Strik benchmark sources

I’m trying to run the Strike benchmark on some devices and it looks like some tweaks are needed - are the sources available?

hi @elsassph ! The current source of Strike is not open source, but we are planning to open source the tests in the future.

This will likely not be done anymore with Strike 1.0 but as part of our Strike 2.0 roadmap we’re looking into making the tests publicly available for collaboration.

For now, I think its easiest to help you out directly. Can you describe what kind of tweaks you are looking for on your test setup? Are you running into WebGL issues? or is this performance based? Any more details you can provide would help tremendously.

Tagging @Douwe for awareness.

I’d like an option to provide config tweaks, like resolution 720p, disabling web worker, forcing canvas 2D…

Alternatively a downloadable build for self hosting could do the trick.

Seems like the tweaks are mostly about passing in configuration. That’s definitely a good idea for Strike2.0, I think!

Technically we could probably add some basic support for this in Strike1. I’m thinking perhaps passing the options via querystring params. Not sure if we still want to do that for Strike 1, though. I guess it depends on how much effort it would take …

Thinking there may be cases where extra polyfills would be needed also so an offline version would still be desirable.

Interesting requests, though I’m not sure if that’s the original purpose we had for the Strike tool. Creating these many “configurables” makes it hard to establish a baseline.

May I ask how you are using the Strike tool @elsassph? What purpose does it serve in the way you intend to use it like that?

The tool seem to be focused on RDK-based devices and I wanted to test on some other TVs where some adjustments may be needed.

Hi @elassph, the purpose of the Strike tool is to validate the capability of a device to run Lightning Apps. It does this by running a fixed set tests and calculating a score based on how well this particular device performs. That final score should give an idea of what to expect of said device.

It’s not focussed specifically on RDK devices (not sure where you got that impression?). Strike can be run any device. We’ve even ran it on an iPhone for example :slight_smile:

Coming back on the original question, and Wouter’s concern about creating too many configurables: some basic config (a limited, predefined set) might be something useful, if that allows Strike to run or more devices. But making Strike fully configurable, kind of defeats the purpose of the tool, which is comparing devices against the baseline of tests.

It would be helpful if you could give a list of examples of issues you’re running into that could be solved via passing some configuration, keeping in mind Strike’s initial purpose. The 720p option for example, sounds like something reasonable to make configurable.

Sure I don’t really want to change anything in particular, just make it run on some devices where it doesn’t work directly (like webOS) and I’d like a downloadable build of strike to troubleshoot it.