Back to all dispatches
Benchmarks March 05, 2026· 7 min read

Benchmarking Playwright vs. Appium on Physical Apple Silicon

A rigorous 1,000-run comparison of session initialization, assertion throughput, memory overhead, and thermal throttling.

BG
Benchmarks Guild
Gravity QA Performance Team

Test Environment & Methodology

To measure real-world performance differences, we automated a standard 14-step e-commerce checkout flow across 1,000 consecutive runs on identical physical hardware: a Google Pixel 8 connected via USB 3.2 to an Apple M3 Max (36GB RAM).

We compared two stacks: (A) Standard Appium 2.0 with uiautomator2 driver running in Node.js, vs (B) Gravity QA native gRPC orchestrator.

The Benchmark Numbers

The performance delta was immediate across every metric:

textBenchmark results averaged over 1,000 consecutive automated test runs.
----------------------------------------------------------------------
METRIC                       APPIUM 2.0 (UIA2)     GRAVITY QA (NATIVE)
----------------------------------------------------------------------
Session Initialization Time   18.4 seconds           0.8 seconds (23x faster)
Average Element Query Time    184 ms                 14 ms       (13x faster)
Full Checkout Suite (14 steps) 48.2 seconds          5.6 seconds (8.6x faster)
Memory Footprint (Workstation) 640 MB (Node+JVM)     84 MB (Rust Core)
Flakiness Rate (1,000 runs)    6.8% false failures   0.1% false failures
----------------------------------------------------------------------

The 23x difference in session startup is primarily due to Appium reinstalling helper APKs and establishing multiple serial HTTP proxies, whereas Gravity QA maintains persistent daemon sockets that connect instantly.

BG

Benchmarks Guild

Gravity QA Performance Team

The Gravity QA Engineering team is dedicated to building low-level, desktop-native testing pipelines, compiler-level heuristics, and high-performance device orchestrators.

Further Dispatches

Founder's Story5 min read

“I didn’t plan to build a QA platform. I just wanted to solve the problems I kept facing.”

From discovering SQA during university with Cypress to leading QA teams across manual, automation, API, and mobile testing—and finally asking: “Why are we still doing this this way?” The real journey behind Gravity QA & Graviton.

Architecture8 min read

Why Cloud Device Farms Are an Architectural Dead End

Renting remote virtualized phones in a distant data center adds 400ms of WAN latency and network jitter to every assertion. Here's why desktop-native execution is the only architecture that scales.

Ship with fearless velocity

Experience the desktop-native testing platform built by engineers who lived through the pain.