Is there a DOTS based solution for navigation?

Or will Navmesh work and scale well with thousands or millions of DOTS based entities navigating?

 

Or is specifically tight to CPU computing power, complexity of navmesh, size of navmesh.
And you know that. Your questions are just arbitrary.

And again, you should make, or bring some benchmarking, before posting such questions out of thin air.

Expecting that average Joe, without fancy algorithmic optimizations, should experience 10k to 100k agents moving. Assuming there is no additional logic on the game.

 

As far as I know Unity's making a DOTS version of navmesh/recast.

 

Making or planning? Anyway the roadmap just states its in the planning phase which would suggest we're still years out.

 

Making according to the author of it, I read a comment of his over a year ago. Can't remember if it was on these forums or over on github where the mono/component workflow version of it currently resides.

 

Was it this thread? My line of thinking is that the suite of roadmap pages for animation and urp are pretty on target and up to date, actually more regularly than I originally would've thought.

 

Think it was both but they did clearly state in the thread you linked:

"We continue working on aspects of DOTS NavMesh but we don't have news to share at this point" a year ago, so sounds roughly right to me.

 

That was a year ago, and things can change. From the roadmap I'm just speculating that perhaps development was stalled/restarted.

 

I remember reading this post somewhere.
However, seems the statement applies to this year as well.

 

Q: How much code is needed to make a DOTS navmesh solution and how much code can a Unity level developer write in month?

Is this something they could pass onto the community e.g. here's the Navmesh code go on DOTSify it for yourselves?

 

Sure they can. And it is actually dangerous, as people alelready did it, and Unity may stick Middle Finger to us, and never release proper DOTS navmesh, because of that.

The precision of the answer is a bit vague. How much code probably you'd Nedd to write, is something minimum in few thousands of lines. The thing is, there is multiple ways to do so. Even in DOTS.That depending how deep in rabbit hole you want to go. But for example, I wrote my node base path planning systems, probably between 2 and 3 weeks, with testing. But also, this was something new to me. I could probably reevaluate na mesh instead, during that time.

The time will grateley vary, based on devs expertise. If someone did navmesh in past, understanding the problem and what is needed, it will accelerate development.
But for someone new, it easily can exeed months.

Will dev use existing navmesh, to expand upon. Or make it from scratch.
Next challenge is, knowladge of mesh vertices and normals, how to handle them efficiently.
On top of that, need to know DOTS already.
And even then, you can spend further more time, trying figure out bes optimal approach, to get best performance.

Unity team, probably could pull out DOTS nav mesh in few short months, if they really want to, with their inhouse expertise. Even if rewriting totally, existing UnityEngine na mesh. But I will not comment further on that.

 

Any progress with DOTS to GPU as this could be amazing for larger spaces and LODding of navmesh for distant Units.