Skip to benchmark

COUNTER-STRIKE 2 · DISPLAY MODE · SINGLE SYSTEM

Does Borderless Cost FPS in CS2? We Tested It.

Fullscreen versus borderless fullscreen, measured in the same CS2 benchmark scene with PresentMon.

4 tests per mode4 paired comparisons19 Oct 2026

Fullscreen had higher average FPS in this test

PresentMon · median of four tests

Fullscreen recorded a median of 188.57 FPS; borderless recorded 179.45 FPS. Borderless was 4.84% lower relative to the fullscreen median. All four paired average-FPS results favored fullscreen, and the observed ranges did not overlap.

Average FPSMedian · higher is better · zero-based scale
Fullscreen
188.57 FPS
Borderless
179.45 FPS

Fullscreen range: 187.96–194.45 · Borderless range: 178.71–179.76

1% low FPSMedian · higher is better · zero-based scale
Fullscreen
76.40 FPS
Borderless
72.08 FPS

Fullscreen range: 71.18–81.47 · Borderless range: 68.99–77.37

p99 frame timeMedian · lower is better · zero-based scale
Fullscreen
10.89 ms
Borderless
12.00 ms

Fullscreen range: 10.14–12.13 · Borderless range: 11.16–12.36

Limits: the 1% low and p99 ranges overlap, and paired results do not show a consistent advantage across those metrics. This is one PC, one CS2 build and one synthetic map test; it does not predict every match or measure input latency.

Four paired comparisons in chronological order: Fullscreen–Borderless, Borderless–Fullscreen, Borderless–Fullscreen, Fullscreen–Borderless. No measured runs were removed.

Eight measured 60-second PresentMon windows; warmup runs are not included.
Pass / pairDisplay modeAverage FPS1% low FPSp99 msFrames
1Pair 1Fullscreen194.4581.4710.1411,667
2Pair 1Borderless178.7168.9912.3010,723
3Pair 2Borderless179.7677.3711.1610,785
4Pair 2Fullscreen188.0571.1812.1311,283
5Pair 3Borderless179.1672.4312.3610,749
6Pair 3Fullscreen187.9679.0610.5711,277
7Pair 4Fullscreen189.0973.7311.2011,345
8Pair 4Borderless179.7571.7411.7010,785
VProf full-benchmark values (separate from PresentMon)

VProf summarizes the full in-game benchmark. The PresentMon values above use a fixed 60-second window inside that run, so the two tools' FPS values are not directly comparable. VProf P1 is also a separate metric from PresentMon 1% low.

Values are included for completeness and are not used in the PresentMon charts.
Pair / passDisplay modeVProf average FPSVProf P1
1Pair 1Fullscreen374.1093.00
2Pair 1Borderless344.40110.90
3Pair 2Borderless358.00105.00
4Pair 2Fullscreen370.30113.40
5Pair 3Borderless341.5076.90
6Pair 3Fullscreen368.90121.80
7Pair 4Fullscreen370.80130.00
8Pair 4Borderless358.7097.80

Test setup & definitions

CPU
AMD Ryzen 5 7600X3D
GPU
NVIDIA GeForce RTX 4060
DRIVER
32.0.16.1742
OS
Windows 11 Pro · build 26300
CS2 BUILD
25966348
DISPLAY
1440 × 1080 at 144 Hz
SCENE
FPSHeaven Ancient · Workshop map 3472126051
POWERPLAN
Balanced
GAMEMODE
On
FPSCAP
Temporarily removed for the benchmark and restored afterward
CAPTURE
PresentMon · fixed 60-second window starting 10 seconds after observed VProf start
CAPTUREWINDOW
CPU frame-start intervals from the same map and scene
ORDER
Fullscreen → Borderless · Borderless → Fullscreen · Borderless → Fullscreen · Fullscreen → Borderless

Fullscreen and borderless modes were measured four times each in a balanced order. Power plan, Game Mode, resolution, refresh rate and the remaining saved graphics settings were held constant. The FPS cap was temporarily removed for the series and restored afterward.

Average FPS = 1,000 ÷ mean CPU frame-start interval in milliseconds. 1% low = 1,000 ÷ mean of the slowest ceil(1%) intervals. p99 is the nearest-rank 99th percentile frame interval. These are rendered-frame timing measurements, not displayed-frame or latency measurements. See the PresentMon documentation.

Data quality & downloads

Download report JSON ↓

The complete 30-test series passed its final restoration check: CFG hashes and attributes, power plan, Game Mode, display mode, NVIDIA state, game closure and collector completion were verified. This page includes only the eight display-mode measurements.

PresentMon reported Hardware: Independent Flip in both fullscreen and borderless runs. That observation does not establish that fullscreen used exclusive presentation. The original frame trace and private local logs remain local; downloads contain derived values only.

  • One PC, one game build, one map, and four measured passes per display mode.
  • Fullscreen median average FPS was higher in this series; the 1% low and p99 ranges overlap and their paired results do not establish the same consistent advantage.
  • PresentMon reported Hardware: Independent Flip for both modes. This does not prove exclusive fullscreen presentation.
  • FPS and frame-time measurements do not measure input latency or guarantee results in other games, maps, or PCs.
  • The public downloads contain derived measurements, not the original frame trace or private local logs.