Streamline benchmark delivery to toolchain

Michael Hope michael.hope at linaro.org
Wed Jan 11 01:01:06 UTC 2012


On Wed, Jan 11, 2012 at 10:20 AM, Zach Pfeffer <zach.pfeffer at linaro.org> wrote:
> On 10 January 2012 14:54, Andy Doan <andy.doan at linaro.org> wrote:
>> On 01/10/2012 01:55 PM, Zach Pfeffer wrote:
>>> We've started to spin up an effort to automate Android benchmarks in
>>> LAVA (thanks Andy and Bero). The toolchain group is interested in
>>> these results and would like the results in raw text form so that they
>>> can extend their parser to read it. I figured we should just shoot for
>>> this first since getting benchmark results to the toolchain group is
>>> very important.
>>>
>>> Does this sound good to you Andy and Bero?
>>
>> Sure.
>>
>>> Michael/Åsa would you guys mind sending a sample of a preferred format
>>> and how you'd like it delivered?
>>>
>>> Paul, what ways can LAVA push out a text file?
>>>
>>
>> I might be able to explain some of this. I'm pushing results to a
>> "bundle stream" in the validation server like:
>>
>> <http://validation.linaro.org/lava-server/dashboard/streams/anonymous/doanac/bundles/f03d177c4318c2e206b29e1d8b9b028f3410319b/>
>>
>> The bundle streams are in a pretty easy to manage JSON format. The page
>> itself shows an example of how to download the stream. However, I like
>> to just use python to deal with them. I'm attaching an example script
>> I've written that looks at a bundle stream and produces a summary of
>> averages.
>
> Michael/Asa, is the JSON format okay?

Yip.  Text is good.  My parser is Python based so we can pull this in easily.

-- Michael



More information about the linaro-android mailing list