How to Use the Performance Guide
Apply quick wins, then tune graphics and base layout using the data-backed heavy hitter lists.
Performance Optimization
All recommendations on this page are derived directly from Star Rupture game files (entity configs, building DataAssets, LOD parameters). Settings are matched to the engine's actual LOD thresholds and simulation cost values — not estimated.
Apply quick wins, then tune graphics and base layout using the data-backed heavy hitter lists.
Engine LOD system
Source: DA_GenericBuilding — BaseLODDistance values. The Mass Entity system lowers update frequency at each threshold; large bases keep more buildings in LOD0 simultaneously, raising CPU load.
Max replicated buildings per viewer: 2,500. Buffer hysteresis: 20% (prevents LOD pop when moving near thresholds).
Simulation cost — extracted from game files
High electricity draw and stability cost = more simulation work per tick. Cluster these buildings so the LOD system can cull them together when the camera moves away.
Power draw (top consumers)
Stability load (top 7)
Largest footprint
Full complexity reference
Electricity draw and stability cost extracted from building DataAsset files. Higher values = more CPU work per entity tick.
| Building | Electricity Draw | Stability Cost | Footprint | Notes |
|---|---|---|---|---|
| Teleporter | 80 MW | 90 | 5x6 | Largest interaction radius (800 units). Active teleport triggers full entity re-registration. |
| Molecular Binder | 80 MW | 90 | 7x7 | Heaviest single footprint among high-power consumers. |
| Constructorizer v2 | 80 MW | 90 | 7x7 | Identical load profile to Molecular Binder. |
| Facturer | 80 MW | 90 | 6x6 | Advanced manufacturing; high crafting-loop overhead. |
| Constructorizer | 80 MW | 90 | 6x3 | Same electricity draw as v2 but smaller footprint. |
| Oil Extractor | 80 MW | 0 | varies | Continuous fluid extraction; keep away from dense building clusters. |
| Assembler | 60 MW | 90 | 6x3 | 8.3 s crafting cycle. Input/output socket count adds conveyor overhead. |
| Compounder | 60 MW | 90 | 6x6 | Same load tier as Assembler. |
| Compounder T2 | 60 MW | 90 | 6x6 | Identical profile to Compounder. |
| Pyro Forge | 60 MW | 90 | 6x6 | High thermal simulation overhead. |
| Defense Tower | 50 MW | 70 | 3x3 | AI targeting runs every tick during waves. Bunching multiple towers spikes CPU. |
| Cargo Dispatcher | 40 MW | 80 | 4x4 | Rail logistics scheduling adds entity pathfinding load. |
| Cargo Receiver | 40 MW | 80 | 4x4 | Paired with Dispatcher; each active route doubles pathfind cost. |
| Refinery | 40 MW | 90 | 6x6 | 4 s crafting cycle (fastest). High throughput = more item-spawn events per minute. |
| Mega Press | 30 MW | 90 | 4x5 | Heavy manufacturing; combine with other 90-stab buildings sparingly. |
| Furnace | 20 MW | 80 | 4x3 | Moderate load; safe to stack multiple if power budget allows. |
| Fabricator | 10 MW | 70 | 3x3 | Light consumer; good building block for compact bases. |
| Chemical Processor | 10 MW | 40 | 3x3 | Low stability cost makes it one of the friendliest processors. |
| Smelter | 5 MW | 80 | 4x3 | Low power but high stability; early-game bottleneck. |
Data source: DA_GenericBuilding, DA_GenericReplicatedBuilding, DA_GenericMegaBuilding, DA_Factory — Hotfix 0.1.2
Graphics tuning
Each preset targets a specific bottleneck. Apply one setting at a time and observe frame pacing before making further changes.
Recommended Start
Stable 60+ FPS while maintaining visual readability across all biomes.
In-game V-Sync introduces double-buffer latency. Apply a frame cap at driver level (NVIDIA Control Panel / AMD Radeon Software) or via Rivatuner Statistics Server instead.
Why: External caps produce consistent 16.6 ms frame pacing without the input-lag penalty of in-game V-Sync.
