TARS COUNTER-STRIKE 2 · SINGLE SYSTEM
CS2: Balanced vs High Performance Power Plan
CS2: Balanced vs High Performance Power Plan. Four measured tests per setting, paired PresentMon frame timings, methodology and downloadable results.
No clear performance advantage in this test
Median and observed rangeHigh performance recorded a 0.48% higher median average FPS, a 3.80% lower 1% low and a 2.54% higher p99 frame time. Ranges overlapped and paired results were mixed; no metric met the declared decision rule.
Balanced: 188.56–190.94
High performance: 188.58–192.03
Balanced: 75.44–81.72
High performance: 65.48–82.92
Balanced: 10.03–10.96
High performance: 9.66–13.05
Input latency: not measured. FPS and these source-specific timing metrics do not establish click-to-photon latency or guarantee smoother gameplay.
Every measured test
Download measurements CSV ↓| Pass / pair | Setting | Average FPS | 1% low FPS | p99 frame time (ms) | Frames |
|---|---|---|---|---|---|
| 3Pair 1 | Balanced | 190.94 | 81.72 | 10.03 | 11,456 |
| 4Pair 1 | High performance | 189.03 | 76.59 | 11.06 | 11,341 |
| 5Pair 2 | High performance | 192.03 | 82.92 | 9.66 | 11,521 |
| 6Pair 2 | Balanced | 188.90 | 78.01 | 10.77 | 11,334 |
| 7Pair 3 | High performance | 188.58 | 65.48 | 13.05 | 11,315 |
| 8Pair 3 | Balanced | 188.56 | 75.44 | 10.96 | 11,314 |
| 9Pair 4 | Balanced | 188.62 | 81.53 | 10.26 | 11,317 |
| 10Pair 4 | High performance | 190.32 | 76.88 | 10.51 | 11,419 |
Warmups and separate VProf results
VProf reports the full benchmark. Its average FPS and source-defined P1 are separate from PresentMon measurements; absolute values across these windows are not interchangeable.
| Pass | Setting | Purpose | VProf average FPS | VProf P1 |
|---|---|---|---|---|
| 1 | Balanced | Warmup | 366.60 | 121.00 |
| 2 | High performance | Warmup | 369.60 | 115.70 |
| 3 | Balanced | Measured | 371.90 | 121.60 |
| 4 | High performance | Measured | 370.50 | 121.30 |
| 5 | High performance | Measured | 374.00 | 124.20 |
| 6 | Balanced | Measured | 371.90 | 122.40 |
| 7 | High performance | Measured | 357.80 | 120.60 |
| 8 | Balanced | Measured | 370.90 | 122.70 |
| 9 | Balanced | Measured | 370.70 | 117.90 |
| 10 | High performance | Measured | 372.50 | 118.30 |
Test setup & definitions
- CPU
- AMD Ryzen 5 7600X3D
- GPU
- NVIDIA GeForce RTX 4060
- Driver
- 32.0.16.1742
- OS
- Windows 11 Pro 10.0.26300
- CS2 build
- 25966348
- Scene
- FPSHeaven Ancient · workshop 3472126051
- Display
- 1440 × 1080 · 144 Hz
- Fixed controls
- Game Mode on · fullscreen · 4× MSAA
- Capture
- PresentMon 2.6.0 · fixed60-offset10-v1
Only the Windows power plan changed in this comparison. Other controls remained fixed. The game relaunched for each test.
Two warmups (A → B), followed by four measured pairs: A → B, B → A, B → A, A → B. Warmups never enter the summaries.
PresentMon covers exactly 60 seconds, starting 10 seconds after the observed VProf start. Average FPS = 1,000 / mean consecutive CPU frame-start interval in ms; 1% low = 1,000 / mean of the slowest ceil(1%) intervals; p99 uses the nearest-rank 99th percentile.
The declared decision rule requires at least a 2% median difference, all four paired differences in the same direction and non-overlapping ranges. Failure to meet it does not prove equivalence or statistical significance.
Data quality & downloads
The combined three-comparison session completed 30 tests: 6 warmups and 24 measurements. Earlier incomplete sessions were kept separate.
All 10,185 sampled foreground observations matched the owned game process; maximum observation gap was 0.388 seconds. Sampling cannot rule out brief changes between checks.
Collectors completed. Original CFG hashes and attributes, Windows power plan, Game Mode, NVIDIA state and display state were independently verified after restoration.
- One PC; four measured passes per condition; not statistical certainty
- Existing power plans, not assumed factory defaults
- Game Mode preference verified in UI; scheduler activation not independently instrumented
- No input latency or unsupported power-consumption claims
- VProf full interval differs from PresentMon fixed60-offset10-v1
- Non-windowed Game Bar hosts permitted; recording and visible overlays prohibited
Downloads contain allowlisted derived measurements, including warmups. Original captures, account identifiers, local paths, device identifiers and configuration contents remain private.
Source integrity
Three independent CS2 tweaks · completed 19 October 2026
SHA-256 of the original local analysis artifact (not the downloadable derived report):
8b8b89e14e1c6cacc0ee708723747d77ef097df7d3d80e6d11736ba6596e650bDownload report JSON