Why is TGE 1.5 So slow
by Thanhda Tie · in Torque Game Engine · 12/23/2006 (8:50 pm) · 11 replies
I'm just curious why is TGE 1.5 soo much slower then TGE 1.4 or TLK 1.4. I mean besides increasing the terrain geometry. what makes it soo much different.
TGE 1.5 - 5~12fps
TGE 1.4 - 60~80fps
TLK 1.4 - 50~60fps
I havnt been actived with Torque in a while, but has there been any recent updates to this issue?
I was thinking of porting my game over to 1.5, but right now i dont see any reason why should i, besides the arcane fx stuff.
TGE 1.5 - 5~12fps
TGE 1.4 - 60~80fps
TLK 1.4 - 50~60fps
I havnt been actived with Torque in a while, but has there been any recent updates to this issue?
I was thinking of porting my game over to 1.5, but right now i dont see any reason why should i, besides the arcane fx stuff.
#2
Are you comparing the same mission ??
The demo mission in 1.5 has pretty lots of stuff in it, making it very slow.
I want to know the comparing fps numbers of 1.4 and 1.5 running the same mission too.
:)
Aun.
12/23/2006 (10:44 pm)
@ T Squared Are you comparing the same mission ??
The demo mission in 1.5 has pretty lots of stuff in it, making it very slow.
I want to know the comparing fps numbers of 1.4 and 1.5 running the same mission too.
:)
Aun.
#3
12/24/2006 (7:37 am)
I have realized that too. My average fps in TGE 1.5 is 40~50fps, the same i get running games like Half Life 2 and FEAR. I think there's something wrong with that.
#4
12/24/2006 (12:46 pm)
For some reason it is the other way with me TGE 1.5.0 is running alot better then TGE1.4.0 and TGE1.4.2.
#5
12/24/2006 (3:14 pm)
Another thing that you may want to try is updating both your video card and sound card drivers.
#6
12/24/2006 (3:23 pm)
For best results in tracking down the problem, run the profiler for a bit and tell us where the time is going. "It's slow" isn't nearly as helpful as "the such and such block is taking 50% of frametime."
#7
12/24/2006 (3:24 pm)
What is your compiler, VS Express does not come with optimization, which can result in a slow exe
#8
12/24/2006 (5:39 pm)
I took the TGE 1.5 demo and also placed the TLK 1.4 exe file directly inside the same map. I see a big difference in FrameRate.
#9
If you want a good side by side example, make your own mission- using your own art, dont do any fancy lighting or use anything 'extra' from either build. Then run them side by side.
What is did, for my 1.5/1.4 test mission was make some HIGH poly shapes (10,000 poly each 'ball' shape). Some of them i exported with transparent textures(2 sided), some with environment mapping, others with just simple ONE texture(all textures 512x512). I put 100 of them in a mission, in different configurations in diffrent corners. I then put in a SUN light, and used volumeLight for the only light type in the map (I expect there are no changes from 1.4 Volume light to the 1.5 volume light...)
My conclusion, as far as i can figure it, 1.5 IS slower on 'most' of my test systems. (I have 3 computers and a laptop, slowest is a p3 1000mhz with GF2MX video card) The better system run 1.5 just fine (some odd stuttering and slow slow loading, then a lot of stuttering when switching from views with NOTHING to heavy poly views), but the FPS is near equal to 1.4 for the most part but 1.4 'seems' far faster. 1.5 hardly runs on the p3 with the 1.4 build getting a steady ~20FPS.
I use the 1.4 TLK demo build and the 1.5 demo build to run the tests.
12/24/2006 (9:08 pm)
"T^2" that will not work. (as i recall reading from some other post) 1.5 use some different lighting types, that will be skipped under the 1.4 exe. If you want a good side by side example, make your own mission- using your own art, dont do any fancy lighting or use anything 'extra' from either build. Then run them side by side.
What is did, for my 1.5/1.4 test mission was make some HIGH poly shapes (10,000 poly each 'ball' shape). Some of them i exported with transparent textures(2 sided), some with environment mapping, others with just simple ONE texture(all textures 512x512). I put 100 of them in a mission, in different configurations in diffrent corners. I then put in a SUN light, and used volumeLight for the only light type in the map (I expect there are no changes from 1.4 Volume light to the 1.5 volume light...)
My conclusion, as far as i can figure it, 1.5 IS slower on 'most' of my test systems. (I have 3 computers and a laptop, slowest is a p3 1000mhz with GF2MX video card) The better system run 1.5 just fine (some odd stuttering and slow slow loading, then a lot of stuttering when switching from views with NOTHING to heavy poly views), but the FPS is near equal to 1.4 for the most part but 1.4 'seems' far faster. 1.5 hardly runs on the p3 with the 1.4 build getting a steady ~20FPS.
I use the 1.4 TLK demo build and the 1.5 demo build to run the tests.
#10
12/24/2006 (9:18 pm)
Incidentally, we have some bugfixes going out in the next TGE update that might enhance performance in some cases. You might try mucking with TimeManager::process() - that seems to have caused some performance problems.
#11
Following this example usually shows the versions are identical in performance (though some people have seen better performance in 1.5).
12/24/2006 (10:07 pm)
This thread explains how to test the performance between TGE 1.4 and 1.5, and why: Torque 1.5 performace hit?Following this example usually shows the versions are identical in performance (though some people have seen better performance in 1.5).
Torque 3D Owner Stephen Zepp
2) The demo by default has all lighting and effects turned on. You will want to fine tune what types of lighting, shadows, and other performance intensive things in the code are turned on and off for your game.