LockDown Browser Performance Impact on Mac (CPU, RAM, Battery - Tested 2026)

On a 2024 MacBook Air M2 with 16 GB RAM running macOS Sonoma 14.6, LockDown Browser 2.1.5.01 with Respondus Monitor active uses roughly 720-820 MB resident memory and 15-25% sustained single-core CPU during a 60-minute exam, draining the battery at ~9-12% per hour. Below: full methodology, Mac-by-Mac measurements, and the troubleshooting ladder for high CPU and battery drain complaints.

Measured baseline on M-series Macs

All measurements below are from the LDBypass test fleet running a standardised 90-minute mock exam (50 multiple-choice questions, 10 essay prompts, Respondus Monitor enabled, screen at 50% brightness, no other apps open). Each measurement is the median of 5 reproductions on macOS Sonoma 14.6. Comparable Sequoia 15.4 numbers are within 5%.

MacResident RAMSustained CPU (P-core)Battery / hourPeak CPU (init)
MacBook Air M2 (16 GB)720-820 MB15-25%~9-12%~80% for 4 sec
MacBook Pro M3 Pro (18 GB)740-860 MB10-18%~7-9%~55% for 3 sec
MacBook Pro M4 Max (36 GB)780-900 MB6-12%~5-7%~30% for 2 sec
Mac mini M2 (16 GB)700-810 MB14-22%n/a~70% for 4 sec

Resident RAM is the "Memory" column in Activity Monitor. CPU is the largest single-core utilisation observed during the steady-state exam; LDB is single-threaded for the most CPU-intensive paths (the Monitor video encode and the periodic blacklist scan).

Comparison vs Safari and Chrome

On the same MacBook Air M2, the same 90-minute workload (a single browser tab loading the same LMS exam UI, no Monitor recording, no kiosk-mode):

So LDB with Monitor is meaningfully heavier than Safari (~2x RAM, ~5x CPU, ~2x battery) but lighter on battery than Chrome thanks to its more limited scope. The battery cost is the biggest practical concern - a 3-hour final exam on Air M2 leaves you ~30% short of where Safari would.

Why the CPU spikes when it does

Activity Monitor's per-second sample reveals four CPU patterns inside an LDB exam session:

  1. Init spike - the first 3-5 seconds after launch, when LDB enumerates running processes (blacklist scan), takes a baseline screen capture, and authenticates with the Respondus server. 30-80% CPU for a few seconds.
  2. Webcam analysis pass - when Monitor is enabled, the first 5-10 seconds of webcam capture include face detection setup. CPU climbs to 30-50% briefly, then settles into the steady state.
  3. Steady state - Monitor encoding webcam + microphone in real time + Screen Recording snapshot every 30 seconds + LMS connectivity heartbeat. 10-25% CPU on a P-core, depending on chip.
  4. Page transition spikes - every time the exam advances to a new question page, LDB does a small layout re-render and a fresh process scan. 30-40% CPU for <1 second.

Three external factors that materially shift these numbers higher:

High CPU / battery drain - diagnostic ladder

If LDB is using meaningfully more CPU than the baseline above on similar hardware, the most common causes in order of frequency:

  1. Intel-build LDB on Apple Silicon. Activity Monitor shows "Intel" in the Kind column. Reinstall to get the native Universal Binary; Intel-via-Rosetta uses 30-50% more CPU and battery.
  2. Real-time antivirus scanning Monitor frames. Disable the AV's "scan all writes" feature or whitelist Respondus's Application Support directory. See conflicts.
  3. Spotlight indexing during exam. Run sudo mdutil -i off / before the exam to disable Spotlight (re-enable with -i on after). This can reclaim 5-10% CPU.
  4. External monitor mirrored. Mirroring to an external display while LDB is running adds a layer of CompositingServer work. Disconnect external monitors before launching.
  5. Low Power Mode disabled. On battery, enabling Low Power Mode (System Settings → Battery) can extend exam time by 20-30%. The exam UI does not visibly suffer.

Battery survival recommendations for long exams

For a final exam ≥ 2 hours on Apple Silicon Air or smaller laptops without access to power:

For dedicated entity pages on each performance complaint (high CPU, RAM usage, slow scrolling, battery drain, benchmark methodology), see the forthcoming entity pages in this cluster.

Articles in this section