当前位置:网站首页>Heap concept in JVM
Heap concept in JVM
2022-07-04 07:37:00 【xmh-sxh-1314】
One . Overview of heap
One JVM Instance has only one heap memory , So is heap Java The core area of memory management , Pile on JVM Create at startup , Its size is also created , yes JVM The largest memory space in the , All threads share Java Pile up , Physically discontinuous, logically continuous memory space , Almost all instances allocate memory here , At the end of the method , Objects in the heap are not deleted immediately , Only deleted during garbage collection , The pile is GC( Garbage collector ) Key areas for implementing garbage collection .
Two . Heap space subdivision
Java7 As before, heap space is logically divided into three parts : New Area + Retirement area + Forever
Java8 And later, the heap memory is logically divided into : New Area + Retirement area + Meta space
The new generation :
1. The new generation uses replication algorithm
2. The new generation is gc The key object of , Officially tested 70% The life cycle of objects will end in the new generation
3. The new generation is divided into eden、survivor1、survivor2, Object creation is first placed in eden in , After a certain period of time, it will be placed in the survivor area
4. The memory ratio is assumed by default :8:1:1
5. Cenozoic collector :Minor GC/Young GC
eden( New Area )
When the object is initially loaded, it will enter the new area
survivor( Survival zone )
The survival zone is further divided into from and to — Who is empty, who is to , There will always be an empty area .
The surviving area will not actively recycle garbage , It's just eden Only when recycling gc
When the threshold in the surviving area reaches 15 after ( Default 15 Modifiable ) Will automatically enter the elderly generation
When the new area (eden) When there is insufficient memory , Will be carried out in YGC, Then the object without pointer will be recycled , And the object to which the pointer is directed is put survivor1 perhaps survivor2 In the region ,eden Empty , Put the data into a survivor in .— When the second time gc It will be eden Put the data into another empty survivor in , And will present survivor Valid data in , Put the empty survivor in , One analogy .
Old age
1. Larger object data will be put into the elderly
2. The data of the s are relatively persistent and infrequent gc
3.(MajorGC / Old GC) It's going on majorgc At least once minorGc , and majorgc Is more efficient than minorGc slow 10 Times
4. Old age collector :MajorGC / Old GC Distinguish between and Full GC
When an object enters memory Will enter eden, If the full (YGC Recycle what is not referenced , If there are still references ) Will be put into the s1 perhaps s0 This involves to from Which is empty is to,( The next time eden When it is full again, there will be data 【 give an example s1】 in Data into s0, And iterative version ) And so on , When the number of iterations of an object reaches the default 15 thereafter ,( Of course, there will be special optimization : As if survivor The total memory of the same age in the region is greater than survivor Half of the memory , Objects older than or equal to the average age will be placed in the elderly generation in advance ) It will be put into the old generation About YGC The entire (YoungGC) It can also be for (Minor GC) s1,0 There will be no separate gc Recycling will only passively rely on eden Of gc When eden Conduct gc It will be recycled automatically s1,s0
yangGC Only in eden When the area is full , Not in survivor When the area is full ,eden District GC I will also put survivor Area for GC, When survivor in age=15 The data will be put into the elderly area .
Why three generations
Four memory allocation strategies ( Or object promotion rules )
If the object is EDEN Was born and passed through once MinnorGC Still alive after , And can be Survivor If it can be accommodated , Will be placed in the survivor area , And set the target age to 1, Every time I get through MinnorGC,age Add one year , When age Add to 15 when , Will be put into the old generation
5、 ... and .TLAB
Vi. escape analysis
What is escape ?
That is, if you create an object in a method , also return Outgoing , Or assign values to external variables , Then we run away .
-XX:+DoEscapeAnalysis (JDK1.8 Then it turns on by default )
-XX:+DoEscapeAnalysis( close )
Escape analysis includes the following
On the stack
That is, allocate objects directly to the stack , Die with the death of stack , Less gc( There is no... In the stack gc), Improved performance 、 Speed .
Synchronous ellipsis
Because it is unique to each stack , A stack is also a thread, so there is no synchronization safety problem .
Separate objects or scalar substitution
expand : A class represents a : The amount of polymerization , Scalars are the smallest data that cannot be analyzed , Aggregate quantity can be analyzed as scalar quantity , That is, analyze attributes
That is, when loading a pojo Class time , Instead of creating objects , Scalar substitution is analyzed into small attributes , Reduced memory , Improved performance .
But based on hotSpot The technology of virtual machine is not mature , Because we still need to judge whether Belong to escape , If there is no escape , It may waste judgment time and other problems .
But in the end, scalar substitution refers to hotSpot In the virtual machine
边栏推荐
- [thread pool]
- L1-028 judging prime number (10 points)
- Advanced MySQL: Basics (5-8 Lectures)
- Two years ago, the United States was reluctant to sell chips, but now there are mountains of chips begging China for help
- It's healthy to drink medicinal wine like this. Are you drinking it right
- Take you to master the formatter of visual studio code
- Node connection MySQL access denied for user 'root' @ 'localhost' (using password: yes
- Label management of kubernetes cluster
- Boast about Devops
- Zephyr 学习笔记1,threads
猜你喜欢

Amd RX 7000 Series graphics card product line exposure: two generations of core and process mix and match

MySQL storage engine
![[kubernetes series] kubesphere is installed on kubernetes](/img/2b/eb39cf78b3bb9908b01f279e2f9958.png)
[kubernetes series] kubesphere is installed on kubernetes

flask-sqlalchemy 循环引用

Introduction to deep learning Ann neural network parameter optimization problem (SGD, momentum, adagrad, rmsprop, Adam)

User login function: simple but difficult

This article is enough for learning advanced mysql

University stage summary

Guoguo took you to write a linked list, and the primary school students said it was good after reading it

Easy to understand: understand the time series database incluxdb
随机推荐
Redis - detailed explanation of cache avalanche, cache penetration and cache breakdown
Zephyr 学习笔记2,Scheduling
L2-013 red alarm (C language) and relevant knowledge of parallel search
Transition technology from IPv4 to IPv6
University stage summary
[C language] open the door of C
L1-030 one gang one (15 points)
SQL注入测试工具之Sqli-labs下载安装重置数据库报错解决办法之一(#0{main}thrown in D:\Software\phpstudy_pro\WWW\sqli-labs-……)
PCIE知识点-010:PCIE 热插拔资料从哪获取
The crackdown on Huawei prompted made in China to join forces to fight back, and another enterprise announced to invest 100 billion in R & D
Blue Bridge Cup Quick sort (code completion)
How to buy financial products in 2022?
Electronic Association C language level 1 34, piecewise function
A real penetration test
One of the general document service practice series
Data double write consistency between redis and MySQL
Valentine's Day is coming! Without 50W bride price, my girlfriend was forcibly dragged away...
Introduction to sap commerce cloud B2B organization function
tornado之目录
时序数据库 InfluxDB 2.2 初探