memory thrashing

One . Memory jitter

1. Memory jitter ： It refers to the phenomenon that a large number of objects are created or recycled in a short time .

2. Causes of memory jitter ： Mainly frequent ( Very important ) Create objects in loops ( A large number of objects are created in a short time , Because new objects occupy memory space and are frequent , If you create an object in a loop once or twice, it has little effect on memory , It will not cause serious memory jitter, which is acceptable and inevitable , Frequently, the memory jitter is very serious ), The effect of memory jitter is if the jitter is frequent , Will cause the garbage collection mechanism to run frequently ( Generate a large number of objects in a short time , Need a lot of memory , And still shake frequently , You may need to reclaim memory for generating objects , The garbage collection mechanism will naturally run frequently ).

3. Memory jitter affects ： Frequent memory jitter can cause garbage collection to run frequently , Cause the system to jam .

Solution

Where are the common memory jitters ？

1.Android Of onDraw, onMeasure Method .

here , The authorities do not recommend creating objects here , Why? ？ I see , It is frequently called by the bottom layer ,（ You can learn about the Android refresh mechanism ）, If you create a large number of objects , Frequent recycling , Memory jitter is easy . If it is bitmap, His memory is in native The distribution of , Processing needs to go through the kernel , Make the waiting time longer .

2. Handle io The place of

Read and write files , The memory stream creates a large number of byte[] The object will appear .

3. Produce components of the consumer model , These frequently produce a large number of objects , Like logs , Embedded point components, etc

4. Complex type json analysis .

such as Gson. By default , Is to generate new instances through reflection . Using the object pool through the adapter can solve

Recommend solutions ：

adopt Reuse objects Use to solve memory jitter ,

For example, enjoy yuan design pattern , Object pool scheme . But object pooling does not reduce memory , You need to limit the memory , And each object scene is different , You may need to filter . need lru Data structure of the algorithm, etc . May refer to handler,glide Object pool principle .

The principle of object pooling is simple , It is to create objects and store them in memory after use , For example, save it in the linked list , The next time you use an object , Look it up in the linked list first , If there is, take it out , If not, create , This avoids the creation of a large number of objects