Steam News29 July 202616d ago

Hot Fixes

A performance and presentation release. Two long-standing frame hitches are gone - one of them a freeze you felt every run - the game now runs noticeably smoother at high enemy counts, and a set of interface problems ar

In this update3

Full notes

Full Slot Theory update

Read the full published notes in a cleaner layout. The original post stays linked below.

What changed

2 fixes3 additions4 changes1 removal

UI and audio
Performance

fixedA performance and presentation release. Two long-standing frame hitches are gone - one of them a freeze you felt every run - the game now runs noticeably smoother at high enemy counts, and a set of interface problems are fixed, several of which only ever showed up on Android.

addedPerformanceImpact sparks no longer cripple the frame rate at high speed. Every projectile hit throws a small burst of sparks, and those bursts are recycled from a pool so the game does not have to build new ones constantly. The pool held 400. On a busy board at 5x or 20x speed the game was asking for far more than that at once - measured at over 5,700 live at the same moment - and every burst past the 400th was built from scratch and thrown away again on the same frame. The pool was doing the exact opposite of its job, at the worst possible time. Bursts are now capped at a number the game can actually sustain, and the pool is sized to match, so the recycling never breaks down. On the internal benchmark this cut the average frame time by 41% - from 11.2 ms to 6.7 ms - while the measured run was carrying * more * enemies than the one before it. Nothing looks different: normal play never came close to the new ceiling. Worth noting because it was a genuine surprise: this was expected to help only the worst frames. It improved the average by 41%, because the breakdown was happening on ordinary frames too, not just visibly bad ones.

fixedPerformanceFixed a freeze the first time Wildfire or impact sparks appeared in a run. Graphics work has to be compiled the first time the game actually draws it, and that compilation was landing on a live gameplay frame - the moment the first Wildfire trail or spark burst appeared on screen. The result was a 127 ms freeze mid-run, once per session, at whatever moment those effects happened to first proc. The game now compiles all of it off-screen during loading instead, before you reach the board. The worst frame in a session dropped from 127 ms to 56 ms , and the mid-run freeze is gone entirely. Loading is not longer in any way you will notice - the work was always being done, just at the wrong time.

changedPerformanceBuilds using Monomania or Crosslink Array are cheaper to run. Both cards check what towers you have placed, and that check was being recomputed from scratch on every single hit. It is now computed once and reused until you actually change your board. Purely internal - the damage numbers are identical.

changedPerformanceOverall: on a wave-20 crossfire run with over 550 enemies alive, 99% of frames now land under 13 ms, against about 18 ms before. Average frame time and worst-case frames both improved substantially.

addedNew settingShow FPS. A new toggle in Settings under DISPLAY, off by default. When on, a live frames-per-second readout appears in the bottom-right corner during a run. It stays out of the way of the operator badge, the surge meter and the wave readout, and it does not appear on menus.

worst frame in a session dropped127ms56ms

Slot Theory changes

addedImpact sparks no longer cripple the frame rate at high speed. Every projectile hit throws a small burst of sparks, and those bursts are recycled from a pool so the game does not have to build new ones constantly. The pool held 400. On a busy board at 5x or 20x speed the game was asking for far more than that at once - measured at over 5,700 live at the same moment - and every burst past the 400th was built from scratch and thrown away again on the same frame. The pool was doing the exact opposite of its job, at the worst possible time. Bursts are now capped at a number the game can actually sustain, and the pool is sized to match, so the recycling never breaks down. On the internal benchmark this cut the average frame time by 41% - from 11.2 ms to 6.7 ms - while the measured run was carrying * more * enemies than the one before it. Nothing looks different: normal play never came close to the new ceiling. Worth noting because it was a genuine surprise: this was expected to help only the worst frames. It improved the average by 41%, because the breakdown was happening on ordinary frames too, not just visibly bad ones.

