Search found 66 matches

by raven
03 Dec 2016, 12:20
Forum: Support
Topic: 3-8 fps in Sadrith Mora. Physics out the roof
Replies: 34
Views: 14526

Re: 3-8 fps in Sadrith Mora. Physics out the roof

I did a bit of logging just to see how bad it looks. The worst case I've seen are 26 rays (all 8 solver iterations exhausted) per simulation step! This totally kills the performance. The step length in those cases doesn't change much, which hints at the solver being stuck: s0 5.77191 s1 2.44885 s2 2...
by raven
01 Dec 2016, 11:25
Forum: Support
Topic: 3-8 fps in Sadrith Mora. Physics out the roof
Replies: 34
Views: 14526

Re: 3-8 fps in Sadrith Mora. Physics out the roof

I've been (very) slowly digging through the code. I think the major issue is the very generous use of ray/convex casts. They really should be used only when necessary, as they have to traverse the whole scene every time, don't benefit from time coherence (small changes from one step to the next). It...
by raven
25 Nov 2016, 10:23
Forum: Feature Requests and Suggestions
Topic: Feature request: other image projections
Replies: 2
Views: 1759

Re: Feature request: other image projections

I think your specific use case (multiple screens) should be easy to support, by enabling the functionality in osg and adding options to set up the additional screens. Each screen would still be using rectilinear projection. But with a reduced screen fov it should be good enough.
by raven
01 Nov 2016, 09:26
Forum: General
Topic: Extremely low performance
Replies: 9
Views: 5257

Re: Extremely low performance

I also get 30-80 exterior fps on linux with a GTX 960, with water shader off. Usually over 100fps indoors. I'm curious what current specific performance bottlenecks are. I'm still reading about how OSG works, trying to understand the OMW codebase. But considering draw performance tanks when viewing...
by raven
18 Oct 2016, 08:56
Forum: Feature Requests and Suggestions
Topic: Make Fortify Health safe to use
Replies: 5
Views: 2756

Re: Make Fortify Health safe to use

Actually the intermediate calculations can be skipped altogether.

When removing fortify health effect just scale current hp by:
hp = hp * (1 - fortify / hpmax)

For the first example:
hp = 100 * (1 - 100 / 200) = 50

Second example:
hp = 150 * (1 - 25 / 150) = 125
by raven
18 Oct 2016, 08:08
Forum: Feature Requests and Suggestions
Topic: Make Fortify Health safe to use
Replies: 5
Views: 2756

Re: Make Fortify Health safe to use

Or it could be converted (internally) to a scale factor which should result in consistent/predictable mechanics. potion/spell/on-use enchant: hp = 100 fortify health 100 -> health factor = hp + fh/hp = 2 hp = 200 -> 100 * 2 take 100 damage hp = 100 -> 50 * 2 fortify health effect expires -> health f...
by raven
04 Oct 2016, 17:38
Forum: General Development
Topic: AMD OpenGL performance
Replies: 4
Views: 2587

Re: AMD OpenGL performance

scrawl wrote:You can set environment variables within gdebugger in the project's "Debug settings". These may override environment variables you have set in the OS.
That was the first option I've tried.
You'd need to recompile the engine.
I was afraid I'd have :)
by raven
01 Oct 2016, 23:10
Forum: General Development
Topic: AMD OpenGL performance
Replies: 4
Views: 2587

Re: AMD OpenGL performance

OSG_GL_ERROR_CHECKING=OFF doesn't seem to work in conjunction with gDEBugger, might be a gDEBugger issue though. Would be nice if the debugger allowed to skip certain call types, but oh well... Is there eventually a simple way to disable lighting setup from the engine side, just to see how it perfor...
by raven
01 Oct 2016, 16:35
Forum: General Development
Topic: AMD OpenGL performance
Replies: 4
Views: 2587

AMD OpenGL performance

There have been quite a few posts about low frame rates with AMD cards. I've got one around and thought to have a look. In the start region I am getting 45-75 fps on my system (Win7x64). Running the game in gDEBugger I see about 55fps. Looking at call counts, there are a bit more than 1000 draw call...
by raven
24 Sep 2016, 20:53
Forum: Content Development
Topic: Workaround for Bug #806
Replies: 6
Views: 4031

Re: Workaround for Bug #806

From my understanding bullet collision mesh code uses triangle vertex positions only, can not differentiate between inside and outside. One would have to implement an extra variant (that uses vertex normals?) to handle one sided collisions.