Introduction
Why?
If you ever scrolled through popular mods for most Bethesda titles, you might have noticed the sheer amount that either claims a performance improvement, or improves visuals while ignoring the performance implications (or sometimes lying about it for downloads).
Hardware & Software Used
- All tests are conducted on the following system at 1440p (lower resolution for CPU bound tests):
- 5800X3D with -30 Curve Optimizer voltage offset to prevent any thermal throttling.
- 6700XT QICK 319, stock settings, and driver-only install.
- P41 Plus SSD with Solidigm Synergy installed for improved I/O performance.
- 2x16GB @ 3600Mbps.
- Windows 11 Pro.
- Most benchmarks are done using CapFrameX, captures will be provided through download.
- The mods used as a base are listed below.
- Due to the frequent updates, the driver and Windows version used will be specified by each benchmark.
Any test that differs from this setup will have a section dedicated to a similar list.
Methodology & Info
- Don't give much consideration to 1% or 2% differences and a second or less of loading time, there is always margin of error.
- Any outlier result is corrected by aggregating more runs and leaving out weird ones.
- Testing the effect of a mod on CPU-side performance requires being completely limited by the CPU while profiling, this can be done with a low enough resolution and/or a fast enough GPU.
In 99% of cases, you can tell you're CPU bound when the GPU core usage is not maxed out (99/100%).
Refer to this video for more information. - Even if some of the performance results seem so high in FPS, remember that performance gains/losses largely scale the same across different hardware, obviously assuming the testing enviroment is comparable and nothing weird is going on.
Always refer to performance differences in percentage and not absolute values. - When testing performance you want to remove any factor unrelated to what the mod being tested affects. For this reason, AI is disabled during testing.
In the case of Fallout 4, the Papyrus VM performance is not an issue thanks to Buffout 4 NG which allows more operations per frame and High FPS Physics Fix which makes the timing dynamic and prone to less breakage at high framerates.
- AI and scripting are too variable and would require a different (also far more complex) testing methodology, so no tests of that kind are considered.
FO4 Benchmarks
Modlist Used (OG EXE + NG Content)
- Bethini Pie Ultra
- F4SE
- Plugin Preloader
- Address Library
- Backported Archive2 Support System
- Better Console
- High FPS Physics Fix
- Unofficial Fallout 4 Patch
- Community Fixes Merged
- Buffout 4 NG
PRP vs Roombounds
❮
❯
Date of testing: 16/11/24
Conclusion and notes:
- All runs were CPU bound.
- The drop in performance from PRP is due to some objects being excluded from Precombination, which is necessary to fix visual bugs.
- The Previs system is far more granular than roombounds, so the Roombounds mod will never perform better unless it is hiding enough NPCs (which Previs can't do) to make up for the worse occlusion, as seen in College Square.
Mod links and versions:
- Previsibines Repair Pack Stable Branch - PRP (74.16)
- Optimized Room Bounds and Portals - No PreVis and PreCombines for Interior Cells (002)
- Windows 11 Pro Build 26100
- Driver 24.10.1
Captures
High FPS Physics Fix vs Long Loading Times Fix
ENB Presets
❮
❯
GPU Bound
Date of testing: 28/11/24
Conclusion and notes:
- 0.501 is reported to have bugs so 0.496 is used instead.
- The weather mods recommended by each ENB preset aren't used in order to avoid extra factors, visual comparison isn't the point here as it is highly subjective.
- DECENT ENB needed the DoF fix pinned in the post section.
- DECENT ENB for some reason has awful lows in the CPU bound test (confirmed with more runs).
Mod links and versions:
- ENB (0.496)
- Spectrum - ENB for Lightweight Lighting (1.2)
- DECENT ENB - Natural and Performance Friendly (1.5.1)
- REACTOR - A FALLOUT4 ENB (1.4.4)
- ENB Helper for Fallout 4 (1.0.2)
- Windows 11 Pro Build 26100
- Driver 24.10.1
Captures