Dev Update: Work in Progress

It’s been a few weeks since we provided you with an update on what the team is up to and how the game is progressing. As we mentioned in our last post (linked below), we realized that overall game performance was not up to our standards. Since then, we’ve focused the majority of our development on improving the performance and stability of the game experience. We still have a lot of work to do in that area but we’re also moving with urgency and focus.

[dynamiclink href="https://store.steampowered.com/news/app/2550530/view/538883784384185203?l=english"]

Here are a few highlights of our progress.

I realize some of this might be too technical for some of you and not technical enough for others. The high-level goals here are getting a higher player count per server and having the play experience be as smooth as possible.

• We optimized our database persistence layer, reducing allocation counts by 95% and saving 20GB of memory for playtest servers.

• Refined our automation system to cover more player behaviors and generate server performance reports nightly.

• Addressed performance spikes across a number of systems, including land claims, Outpost/Zone/Stronghold discovery, harvestables (trees and rocks), and animal spawning. We've created an automated system to stress test these more accurately, and will address more issues as server performance reports roll in.

• Dramatically cut down the memory cost and performance overhead for AI pathdata generation.

• Dramatically cut down memory cost for game simulation.

• Cut collision data in half for Strongholds.

• Compressed animation data by a factor of 6.

• Optimized animation graph loading.

• Reduced voxel navigation data by a factor of 6.

• Optimized ranch building operations.

• Server browser foundation replaced to make it extensible for better UI/UX.

• Per server login queue reliability improved.

• Character create flow improved, eliminating some fringe bugs that could prevent you from playing on a specific server.

• Reduced texture memory usage on the client by 40%.

For those of you less familiar with game development, this is just one small example from above of the work that’s involved. In this particular case, you can see how we approached collision optimization.

“We modified Unreal so we could simplify the collision on this forge from 22,586 triangles to 14 simple shapes while retaining three different collision materials: stone, brick, and metal.” - Erin Catto

Progress has been great. But we still have a long way to go. As of the writing of this message, we’re currently working on these things as well:

• Rewriting how items/storage works (again) to drastically reduce memory, save file size, and compute cost for large storage operations.

• Making significant CPU optimizations to how we track moving entities.

• More game sim memory cost & perf spike optimizations.

• More memory and performance optimizations for AI Pathdata.

• Working on reducing navigation data size for Strongholds.

Performance and optimization aren’t one-time tasks that you do and are done forever. They are constant responsibilities and every new feature and all new content requires us to revisit how the game is running.

But let’s talk about some fun stuff too.

Even though the next public test is mainly focused on improving how the game runs, there is a lot of new stuff the team is working on as well. This won’t be the most content or feature-heavy patch we do. But we thought it would be fun to share some of the progress so far.

• We have two new “medium” sized enemy Strongholds that should make the next public test. As a reminder, “medium” size is the equivalent of McCarthy, Observatory, Hacienda, Yellow Pine etc. One of