Strategies for the Allocation of Memory

Contents[Show]

There a lot of different strategies for allocating memory. Programming languages like Python or Java request their memory from the heap at runtime. Of course, C or C++ has also a heap but prefers the stack. But these are by far not so only strategies for allocating memory. You can preallocate memory at the start time of the program as a fixed block or as a pool of memory blocks. This preallocated memory can afterward be used at the runtime of your program. But the key question is: What are the pros and cons of the various strategies to allocate memory?

At first, I want to present four typical memory strategies.

Strategies for the allocation of memory

Dynamic allocation

Dynamic allocation also called variable allocation is known to each programmer. Operations like new in C++ or malloc in C request the memory when needed. On contrary, calls like delete in C++ or free in C release the memory when not needed anymore.

 int* myHeapInt= new int(5);
 delete myHeapInt;

The subtle difference in the implementation is if the memory will be automatically or not automatically be released. Languages like Java or Python know general garbage collection, C++ or C is contrary not. (Only for clarification: That is of course not totally true because we have the Boehm-Demers-Weiser conservative garbage collector that can be used as a garbage collecting replacement for C malloc or C++ new. Here are the details: https://www.hboehm.info/gc/).

Dynamic memory management has a lot of pros. You can automatically adjust the memory management to your needs. The cons are that there is always the danger of memory fragmentation. In addition, a memory allocation can fail or take too much time. Therefore, a lot of reasons speak against dynamic memory allocation in highly safety-critical software that requires deterministic timing behavior.

Smart Pointers in C++ manage their dynamic memory by objects on the stack.

Stack allocation

Stack allocation is also called memory discard. The key idea of stack allocation is that the objects are created in a temporary scope and are immediately freed if the objects go out of scope. Therefore, the C++ runtime takes care of the lifetime of the objects. Scope are typically areas like functions, objects, or loops. But you can also create artificial scopes with curly braces.  

{
  int myStackInt(5);
}

 

A very beautiful example for stack allocation is the smart pointers std::unique_ptr and std::shared_ptr. Both are created on the stack in order to take care of objects that are created on the heap.

But what are the benefits of stack allocation? At first, memory management is as easy as possible because the C++ runtime manages it automatically. In addition, the timing behavior of memory allocation on the stack is totally deterministic. But there is a big disadvantage. On one hand, the stack is smaller than the heap; on the other hand, the objects should often outlive their scope and C++ supports no dynamic memory allocation on the stack.

Static allocation

Most of the time static allocation is called fixed allocation. The key idea is that at runtime required memory will be requested at the start time and released at the shutdown time of the program. The general assumption is here that you can precalculate the memory requirements at runtime.

char* memory= new char[sizeof(Account)];
Account* a= new(memory) Account;

char* memory2= new char[5*sizeof(Account)];
Account* b= new(memory2) Account[5];

 

The objects a and b in this example are constructed in the preallocated memory of memory and memory2.

What are the pros and cons of static memory allocation? In hard real-time driven applications, in which the timing behavior of dynamic memory is no option, static allocation is often used. In addition, the fragmentation of memory is minimal. But of course, there are pros and cons. Often it is not possible to precalculate the memory requirements of the application upfront. Sometimes your application requires extraordinary memory at runtime. That may be an argument against static memory allocation. Of course, the start time of your program is longer.

Memory pool

Memory pool also called pooled allocation combines the predictability of static memory allocation with the flexibility of dynamic memory allocation. Similar to the static allocation you request the memory at the start time of your program. In opposite to the static allocation, you request a pool of objects. At runtime, the application uses this pool of objects and returns them to the pool. If you have more the one typical size of objects, it will make sense to create more than one memory pool.

What are the pros and cons? The pros are quite similar to the pros of static memory allocation. On one hand, there is no fragmentation of memory; on the other hand, you have predictable timing behavior of the memory allocation and memory deallocation. But there is one big advantage of memory pool allocation to static allocation. You have more flexibility. This is for two reasons. The pools can have different sizes and you can return memory back to the pool. The cons of memory pools are that this technique is quite challenging to implement.

What's next?

Sometimes, I envy Python or Java. They use dynamic memory allocation combined with garbage collection and all is fine. All? All the four presented techniques are in use in C or C++ and offer a lot. In the next post, I will have a closer look at the difference between the four techniques. I'm in particular interested in predictability, scalability, internal and external fragmentation, and memory exhaustion.

 

 

Thanks a lot to my Patreon Supporters: Matt Braun, Roman Postanciuc, Tobias Zindl, Marko, G Prvulovic, Reinhold Dröge, Abernitzke, Frank Grimm, Sakib, Broeserl, António Pina, Sergey Agafyin, Андрей Бурмистров, Jake, GS, Lawton Shoemake, Animus24, Jozo Leko, John Breland, Louis St-Amour, Venkat Nandam, Jose Francisco, Douglas Tinkham, Kuchlong Kuchlong, Robert Blanch, Truels Wissneth, Kris Kafka, Mario Luoni, Neil Wang, Friedrich Huber, lennonli, Pramod Tikare Muralidhara, Peter Ware, Daniel Hufschläger, Alessandro Pezzato, Evangelos Denaxas, Bob Perry, Satish Vangipuram, Andi Ireland, Richard Ohnemus, Michael Dunsky, Leo Goodstadt, Eduardo Velasquez, John Wiederhirn, Yacob Cohen-Arazi, Florian Tischler, Robin Furness, Michael Young, Holger Detering, Bernd Mühlhaus, Matthieu Bolt, Stephen Kelley, Kyle Dean, Tusar Palauri, Dmitry Farberov, Ralf Holly, Juan Dent, George Liao, and Daniel Ceperley.

 

Thanks in particular to Jon Hess, Lakshman, Christian Wittenhorst, Sherhy Pyton, Dendi Suhubdy, Sudhakar Belagurusamy, Richard Sargeant, Rusty Fleming, Ralf Abramowitsch, John Nebel, and Mipko.

 

 

My special thanks to Embarcadero CBUIDER STUDIO FINAL ICONS 1024 Small

 

My special thanks to PVS-Studio PVC Logo

 

Mentoring Program

My new mentoring program "Fundamentals for C++ Professionals" starts on the 22nd of April. Get more information here: https://bit.ly/MentoringProgramModernesCpp.

Seminars

I'm happy to give online seminars or face-to-face seminars worldwide. Please call me if you have any questions.

Bookable (Online)

German

Standard Seminars (English/German)

Here is a compilation of my standard seminars. These seminars are only meant to give you a first orientation.

New

Contact Me

Modernes C++,

RainerGrimmSmall

Comments   

0 #1 Rosalyn 2017-07-22 00:12
I conceive this internet site contains some very excellent
information for everyone :D.
Quote
0 #2 Torsten 2017-07-29 03:17
Excellent post. I used to be cheecking constantly this weblog and
I am impressed! Verry helpful info specially tthe remaining phase :)
I care for such inbfo much. I used to be looking for this certain information forr a long time.
Thank you and best of luck.
Quote

My Newest E-Books

Course: Modern C++ Concurrency in Practice

Course: C++ Standard Library including C++14 & C++17

Course: Embedded Programming with Modern C++

Course: Generic Programming (Templates)

Course: C++ Fundamentals for Professionals

Interactive Course: The All-in-One Guide to C++20

Subscribe to the newsletter (+ pdf bundle)

Blog archive

Source Code

Visitors

Today 2355

Yesterday 7881

Week 35010

Month 126164

All 9406187

Currently are 211 guests and no members online

Kubik-Rubik Joomla! Extensions

Latest comments