Why aren't the worms just non-ticking/deactivated unless there is a player owned entity or radar coverage over anywhere in their terrain (if idling and not attacking and in their terrain of course)?

Turning on the debug overlay shows almost a hundred demolishers in the dark unexplored area of the ...

Sadly no mods or commands here, it's a completely vanilla server with no admin access to /c (to keep achievements for those that like going for them).

And yeah, as demolishers are killed, more ones spawn further out, and they all load back in when the game is loaded (didn't it used to be just ...

Currently [2.0.20], SmokeWithTrigger is eating around 1.2ms when normal in the game, and spikes to over 3ms (over 5ms in a few cases) when someone decides to go attack a demolisher or more and persists until a minute or so after they're dead. That's enough to cause it to drop below 60UPS for some ...

Greets, finished a few hundred hours in two factorio servers from the steamdeck, was hoping I'd eventually be able to learn the controls well enough over the mouse/keyboard that I could disable the helper key overlay in the lower-left corner, however I somehow still can't keep remembering even those ...

may I interest you in the Discovery Tree mod?


You just did ! I was sarcastic in my previous comment and have actually 0 reserve for the research queue being always on. This mods sounds pretty interesting but i'm also an avid FNEI user when playing with large unknown mod and i think the latter ...

posila wrote: Sat Jan 15, 2022 10:36 am I measured the save in 1.1.51 and it is pretty much the same as in 1.1.50, so that's not bad

Indeed a very good sign considering the worry mentioned earlier, guess the CPU is able to prefetch that new change well or it gets optimized out or something. ^.^;

Since there are 4 unique entity types the logic does not know what type of entity it might get when iterating the chunk lists and since we need to keep addition and removal of an entity from these lists O(1) they can't be pooled into any kind of memory pool.


This is precisely the kind of issue ...

As requested by posila at viewtopic.php?p=559889#p559889 posting this save to look at Surface::listFriendlyEntities performance issues:
https://overminddl1.com/Factorio/perf/OverSE-E1-22.zip

which mtune, O-level, or other CXXFLAGS are you using?


Purely just -O3 for this test, nothing fancy.



...

I can't download https://overminddl1.com/Factorio/perf/perf-factorio.data.xz, I get 403 Forbidden.


Oh bugger all, didn't have xz files in the allow list, fixed.


But, save would ...

movd + cvtdq2pd + mulsd

Interestingly GCC 11.2 compiles it to:

pxor xmm0, xmm0
cvtsi2sd xmm0, edi
mulsd xmm0, QWORD PTR .LC0[rip]
.LC0:
.long 0
.long 1064304640


And clang 13 compiles it to:

.LCPI0_0:
.quad 0x3f70000000000000 # double 0.00390625
cvtsi2sd xmm0, edi
mulsd xmm0, qword ...

On Windows (so compiled by MSVC), it ends up being movd + cvtdq2pd + mulsd and is inlined everywhere I looked. Which makes me think your profiler is incorrectly attributing instructions to getDouble() as it can't figure out what belongs to what after inlining and link time optimizations.


Even ...

Greets, been having some UPS issues in a bug-heavy game (with mods like rampant, but script times are remaining sub-1.2ms consistently), the in-game timings show that the Entity Update times are (depending on what the bugs are doing) between 15 to 35ms. I took a profile of the game while it was ...

Found the precise frame where it crashed, so yeah I do think it was what was just fixed in 1.1.30. Sorry for the noise.

Oh actually, by random chance I did happen to be streaming it at the time so some friends could watch, my mouse cursor went over the icon for an entity that showed ammo in the tooltip, so this might be related!

So this may have been what was *just* fixed in 1.1.30, I'm updating to 1.1.30 and will ...

EDIT: It looks like it was the ammo crash of 1.1.29 that was fixed in 1.1.30 shortly after, so this issue may be resolved.

Factorio 1.1.29 (1.1.30 came out while in-game, updating to 1.1.30 now though, however the below log doesn't seem to be in the changelog of fixes for 1.1.30, thus still ...

I have a (large) number of mods on my server where some are creating a lot of garbage memory instead of using pools or so as they should be doing. This is causing the LUA Incremental GC time to take almost *4* times the amount of time that LUA script processing itself is taking. Right now the LUA GC ...

Vanilla factorio loads almost instantly, it's mods that slow it down. Something I've noticed and mentioned somewhere before is that I did testing and found that for each additional mod added, even tiny QoL mods that don't adjust prototypes, increases the time for every mod to load. An example with ...

For documentation sake it happened again, same stack trace (though different symbol value, likely because factorio has been updated, 1.1.11 now):


Factorio crashed. Generating symbolized stacktrace, please wait ...
Raw stacktrace: 0xd08b27, 0xf05b3d, 0xd5f425, 0xd5fab1, 0xd5fb69, 0x41950, 0x93d15 ...

I thought that might be the case but the similar error on two computer that are separated by years between when they were built made me wonder... I'll try to update if it happens again and I catch it.

And it loaded on the next attempt for note.

It's worth noting that I've gotten this same (or at least very similar) error in the same place a week ago on another computer but as it wasn't my main computer and it wasn't on the internet at the time so I didn't report it. It was the same set of mods ...