Submit A Tip Alternative Tip Form

LightSpark Vs Flash: Benchmarked

The team behind open-source Flash alternative LightSpark pushed out a new bug-fix release a few days ago.

Seeing as we’re a few releases in, I felt now was a prime time to see just how well LightSpark performs against its proprietary counterpart – particularly so I have a benchmark which  to compare future releases against.

Now, you don’t need to be remotely “on-the-ball” to predict that LightSpark won’t out-whoop the real Flash plug-in, but if users are considering taking a performance hit by switching they might want to know exactly how big a hit it will actually be…

Tests

For my highly unscientific testing I used stock Flash from the repo in both Firefox and Chromium, Google-Chrome’s built in Flash plug-in and, finally, LightSpark’s browser plug-in for Firefox.
When choosing which tests to run I opted for a number of “real world” use cases. Y’know, YouTube videos, flash games and websites, etc. For more concise tests I opted for GUIMark2.
Scores

CPU load scores were taken at three intervals during the respective activities and the average presented below. For the most part the loads hovered nicely around these scores. RAM usages is in MiB and, again, is the average based on three intervals.

GUI Mark tests were each run three times and the average score listed.

Firefox & Flash 10.1.53.64

Firefox 3.6.8 coupled with stock Flash from the Ubuntu repository.  
  • YouTube: CPU 44%, 24MiB
  • Flash game #1: 58% 9.3MiB
  • Flash-heavy website: 58%, 40MiB
  • Flash Game #2: 72%, 36MiB

GUIMARK2

  • Bitmap Gaming: 12.04 fps
This test stresses pushing around lots of bitmap assets that animate each frame.
  • Vector Chart: 15.34 fps
This benchmark is designed to stress the vector apis by simulating a streaming stock chart.

Firefox & Lightspark

Firefox 3.6.8 partnered with the LightSpark browser plug-in.
  • YouTube: CPU 96%, 24.8MiB
  • Flash heavy website: 100% CPU, 29.7MiB; wouldn’t load past intro – no menus etc
  • Flash game #1: wouldn’t load.
  • Flash game #2: wouldn’t load.

GUIMARK2

  • Bitmap Gaming: N/A
  • Vector Chart: N/A

Finally, just for comparison, I ran the same tests again  in Google Chrome & Chromium.

Google Chrome ships with a built-in version of Flash whilst Chromium makes use of the Flash plug-in proper although preliminary support for Lightspark in Chromium is available.

Google Chrome w/built in flash

  • YouTube: 20% CPU, 25.4MiB RAM
  • Flash heavy Website: 50%, 29MiB
  • Flash game #1: 52%, 20MiB
  • Flash game #2: 82%, 49MiB

GUIMARK2

  • Bitmap Gaming: 12.53 fps
  • Vector Chart: 16.65 fps

Chromium Flash 10.1.53.64

  • YouTube: 56%, 29MiB
  • Flash heavy website: 52%, 23MiB
  • Flash game #1: 54%, 15MiB
  • Flash game #2: 86%, 45.6MiB

GUIMARK2

  • Bitmap Gaming: 9.55 fps
  • Vector Chart: 15.51 fps

Notes

The rather poor showing for LightSpark shouldn’t be surprising given its relative freshness. The fact that many of the Flash-based activities we tried couldn’t load shouldn’t take away from the advances it is making but rather highlight to casual users keen to ditch flash on a whim the relative shortcomings of tool.

The resource usage was not only much higher but frequent page lock-ups and stuttery YouTube video playback rendered the few jobs it could handle moot.

LightSpark aside, something else shows up in these results: how much more resourceful Flash was on YouTube in Google Chrome. The CPU load barely ever went above 20%. Flash is a typically a resource hound on Linux so seeing such a relatively good showing is great to see.

Google’s choice to “improve the Flash Player experience in Google Chrome” seems to be paying off. Whilst RAM usage was often ever-so slightly higher than standalone flash the lower CPU draw and over-all feel was very fluid. The GUIMark2 tests also bear this out, receiving better scores than the standalone version.