fixedFixed a freeze the first time Wildfire or impact sparks appeared in a run. Graphics work has to be compiled the first time the game actually draws it, and that compilation was landing on a live gameplay frame - the moment the first Wildfire trail or spark burst appeared on screen. The result was a 127 ms freeze mid-run, once per session, at whatever moment those effects happened to first proc. The game now compiles all of it off-screen during loading instead, before you reach the board. The worst frame in a session dropped from 127 ms to 56 ms , and the mid-run freeze is gone entirely. Loading is not longer in any way you will notice - the work was always being done, just at the wrong time.

changedBuilds using Monomania or Crosslink Array are cheaper to run. Both cards check what towers you have placed, and that check was being recomputed from scratch on every single hit. It is now computed once and reused until you actually change your board. Purely internal - the damage numbers are identical.

changedOverall: on a wave-20 crossfire run with over 550 enemies alive, 99% of frames now land under 13 ms, against about 18 ms before. Average frame time and worst-case frames both improved substantially.

Performance

Impact sparks no longer cripple the frame rate at high speed. Every projectile hit throws a small burst of sparks, and those bursts are recycled from a pool so the game does not have to build new ones constantly. The pool held 400. On a busy board at 5x or 20x speed the game was asking for far more than that at once - measured at over 5,700 live at the same moment - and every burst past the 400th was built from scratch and thrown away again on the same frame. The pool was doing the exact opposite of its job, at the worst possible time.
Bursts are now capped at a number the game can actually sustain, and the pool is sized to match, so the recycling never breaks down. On the internal benchmark this cut the average frame time by 41% - from 11.2 ms to 6.7 ms - while the measured run was carrying * more * enemies than the one before it. Nothing looks different: normal play never came close to the new ceiling. Worth noting because it was a genuine surprise: this was expected to help only the worst frames. It improved the average by 41%, because the breakdown was happening on ordinary frames too, not just visibly bad ones.
Fixed a freeze the first time Wildfire or impact sparks appeared in a run. Graphics work has to be compiled the first time the game actually draws it, and that compilation was landing on a live gameplay frame - the moment the first Wildfire trail or spark burst appeared on screen. The result was a 127 ms freeze mid-run, once per session, at whatever moment those effects happened to first proc. The game now compiles all of it off-screen during loading instead, before you reach the board. The worst frame in a session dropped from 127 ms to 56 ms, and the mid-run freeze is gone entirely. Loading is not longer in any way you will notice - the work was always being done, just at the wrong time.
Builds using Monomania or Crosslink Array are cheaper to run. Both cards check what towers you have placed, and that check was being recomputed from scratch on every single hit. It is now computed once and reused until you actually change your board. Purely internal - the damage numbers are identical.
Overallon a wave-20 crossfire run with over 550 enemies alive, 99% of frames now land under 13 ms, against about 18 ms before. Average frame time and worst-case frames both improved substantially.

New setting

Show FPS. A new toggle in Settings under DISPLAY, off by default. When on, a live frames-per-second readout appears in the bottom-right corner during a run. It stays out of the way of the operator badge, the surge meter and the wave readout, and it does not appear on menus.

Interface

Sealed slots are visible on every platform. A locked hardpoint was drawn using a padlock emoji, which the game does not ship a font for. Windows quietly substituted one from the operating system; Android had none to offer, so a sealed slot looked exactly like an empty one. Locked slots are now drawn by the game itself, like every other battlefield marker, and read correctly everywhere.
All interface symbols now render identically on every platform. The underlying problem behind the missing padlock was broader than one icon: the game's main typeface carries no symbols at all - no arrows, checks, stars or chevrons - so 68 places in the interface were relying on whatever the host operating system happened to supply. The game now bundles its own symbol font, so every platform draws the same shapes.
Long run names are no longer cut short on the leaderboard. Expanding a leaderboard entry now shows the full generated build name across the full width of the row, instead of truncating it after 30 characters ("Patchwork Overclocked Voltg...").
The end-of-run screen is readable again. The results column was drawn on an almost fully transparent surface, so the live board and HUD showed straight through your stats. It now has a proper dark backing. The stat column is also wider, with the content evenly inset.

Source

Steam News / 29 July 2026

Open original post

Changelog.gg summarizes and formats this update. How we read updates.