Use PerfDog analyze game performance bottlenecks
first clear the purpose of testing
the test report parsing
the first clear test purposes
recent quality check quality of the game, find fluency relatively poor, more Caton game,
first of all we must be clear where performance bottlenecks, that is the the purpose of our tests;
common games bottlenecks such as
CPU, GPU, memory, through Perfdog can easily get the data indicators; but first make sure your phone and computer to connect properly, for example, you can view by adb devices whether the phone connected to the computer;
like this:
Then remember to set up the data you want to capture, click on the bottom right corner of + Check the operation you want;
but note that, unless it is necessary to check to be captured according to your own needs data, after all, each additional a data, the impact will be more number of mobile phone performance, such as electricity, but do not worry be cpu, memory, gpu and other impacts, after all perfdog is obtained for the application, instead of capturing global data;
use of specific methods can refer
perfdog operation explanation
Of course there is also the meaning of some specific indicators representative, or you can find some description here
Perfdog support
Analytical test report
This is a low-end models selected
Here is a CPU data, it seems no problem, unlike the bottleneck.
This is a memory data, memory has been on the rise, rise, some danger, there may be a memory leak, and memory here is PSS memory data, high memory usage, total memory for a total of 1.8G mobile phone, the memory already high;
it can be regarded as a bottleneck point;
This is the GPU data, it can be seen, there is a clear high consumption area, 
a further determination, at this node CPu temperature rises suddenly, after Well GPU has burst table;
Continue to follow up the positioning at the time of the game, and found that just entered into another map, the map is then whether the problem then?
The GPU segment to see lower consumption point in time, find the time to enter the map 3, GPU consumption is reduced, the above data, when the map 2 GPU consumes a sharp rise burst table, so we conclude that there are no maps to optimize good 2 , may be some art resources is not optimized, or have any other show operating within this map;
So there are two conclusions of this test,
a large amount of memory, has been showing an upward trend, there may be a memory leak;
map within 2. Game 2 there is the problem of excessive consumption of GPU usually do not map well to optimize, but also does not exclude other Sao operation;
In summary, our analysis must first define their goals;
1. General overview report about the analysis;
2. take itemized comparison of the data indicators of their products;
such as our Android memory indicators
3. find the more obvious qualities region
if there is no obvious area can only rely on the experience a little bit of analysis slightly,
4. analysis concluded