Jerking rtsplayers after porting to 1.52
by Thomas Huehn · in RTS Starter Kit · 03/26/2008 (3:47 am) · 12 replies
I noticed, that the rtsplayer run not very smooth after porting to 1.52. I repeated the patch on a clean engine but the problem is still there. Is it only me or did anybody else notice this ?
About the author
Contact: torque [AT] ohmtal [DOT] com
#2
03/26/2008 (9:52 am)
They both look the same to me on windows side. Also maybe try switching to directx/opengl and see if any difference. 
#3
03/26/2008 (6:08 pm)
Thomas could you give some details on the weird behaviour you are getting?
#4
FPS is ok so i think it must be something else.
03/28/2008 (4:43 am)
I dont know how to better describe it ;) I start the default mission, select units and move them. On the 1.3 engine it runs smooth and with the 1.52 patched exe the units and the selection circle move unsmooth. FPS is ok so i think it must be something else.
#5
And maybe you should make the port process once more, paying atention to the patch DOS window, to see if the patching process is beeing made without errors.
03/28/2008 (3:28 pm)
Its weird. I would start with the classic, first delete all .dso files and try again. If the problem persist, clean, recompile, and try again.And maybe you should make the port process once more, paying atention to the patch DOS window, to see if the patching process is beeing made without errors.
#6
05/01/2008 (4:06 am)
Finally I got it working. This time I did not use the patch and made all changes myself digging the code with "trial and error" ;) 
#7
For a comparison, I quickly changed the RTSkit to use AIPlayers instead of RTSUnits, and they move silky smooth.
I have been porting the kit over to TGEA 1.7 this week, and I'm noticing the movement issues even more with that engine.
My gut feeling is that this is some kind of interpolation/prediction/warp issue, but I'm not exactly sure.
Any chance you could share your findings?
(BTW - I have not yet tried the official 1.5.2 RTSKit download)
05/01/2008 (9:10 am)
After looking closely at my RTS 1.5.2 builds and comparing the units movement to 1.3, I know what you're talking about. The motion of the player is not as smooth as it should be. When I zoom in close, there are subtle 'glitches' to the units movement. For a comparison, I quickly changed the RTSkit to use AIPlayers instead of RTSUnits, and they move silky smooth.
I have been porting the kit over to TGEA 1.7 this week, and I'm noticing the movement issues even more with that engine.
My gut feeling is that this is some kind of interpolation/prediction/warp issue, but I'm not exactly sure.
Any chance you could share your findings?
(BTW - I have not yet tried the official 1.5.2 RTSKit download)
#8
I tweaked alot in the rtsunit class to get pathfinding work and removed some independed client movement since the client didnt have the path information and wanted to move a other way than the server.
I also think about to stay with 1.3 for RTS since it runs much faster on my pc than the 1.5.2 Version. It's still getting very slow after many units so I guess I have to change some stuff in processtick or not execute it every tick.
05/01/2008 (10:12 am)
I didnt find the reason, but it's much better than my first implementation with the patch.I tweaked alot in the rtsunit class to get pathfinding work and removed some independed client movement since the client didnt have the path information and wanted to move a other way than the server.
I also think about to stay with 1.3 for RTS since it runs much faster on my pc than the 1.5.2 Version. It's still getting very slow after many units so I guess I have to change some stuff in processtick or not execute it every tick.
#9
Although In all our tests the results were the opposite (1.52 running better than 1.3 version), Stephen Zepp comented about the Lightning Kit as a probable responsable for the performance hit.
Could you Thomas, describe your configuration a bit?
It will be eventually useful if we can have this performance drop cases a bit more documented.
05/01/2008 (11:22 am)
This is the second time I read about cases where the 1.3 version runs faster than the 1.52. Although In all our tests the results were the opposite (1.52 running better than 1.3 version), Stephen Zepp comented about the Lightning Kit as a probable responsable for the performance hit.
Could you Thomas, describe your configuration a bit?
It will be eventually useful if we can have this performance drop cases a bit more documented.
#10
Same map , no units, same camera position:
1.3 ~ 270 fps
1.52 ~ 220 fps
other camera position:
1.3 ~ 130 fps
1.52 ~ 98 fps
While the 1.52 fps is also very good - it's a difference. I didnt test the frame rates, but running multiple clients on same pc also works better than with the new Version on my pc.
05/01/2008 (12:01 pm)
It's an Pentium 4 (3Ghz) with GeForce 6600GT (AGP) and 1GB Ram :Same map , no units, same camera position:
1.3 ~ 270 fps
1.52 ~ 220 fps
other camera position:
1.3 ~ 130 fps
1.52 ~ 98 fps
While the 1.52 fps is also very good - it's a difference. I didnt test the frame rates, but running multiple clients on same pc also works better than with the new Version on my pc.
#11
05/01/2008 (12:29 pm)
I just downloaded/installed the new offical 1.52 RTS Version - Units move smooth - great work novack :) I guess i'll never find out what the reason was.
#12
About the movement, its really weird how that behaviour appear and vanishes. It reminds me one problem Ive got time ago while fixing a bug, with the units warping instead of walking, but cant remember what it was -something obvious nevertheless-.
05/01/2008 (3:16 pm)
Oh good job on that configuration description, thanks. I guess I'll have to make some tests again, and see whats going on.About the movement, its really weird how that behaviour appear and vanishes. It reminds me one problem Ive got time ago while fixing a bug, with the units warping instead of walking, but cant remember what it was -something obvious nevertheless-.
Torque 3D Owner James