats means basically all of them). This is a conscious choice to peddle their own wares, and really screws consumers.  Also 32 gb (23 GB even, if you dont remove the repair partion) of storage is REALLY small for a complete windows install+programs.

 

As far as VP performance goes:  i have not tried prolonged sessions, because of the weird swipe to register touch issue that plagues some legacy applications (including VP).  As stated,  AFM 2.4 runs great, but Apollo 13 is definately not smooth enough. Embryon was great, extremely playable.   If you want all tables to run great you will have to buy a surface pro2 or something for 1000+ euro......

Edited by bosvrucht, 10 January 2014 - 07:51 PM.

[koadic]

As far as VP performance goes:  i have not tried prolonged sessions, because of the weird swipe to register touch issue that plagues some legacy applications (including VP).

It effects more than just your tablet (or other atom tablets) as I experience the same 'swipe to touch' issues on my i5 tablet

[toxie]

whats the 'wipe to register touch issue' about?