Mobile Perf and Responsiveness

Goals

Engineering

Click the above to see the mobile perf "feature" page that helped spark the conversation. This captures where we were a mere month ago and represents the baseline plan. In summary need to work on:

Land on what architecture we want to use: do we stay with dual architecture or do we move to a single, multi-threaded process?

Status: dougt is working on a build using the Java front-end for testing and evaluation by 10/7.

2. Improve startup time

Status: Java architecture helps us with this but we need to continue to plow through mobile startup shrink bugs.

3. Understand Memory Usage

Status: Pending, more info due by 10/7

4. Improve Perceived Performance

Status: Pending, more info due by 10/7

5. Improve Page Load, Panning, Zooming

Status: Pending more info due by 10/7

Big Picture

  • Provide customers with a mobile browser that exceeds customer expectations in terms of responsiveness by March 2012
  •  Be faster than native Android Browsers
  • Be faster and better than iOS Browsers

Testing & Benchmarking

  • Scale our current automated performance testing and benchmarking infrastructure so that it supports continued innovation and yields reliable data for engineering and quality assurance
  • Status: Currently focusing on the reliability of Unit and Talos testing; we will then move onto adding new tests such as RSS. Dates to track incremental progress again is forthcoming.
  • Create standards of testing, measurement, and analysis to provide a concrete view into how Fennec performs to complement the above benchmarks to be then leveraged for ongoing crowdsource testing:
  • Status: We have an "approved" matrix of test cases and metrics we will collect. Methodology for testing and analysis due on 10/4. Aaron Train is doing a QA test day on 10/7.
  • Integrate with Telemetry
  • Status: Work with telemetry team to integrate metrics to crowdsource results; will hopefully land on a plan during the mobile work week.