https://github.com/ReidAtcheson/numaallocator

Simple C++11 and beyond allocator to control
https://github.com/ReidAtcheson/numaallocator

Last synced: 3 months ago
JSON representation

Simple C++11 and beyond allocator to control

• Host: GitHub
• URL: https://github.com/ReidAtcheson/numaallocator
• Owner: ReidAtcheson
• Created: 2016-09-21T19:33:24.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2016-09-21T20:50:27.000Z (about 8 years ago)
• Last Synced: 2024-05-03T06:23:20.073Z (6 months ago)
• Language: C++
• Size: 1.95 KB
• Stars: 7
• Watchers: 2
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-numa - NumaAllocator - A simple C++ header NUMA memory allocator (NUMA-aware memory placement and scheduling)

README

        
# NumaAllocator

This *very* simple code is a C++ allocator that gives some control

over where C++ containers allocate memory.

In a multithreaded code it is often enough to rely on the OS

first touch policy to achieve this implicitly, but sometimes

the different memory "nodes" can have wildly

different performance characteristics requiring programmer

to take more control.

#Dependencies

This simple include-only library does depend

on [libnuma](http://oss.sgi.com/projects/libnuma/).

You will need numa.h header file as well as 

a [shared] object file to link against.

# Example

```c++

#include 

#include "numaalloc.h"

int main(int argc,char** argv){

  int n1=0;

  int n2=1;

  int sz=5000;

  typedef NumaAlloc::NumaAlloc na_t;

  //Allocate vector of floats bound to numa node n1

  std::vector x(sz,0.0,na_t(n1));

  //Allocate vector of floats bound to numa node n2

  std::vector y(sz,0.0,na_t(n2));

  //Do something with x,y in parallel for win.

  return 0;

}

```