avc15

  • Content Count

    935
  • Joined

  • Last visited

Community Reputation

493 Excellent

1 Follower

About avc15

  • Rank
    Senior Member
...

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

Enable
  1. Conjecture, but: Spaced Out wasn't designed for classic starts. The big asteroids were enough to bring any home PC to its knees on their own; adding a full list of moonlets to them was begging to cause this problem again. You can make the game unplayable in a moonlet cluster, too, but it takes a little more work.
  2. Here I am criticizing klei for doing a poor job factoring the workload between contexts. And also objecting to false things being said. I can do both. Process monitor dump from a mid-game save where the main thread is only 30% of the game's workload: This CPU figure displayed here is a thing called "moving load average", not a percentage. The game's performance can become ... unplayable... on an old save. Especially a classic start when you've also established multiple colonies. Given the nature of this game, that main thread should be around 1.5 load average if my different colonies were running in different threads. I was disappointed Klei didn't go that far. It would be plausible to run simulations for each asteroid in different contexts. Not sure why Klei didn't do that; there may be some difficulty because of unity. (I have never worked in unity) But, false is false. Be truthful when making these criticisms. This game runs in a self-scaling number of contexts, though half of them are unity garbage collection and only a third are doing meaningful work. Otherwise ONI might run a lot worse than it does now. The multithreading update was a huge boost in game performance; you maybe have to have been there. My save here is in over 100 threads and the OS is doing its job load sharing. The relevant note: Conduit network recalculation and updating is now multithreaded For more details, the test branch updates leading immediately up to this main launch update elaborate a little bit.
  3. I'm not arguing with that, I'm countering some incorrect things that keep being stated. "the game runs all on one core" - "nobody ever said the game runs all on one core"
  4. Yes, the game simulation is probably run on one thread. At the very least the liquid pipes and gas pipes layers are run in their own threads. Klei made the same enhancements in the conveyor rails layer but ended up rolling that change back (it was a buggy mess for the two testing branch updates with conveyors in a different context) Again, you can find this information in the patch notes, some of the last few updates for the pre-DLC game. And you can also see for yourself using a load monitor like process explorer, which will show that far more than you think is split off into a lot of different contexts. Some of these things come from unity, some of them are added by klei. last edit: context. it's not enough to handle a late save with 60 dupes and 6 colonies.
  5. Except this is a rather common claim on these forums, even sometimes repeated by veterans from early early access. Including the post I replied to here.
  6. I take issue with some of these things. The entire game doesn't run in a single core, I can prove otherwise. But at the very least the multi threading isn't rigorous. It depends on your save but sometimes the "main" thread shrinks to below 50% of the total process load. Now. Certain layers run in their own threads, but not per map or per loop as you have pointed out, rather, the whole layer (probably across all asteroids). To find the patch notes naming these changes I'd have to go dig them up (again; it's non trivial to search the patch history). Also, if you wanted to split the simulation (main layer) up into separate simulations (one per habitable moonlet), you could actually do that. It doesn't matter that the map is one map (another thing that gets stated on this topic frequently is that this it can't be, because there's only one map). It does matter that there is a boundary where the different simulations can only interact through specific interfaces (eg divided by neutronium, can only cross the boundary with rockets, supply teleporters, etc). Because such a boundary exists, each layer could be run in separate threads per moonlet including the main simulation. Klei opted not to do that; they probably had to stop somewhere, and the threading they did was a substantial improvement at the time. Anyway, I have to stop somewhere, too. I'll get off here for now.
  7. what you will notice at very late progression - framerate goes to nothing but cpu utilization also goes down. It's odd.
  8. In this savefile Nikola is in a super-productive state. Just watch him, he completes everything instantly. May have to change priorities to keep him on task. The Rosy Crew.sav
  9. In real life shielding is measured in tenth thicknesses. For each tenth thickness, multiply the radiation level by 0.1. He said it used to be 'linear' meaning each tile reduced a fixed amount. Maybe now each tile reduced by a percentage which would be 'geometric', that's how it works real-world as well.
  10. I just noticed this in my new save, dupes will not bring rot piles to my compost heaps. I can't mark the rot piles for compost, Setting a high priority sweep gives a red flag "storage unavailable" Savefile inc. Spacestation.sav
  11. I feel the same about it, but some very vocal community members have been giving klei a hard time about it. So it's important to someone at least.
  12. Optimistic Laboratory Cycle 2042.sav Interesting that we do not have similar issues with atmo suits or dupes in general. They always find an alternate path, unless they're wearing a jet suit.
  13. @Ipsquiggle thanks! you'll notice that in the specific gif I posted, Ren was not incapacitated. If you reload a couple times and let it run a few cycles you might see my specific outcome too.
  14. This was fixed for some time. Lots of old bugs cropping up from the past... Hassan, tune up that generator. Hassan. Hassan!