Hello all, are there any plans to improve performance in version 2.0? Acording to this topic: http://www.phoronix.com/forums/showthread.php?t=27934 , this game might be also very CPU limited. I've seen some discussion about TORCS not using VBO's cu improve performance, are planning to change that?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yes, because 2.0 will support (and use) 2 CPUs if present, one running the graphics engine, and the other one the physics engine an AI drivers. This gives up to a 2x FPS boost, but only in the cases where the frame rate experienced with only one CPU is low (say under 25-20 FPS).
No, because we won't rewrite this time our graphics engine, which is a bit obsolete now, is a heavy CPU consumer while also a bad GPU feeder (not as bad, though, as calling glDrawElements for each triangle ;-)
Brian already started this rewrite work (based on Open Scene Graph), and got promising performance, but is quite stuck about it right now.
BTW, with a decent 2.0 GHz CPU and a medium range graphic card, you easily get 40 or more FPS when racing against l10 AI opponents … so, unless you enjoy a brand new bionic eye-set, things are not that bad, aren't they ?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you, I have some obsolete computer Sempron 2800+ (Socket A, 2 GHz, 1 core) CPU and Radeon X850 XT video card, Linux OS and while it can play quite right Nexuiz, Sauerbraten, UT2004, VDrift games, is struggling with Speed-dreams, especially on tracks with plenty of objects. I admit Marek's post is a bit exaggerated :-) , but this game is far too heavy on the CPU side. On the bright side development speed and quality (art part) is far better than TORCS.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello all, are there any plans to improve performance in version 2.0? Acording to this topic: http://www.phoronix.com/forums/showthread.php?t=27934 , this game might be also very CPU limited. I've seen some discussion about TORCS not using VBO's cu improve performance, are planning to change that?
Yes and no.
Yes, because 2.0 will support (and use) 2 CPUs if present, one running the graphics engine, and the other one the physics engine an AI drivers. This gives up to a 2x FPS boost, but only in the cases where the frame rate experienced with only one CPU is low (say under 25-20 FPS).
No, because we won't rewrite this time our graphics engine, which is a bit obsolete now, is a heavy CPU consumer while also a bad GPU feeder (not as bad, though, as calling glDrawElements for each triangle ;-)
Brian already started this rewrite work (based on Open Scene Graph), and got promising performance, but is quite stuck about it right now.
BTW, with a decent 2.0 GHz CPU and a medium range graphic card, you easily get 40 or more FPS when racing against l10 AI opponents … so, unless you enjoy a brand new bionic eye-set, things are not that bad, aren't they ?
Thank you, I have some obsolete computer Sempron 2800+ (Socket A, 2 GHz, 1 core) CPU and Radeon X850 XT video card, Linux OS and while it can play quite right Nexuiz, Sauerbraten, UT2004, VDrift games, is struggling with Speed-dreams, especially on tracks with plenty of objects. I admit Marek's post is a bit exaggerated :-) , but this game is far too heavy on the CPU side. On the bright side development speed and quality (art part) is far better than TORCS.
OK, understand (I'm often optimistic about the "decent" word, sorry).
Did you try the hints listed here ?
http://sourceforge.net/apps/trac/speed-dreams/wiki/GraphicCardSettings
Some others :
- avoid using Simu V3, you'll save up to 30 % f/s,
- limit the number of AI opponents.
Thanks a lot for the "bright side" … nice to hear it :-)