Skip to content

Random Memory Benchmark for memory power modelling (originally developed by http://www.roylongbottom.org.uk/)

Notifications You must be signed in to change notification settings

greenlsi/randmem

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Random Memory Benchmark Readme

RandMem is a random memory benchmark originally developed by Roy Longbottom (http://www.roylongbottom.org.uk/). This benchmark has been modified and some features have been added to it in order to develop some memory power/energy models for enterprise servers.

The bechmark supports various architectures (Linux 32 bit, Linux 64bit and Solaris). It has been tested on Ubuntu 32bit, CentOS 6.4 and Solaris 10.

Feel free to contribute to it, but bear in mind that the original author would like any money ever made from this benchmark go to charity.

Installation instructions

To install the benchmark, just run the makefile:

For solaris systems $ make solaris

For 32 bit systems $ make 32bit

For 64bit systems (default) $ make all

Once compiled, just run ./randmem.out If you need some help: $ ./randmem.out --help

More about RandMem

RandMem can be used to stress the system memory in some different ways. The benchmark executes with the following parameters:

  • pattern: test to run, can be serial or random
  • data: type of data to run, can be integer or intfp
  • access: type of access to memory, can be r or rw
  • memsize: memory size in GB to use. Note that the maximum amount of per-process memory in a Linux/Solaris server is 2GB
  • blocksize: maximum size of each memory block (in KB). This indicates the malloc size of the benchmark
  • runtime: minimum time for benchmark to run (in secs)
  • numexecs: number of times the benchmark is executed

On startup, the benchmark will try to allocate as much memory as asked by the user ('memsize') in chunks of size 'blocksize'. If it cannot, it will allocate the maximum amount of memory possible.

If you need more than 2GB of memory, launch multiple instances of randmem

About

Random Memory Benchmark for memory power modelling (originally developed by http://www.roylongbottom.org.uk/)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published