TAA eliminates shimmering on conveyor belts, power lines, and building edges that FXAA cannot resolve at typical base density.
Why: TAA adds ~3 FPS cost on mid-range hardware while providing the best temporal stability — important when cameras pan across large bases.
High shadow cascades cost disproportionate GPU time with minimal perceptible improvement. Medium preserves directional depth cues.
Why: Shadow quality is the single largest FPS variable in outdoor/base scenes. Dropping to Medium typically recovers 12–18 FPS on mid-range cards.
Prevents building pop-in when panning across large bases. Matches the game's LOD2 threshold of 6 000 units (extracted from DA_GenericBuilding).
Why: Medium view distance triggers visible pop-in before the LOD2 buffer activates. High eliminates this at modest FPS cost (~4 FPS).
Rupture storm particles and acid-spitter VFX are the primary overdraw sources during combat. Medium keeps effects readable without frame spikes.
Why: Ultra effects cause overdraw spikes during multi-enemy encounters. Medium stays above 60 FPS during raids of 30+ enemies.
Performance Priority
Maximize frame rate on low-to-mid GPUs or when running very large bases.
Renders the 3D scene below native then upscales. HUD and UI remain at native resolution and stay fully sharp.
Why: The game's internal upscaler preserves base readability well. 85% scale consistently recovers 15–25 FPS on entry-level cards.
Contact shadows are preserved at Low, maintaining minimal depth perception. Dynamic shadow maps are disabled.
Why: Low shadows is the single most impactful change for entry-level hardware — typically +20–30 FPS versus High.
Dense foliage biomes (Forest, Swamp) generate a high number of draw calls. Low reduces visible foliage instances significantly.
Why: Foliage density affects CPU-side draw call count, not just GPU. Low setting reduces frame-time variance on CPU-limited systems.
Particle-heavy combat (explosions, acid, storm debris) causes GPU overdraw spikes. Low keeps particle count below the fill-rate limit.
Why: During 30+ enemy swarms, Low effects avoids the frame dips that occur when particle density exceeds GPU fill-rate capacity.
Motion blur is a post-process pass that costs fill-rate with no gameplay benefit.
Why: Disabling it is free performance. Most strategy/base-builder players disable it regardless of hardware tier.
CPU / Simulation Heavy
Reduce frame-time spikes caused by entity simulation overhead in large bases (200+ buildings).
Reduces the number of buildings ticking at full update frequency simultaneously. Critical once base entity count exceeds ~150.
Why: The game's Mass Entity system (source: DA_GenericReplicatedBuilding) lowers update frequency to 1 s at Medium range, reducing simulation CPU load per frame.
Screen-space reflections recalculate every frame on all wet/metallic surfaces. Cost rises with base size.
Why: During rain or storm events, SSR is re-evaluated for hundreds of building surfaces simultaneously — disabling it removes a variable-cost GPU pass.
SSAO samples scene geometry on every frame. With dense building clusters the sample count scales with polygon complexity.
Why: AO contributes ~5–7 FPS cost in complex base geometry. The visual difference in a top-down factory view is minimal.
Prevents VRAM exhaustion when panning across large bases with many distinct building materials.
Why: Without streaming, loading all building textures simultaneously can exhaust VRAM and cause hitching on cards with 6 GB or less.
Cap slightly below monitor refresh rate (58 for 60 Hz, 117 for 120 Hz) for consistent frame pacing.
Why: Uncapped frames produce variable frame-time intervals that manifest as micro-stutter even at high average FPS. A sub-refresh cap eliminates this.
Base layout
Derived from LOD thresholds and entity update costs in game files. Apply these before reducing render settings.
Buildings with stabilityCost 90 (Teleporter, Molecular Binder, Assembler, Refinery, Facturer, Compounder, Pyro Forge) each contribute full simulation load. Placing them within a 300-unit cluster allows the LOD system to cull them as a single group when the camera moves away, reducing active entity count.
