Lorca Linden provides data on pathfinding and simulator performance

It is fair to say that pathfinding has become one of the most controversial subjects in Second Life. While it has been the subject of a range of issues and problems, noticeably with vehicles, both before and after being deployed to the main grid, it has also become the most pointed-to bug-a-boo when people either are seeing, or believe they are seeing, simulator issues and problems.

Because of the levels of concern raised over pathfinding and its potential impact on simulator performance, Linden Lab has been carrying out comparative studies of simulator statistics recorded both before and after the pathfinding deployment. Speaking after the showing of a Designing Worlds special on pathfinding on Monday October 8th, Lorca Linden, the Pathfinding Producer at Linden Lab, gave the following high-level results of these comparisons:

Private Island simulator average sim fps:

  • Before pathfinding (Saturday, June 23, 2012): 44.43
  • After pathfiinding:
    • Dynamic pathfinding NOT enabled: 44.41
    • Dynamic pathfinding enabled, NO pathfinding objects: 44.29
    • Dynamic pathfinding enabled, at least 1 pathfinding object: 44.25
    • Dynamic pathfinding enabled, at least 10 pathfinding objects: 44.70

Mainland simulator average sim fps:

  • Before pathfinding (Saturday, June 23, 2012): 44.66
  • Dynamic pathfinding NOT enabled: 44.46
  • Dynamic pathfinding enabled, NO pathfinding objects: 44.44
  • Dynamic pathfinding enabled, at least 10 pathfinding objects: 44.79

These figures should not to be taken to mean there are no issues with pathfinding in terms of raised JIRAs relating to vehicles, etc.  However, as high-level as they are, and allowing for the size of the sample taken, they potentially show that pathfinding is not having as heavy an impact on simulators as many fear is the case. Commenting on them, Lorca said, “So in short, while there are certainly cases in which it’s possible for PF to have some performance impact, the data is showing that in the great majority of cases PF is not causing performance harm, since the grid-wide averages are within 1% pre and post PF.”

Also discussing the issue of perceived simulator performance during the Designing Worlds show itself, Falcon Linden acknowledged that part of concerns about simulator performance have been due to the way in which the Lab presented the potential for possible impact. “We didn’t communicate early on about the optimal performance of pathfinding,” he said. “We really wanted to take a conservative approach, so our communications, I think, were almost negative, in a way, where we were telling people what the worst case was, like we were making it seem that was what we expected to happen, but it wasn’t; and so people read from that that things could get bad.”

Lorca Linden (centre left) is joined by Maestro and Falcon Linden and Sandry Logan on a  Designing Worlds pathfinding special shown at the Designing Worlds studio on Monday October 8th (image courtesy of Designing Worlds / Wildstar Beaumont)

Also in the Designing Worlds programme, Lorca and Falcon, together with Maestro Linden, discuss Linden Lab’s thinking on pathfinding, why the Lab felt it to be a valuable resource to have in SL, and explain some of the additional features within it, such as the ability for pathfinding characters to navigate to a specific point in a region – or a point several regions away;  and a function which can be used both with pathfinding and independently of it (for example, it could be used within a HUD which can guide avatars to a specific store within a mall).

Overall, the programme helps to provide further insight into how pathfinding works and how it can be used, with a very practical demonstration by Sandry Logan of the Virtual Kennel Club. As such, for anyone who is curious / worried / may have a use for pathfinding, it is a recommended watch. Catch it on Designing Worlds at Tweet TV.

Related Links


6 thoughts on “Lorca Linden provides data on pathfinding and simulator performance

  1. Lets face it, the problem with path finding is that regions need to be opttimized to use it as LL intend or to not wram performance!
    Now i ask, how many Sim users know hwo to optimize a region as per LL faq?
    Our regions on mainland are all optimized and that mean:
    Not having any prim named object!
    Not having any sculp prim as rout, but instead making sure that any objexct or set of linked prims (be sculpted, mesh or whatever) are using a reg square prim as the rout!


    1. So, according to the logic we must “face”, people don’t know how to optimize the sims. So all those non-optimized sims are measurably running within a 1% performance difference. That seems like a pretty good result.
      Or are you saying that Lorca has deliberately only measured specific, known-to-be-optimized sims to skew the data?


    2. Foneco, your information is incorrect.

      It is not a requirement to rename objects / linksets with the default prim name of “object” (if you have any on your land), it is a recommendation (and actually a very good recommendation for anyone who builds, regardless of whether they are using pathfinding or not).

      Linksets do not have to have a cube prim as the root prim when setting pathfinding attributes. Mesh and sculpts can have attributes set against them equally well.

      Finally, there is a console command for region owners / estate managers to make sim-wide “first pass” optimisation of objects in a sim. It’s not perfect, but it helps get started.


Have any thoughts?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.