Leadwerks 3.1 framerate low ?

[Roland]

One thing I noticed running the steam Leadwerks 3.1 is the low FPS speed.

 

I tested by creating a simple terrain 1024x1024 with 4 textures (diff+norm+disp), a infinitive light and a camera. With that I get 30-35 FPS on my GTX560. Doing the same thing in LE2.5 I got well over 600 FPS.

 

So my question is. Is Leadwerks 3.1 this incredibly slow or is there anything with my installation. What do you other guys get?.

[klepto2]

I get the same. first i thought it were my posteffects, but then i noticed the same without them. I think this behaviour needs to be researched and solved.

[Roland]

Yes indeed. Such low FPS rate with essential an empty scene is more or less a show stopper

[DudeAwesome]

jep I also noticed this. it has something to do with the infinite light and the terrain. you can increase the fps when the camera dont see anything of the terrain (like mountains blocking the area behind). if you have just a flat terrain it have to render everything. hard for flight simulators

[YouGroove]

@Roland: Can you post some level to download so we'll test with same scene ?

 

--------------

 

I had same problem with non terrain scenes , simple cubes with few very low poly characters.

Even without optimisation , making same scene on another 3D app i got double frame rate or more, i don't know if it is rendering or shaders ? or if it is some bottlenecks somewhere ?

I already made some suggestion for optimisation :

http://www.leadwerks.com/werkspace/topic/8589-models-lod-lights-per-object-shadow-qualitymip-mapping/

Let's hope optimisation techniques to appear in LE3 ?

[DudeAwesome]

I dont can try it right now but try to decrease the camera view range

[Josh]

One thing I noticed running the steam Leadwerks 3.1 is the low FPS speed.

 

I tested by creating a simple terrain 1024x1024 with 4 textures (diff+norm+disp), a infinitive light and a camera. With that I get 30-35 FPS on my GTX560. Doing the same thing in LE2.5 I got well over 600 FPS.

 

So my question is. Is Leadwerks 3.1 this incredibly slow or is there anything with my installation. What do you other guys get?.

If you are running in debug mode the error checks will make the program run much slower, so run the release executable (F6).

 

There is still performance optimization I can make in the current version. There's normally a cycle of new development, usage, then optimization as we get more and more real-world usage. The same was true in Leadwerks 2.x. Dave's "The Zone" scene was unplayable when I started, and I got the framerate up by just testing the scene, and that came back and benefited everything else being run.

[gamecreator]

You should test it in Leadwerks 3.

[Roland]

Thanks for the info Josh. I'm OK with that and I trust that you will get up the speed in the optimization phase.

Just wanted be sure that this issue is on the to-do list.

 

Now looking forward to the Leadwerks 3.1 C++ release.

[ChrisV]

Yeah, i have to agree on this too. LE 2.5 ran much faster than current LE version. In LE 2.5, you could run a scene with a large terrain, filled with trees and vegetation, realtime shadows, a bunch of animated characters and plenty of fullscreen shaders on and still have a good framerate. Current version seems to have problems with much less than that.

 

I also noticed that the editor gets slower and slower the longer it runs. It seems to be getting slower on each launch of the game (F6).

 

If you are running in debug mode the error checks will make the program run much slower, so run the release executable (F6).

I always run the game using F6, and i see no speed increase or difference, but i'm sure you'll be able to optimize LE as you did with previous version. And, by doing so, you'll make everyone happy. Thanks!

[cassius]

This needs fixing quickly so people can make progress with their game dev.You can't wait months for this one.

[gamecreator]

Josh is doing almost all of this himself so he has a lot on his plate. The highest priority is likely the fulfillment of the Kickstarter promises, which I think is understandable.

 

That said, if there was even one person who was close to releasing a game made with 3.1, I'm sure that optimization would shoot up on the list of priorities.

[cassius]

Yes. I appreciate the hard work Josh is doing, its great . It depends on what is considered a priority.

[YouGroove]

30 FPS in debug mode, on laptop that runs AAA games with good details and lightening.

As level grows it becomes unplayable.

 

 

We are far from some old game like first Far Cry in terms of 3D Scene and content here, this level is very basic and empty.