Data: Stability values from buildings.json — extracted from game files
The game's LOD2 boundary activates at 6 000 units from camera (source: DA_GenericBuilding). A base spread beyond that radius forces the engine to maintain multiple LOD zones simultaneously, increasing the active entity count and per-frame update budget.
Data: LOD distances from DA_GenericBuilding (BaseLODDistance)
Each rail and conveyor segment is a separate physics entity that ticks every frame. The game's entity system (MassEntityConfigAsset) processes each segment individually regardless of load state.
Data: Mass entity trait structure from DA_GenericReplicatedBuilding
Each container with item contents maintains an inventory-change listener in the simulation. Storage Depot v.1 (2x2) and v.2 (3x3) both draw 5 MW each; Expandable Storage draws 60 MW. Consolidating into fewer containers reduces both power draw and per-frame listener overhead.
Data: Electricity values from buildings.json
The temperature simulation (CrMassTemperatureTrait) runs on every tick for each building. Processors with high CoolingCapacityUsing (Assembler: 150, Refinery: 60, Compounder: 60) require active cooling or the simulation continues evaluating overheat states each tick even when idle.
Data: CoolingCapacityUsing values from DA_Assembler, DA_Refinery
Pressure Power Generator (1 000 MW, 0 stability cost) and Solar Generator v.2 (1 000 MW, 70 stability cost) produce maximum power with the lowest per-unit simulation overhead. Wind Turbine v.1/v.2 have constant rotating-animation ticks that add GPU overhead near the camera.
Data: Electricity output and stabilityCost from buildings.json
System tweaks
All options are reversible. Test one change at a time.
Star Rupture uses Unreal Engine's Mass Entity System which is highly multi-threaded. Adding -USEALLAVAILABLECORES to Steam launch options ensures the thread pool spans all logical CPU cores.
Apply a frame cap at the driver level (NVIDIA Control Panel: 'Max Frame Rate', AMD Radeon Software: 'Frame Rate Target Control') or via Rivatuner Statistics Server. In-game V-Sync uses double-buffering which adds at least one frame of latency.
After game updates, allow the shader cache to fully rebuild before benchmarking. The initial session after a patch will stutter as shaders compile on demand. Clearing the cache prematurely resets this process.
Overlay software (Discord, GeForce Experience, Xbox Game Bar) hooks into the render pipeline and can add 2–8 FPS overhead. Monitoring overlays with lightweight rendering (e.g., hardware-only stat overlays) have minimal impact.
The game streams large base environments and uses texture streaming extensively. Mechanical HDD seek times cause loading hitches when the camera pans across areas outside the preloaded zone. SSD reduces hitching; NVMe eliminates it for typical base sizes.
Common misconceptions
These are frequently repeated tips that contradict actual engine data.
More storage buildings always improve logistics
Each storage container with items holds an active inventory listener in the simulation. More containers with small item counts increases simulation overhead compared to fewer fully-loaded containers. Consolidate storage for better frame times.
Source: Inventory size values from buildings.json; MassEntityConfigAsset trait structure
Raising view distance always improves visibility without cost
Higher view distance forces buildings beyond LOD2 (6 000 units) to run at LOD1 (3 000-unit) update frequency, increasing the number of fully-simulated entities per frame. This has the largest impact on CPU-limited systems with large bases.
Source: DA_GenericBuilding LODParams: BaseLODDistance[2] = 6 000
Wind Turbines are the best power source because of high output
Wind Turbine v.1 and v.2 have continuous animation ticks (AnimRotSpeed: 3.0, AnimDuration: 5.0 from DA_WindPowerGenerator) that add GPU vertex processing overhead. Pressure Power Generator and Solar Generator v.2 deliver equal or greater output with lower per-frame GPU cost.
Source: Electricity values and animation parameters from DA_WindPowerGenerator and buildings.json
The Teleporter only costs FPS when actively used
The Teleporter (80 MW, stabilityCost 90, interaction radius 800 units) registers a persistent interaction zone that the entity system scans every tick, even when idle. One Teleporter adds to simulation overhead continuously.
Source: DA_Teleporter: Radius = 800.0, electricity.value = 80 from buildings.json