0

HTML5 Features and JavaScript Performance Measurement

Armin Brunner

October 19, 2016 12:22

None

Follow

Dear collques,

Out of curiostity I did some measurment of the HTML5 features and JavaScript performance on different (and the newest) Brightsign players with the newest firmware of today:

Tests:
HTML5 Test Score: https://html5test.com/
WebKit SunSpider 1.0.2 JavaScript Benchmark: https://webkit.org/perf/sunspider-1.0.2/sunspider-1.0.2/driver.html
Generation	Player Model	FW Release	HTML5test Score	SunSpider Runtime
1	HD1020	6.1.76	no Browser	no Browser
1	XD1030	6.1.76	408 out of 555	3404 ms
2	XD1032	6.1.76	408 out of 555	2802 ms
2	XD1032	6.2.45	450 out of 555	3860 ms
2	4K1042	6.1.76	411 out of 555	732 ms
2	4K1042	6.2.45	453 out of 555	1226 ms
3	LS423	6.2.45	453 out of 555	1710 ms
3	HD1023	6.2.45	453 out of 555	1380 ms
3	XD1033	6.2.45	453 out of 555	1025 ms
3	XT1143	6.2.45	453 out of 555	944 ms

Very interesting is the high JavaScript performance loss of the XD1032 and 4K1042 between Firmware releases 6.1.76 and 6.2.45 

Has anybody an explanation for that?

 

And the most performant player for web sites with JavaScript is still the "old" 4K player with the "old" FW 6.1.76.

 

Regards

Armin

• Facebook
• Twitter
• LinkedIn

3 comments

• 
  support@brightsign.biz November 18, 2016 20:37 Official comment

  Hello Armin,

  The SunSpider test results you have seen are consistent with our internal testing and do indeed stem from the change from Chromium37 in 6.1 to Chromium45 in 6.2. However, because it focuses on simplistic benchmark measurements of the core JavaScript language--rather than features used in modern web applications--SunSpider is not an accurate benchmark for modern browsers. For this reason, SunSpider has been deprecated for both Chromium and Webkit development (see here for more details).

  For a more accurate measurement of performance changes between 6.1 and 6.2 firmware, as well as Series 2 and Series 3 models, we use the Speedometer and Jetstream tests: Speedometer measures the speed of interactive JavaScript, while Jetstream benchmarks raw JavaScript performance.

  Also note that we optimize our Chromium engine for digital signage applications: Because digital signs typically display one or a few web pages for a long period of time, we prioritize overall performance over initial page load times. Furthermore, Chromium uses a a version of JavaScript that is optimized for modern JavaScript frameworks rather than handcrafted JavaScript code, and this is reflected in the Speedometer scores for our players.

  See below for Speedometer and Jetstream test results for 6.1 vs. 6.2 and Series 2 vs. Series 3. Note that 6.2 exhibits moderate to marginal improvement over 6.1 for the 4Kx42 and XDx32, while Series 3 shows significant improvements over Series 2 in most cases. Since Chromium is optimized for ARM platforms, the XDx33/HDx23/LS423 models offer a huge performance boost over the XDx32/HDx22 models (because the former use ARM and the latter use MIPS). On the other hand, the difference in JavaScript performance between 4Kx42 and XTx43 is less dramatic because both are ARM platforms.

  Note: XTx43 and XDx33 players have much better WebGL/graphics performance than the 4Kx42, but the JavaScript benchmarks do not reflect this fact.

  
  
  

   

• 0
  
  ^UD\_$ October 19, 2016 12:34

  Thanks for sharing your tests results.

  Have you done multiple tests for each combination of model and firmware, and what you are showing is an average of multiple tests?

• 0
  
  Armin Brunner October 19, 2016 12:53

  I corrected and added the figures of the 4K1042 player with different FW releases.

  The browser in FW 6.1.76 is Chromium37, the browser in 6.2.45 is Chromium45.

   

  I executed the tests several times. The HTML5test score is always the same, the SunSpider runtime changes slightly.