 

LE3 and OpenGL4 are too heavy asking in terms of GPU for such tiny demo , it would be necessary to limit shaders and shadows precision in distance, per object also, or like AAA games choose shaders/shadows/Lightening quality so different shaders precision.

 

Minimum optimisation , should be priority after Linux release, at least something to speed up somewhat.

[Marcus]

I agree with the performance optimizations. At this point, I don't think Leadwerks 3.1 could be usable for a serious project. With some major optimizations, it could serve nicely for a project.

[Marcus]

I have done some further testing and I'm not too happy about where the performance is at. I have about 10 CSG brushes and each brush has four textures. My framerate is VERY, VERY low. I replaced my mats with some of the developer mats and it doesn't change at all. I think that there are some serious issues at the core of the engine code. I hope this gets fixed in the very near future. Hard to start a project with the engine with such poor performance.

[DudeAwesome]

thats why I said leadwerks Indie Edition is some kind of Early Access but its sold as finished lua game engine. For me I want to buy the cpp Version but only if its worth. Would I buy the current version with cpp? nope

 

Site kickstarter is also important but i Never had bought leadwerks If i had known about all the Bad Performance/bugs stuff.

 

 

Its Early access and not finished

[YouGroove]

I think, this is more when people are trying making a real small game in LE3 , they will get quickly be stuck on performance.

We'll see in the future what systems have been implemented for performance ? But i agree core engine would need performance after Linux release, actually it's too bad for production , even very small game.

[puki]

This is not good news to hear. In fact, it's potentially devastating.

[Marcus]

This is not good news to hear. In fact, it's potentially devastating.

 

I'm sure Josh can make the necessary improvements. I'm not sure by what factor things can improve, but I will say the engine will have to be multiples more efficient to even do a small project.

[Naughty Alien]

I do not have Leadwerks 3, but im carefully observing forums, as I may look for it, but not at this moment. Having said that and regarding this topic, have you guys maybe pinpointed which part of your scene causing fps drop? I would really like to know.

[shadmar]

With just a camera and one lightsource, LE31=300 fps, LE25 700 fps, LE30=2000fps

[YouGroove]

I changed normal map/spec shaders to emissive simple shader, and no shadows even like that i still in 30 FPS or less. I think it's more LE3 core engine that is problematic.

------------------

I made another test elsewhere: lot of active physic balls with normal map shader and good shadows :167 FPS

Performance is huge using shadows on all objects and no optimisation, nothing LE3 can't compare with.

 

Great graphics and shaders are useless if your game runs slowly as a nightmare on good PC.

 

LE3 will need one more year to bring performance and lacking essential game features ( GUI, grass, better char controller/navmesh , open editor ? etc ... )

[Josh]

Writing an engine is much harder than designing one specific game because the engine needs to be able to accommodate lots of different games I don't have direct control over. There's normally a period of getting real-world usage which then can be used to test and optimize code. It's unrealistic to think that a newly developed renderer is going to run at the same speed 8 weeks after release as it will after 6 months of real-world usage and testing. I also don't know anything about people's setups and whether there is some simple misusage going on.

 

I'm an expert at OpenGL optimization, but it's not possible for me to do all things you would like at once. I don't like it when one group of users attempt to steer development to the detriment of another because they don't care about those features.

[YouGroove]

We won't make dev steer to any direction.

Perhaps i was the only one wanting to keep OpenGL2 support on PC, lightmapping, Lua multi script, simple shader and not animated ones, simple shadows without having to complicate workflow choosing a shader ?

 

--------------------

 

Indeed we are interested in CSG Tools, grass system, GUI, Blender support and others indeed ...

but actually, even using super low settings, and almost empty scene, this is too low frame rate.

So just need to state clearly : Grass, GUI, CSG tools priority 1 , peformance priority 2 , and we'll stop asking performance .

 

LE3 priority is perhaps eye candy stuff or features ask to make sell the engine , and that's ok : better to have some 3D engine sell well than engine not selling going in bad way.

All people that learns the engine, or just testing experimenting indeed are happy. other beginning a real game see the problem.

 

I like LE3 and Lua, i just keep happy and fun prototyping with it, this is very important factor

If LE3 editor and Lua wasn't fun and easy to use wouldn't be using it.

But for a real small game consideration i'll wait some more months indeed and see if performance begins to show up ?