How can I optimize my hardware and system setup for PolySpace products?

4 views (last 30 days)
I would like to know what hardware, platform and operating system I should use to maximize the performance of PolySpace.

Accepted Answer

MathWorks Support Team
MathWorks Support Team on 24 Jan 2012
The recommended hardware and supported operating systems for Polyspace can be found on the Polyspace System Requirements page and in the product documentation at the following pages:
<http://www.mathworks.com/support/sysreq/current_release/polyspace/>
<http://www.mathworks.com/access/helpdesk/help/toolbox/polyspace/polyspace.html>
Here are answers to some particular questions:
1. What is the maximum CPU speed and RAM that can be utilized by Polyspace?
We cannot specify either a maximum CPU speed or RAM size at which Polyspace performance saturates. The recommended CPU speed and RAM size for machines running PolySpace Server 5.1 are 3.2 GHz and 4.0 GB respectively. The recommended SWAP space is at least twice the RAM size.
2. Is it better to have a slower dual core CPU or a faster single core CPU?
Starting R2009a, Polyspace supports multi-core capabilities for C and C++ languages. The performance gain is linked to the number of cores on the machine. The recommended number of cores is four and the recommended amount of RAM is at least 2 GB per core. Please note that beyond 4 cores, the gain is not so important.
3. What is the recommanded hard disk type and capacity? And what are the working folders of Polyspace?
The disk usage can be pretty intensive during complex verifications. This is why a server should have a high speed disk drive.
Hence, we recommend to have a SSD (solid-state disk). As they are costly, it is better to have two disks actually: one SSD for the folders frequently accessed by Polyspace, and one HDD to store the data once they have been calculated (a HDD).
The SSD capacity should be around 150Gb.
temporary folder of the OS:
Windows : %TEMP%
Linux : \tmp
the 'datadir' folder
Windows: c:\Polyspace\Polyspace_RLDatas
Linux:
On a UNIX system, the directory location containing locks and datadir by default is /var/Polyspace.
4. What is the recommended Operating System for Polyspace Server?
We do not recommend a particular operating system over another. The list of supported operating systems for the current release of Polyspace is available at the MathWorks support website at the above link.
5. Why is Red Hat 9 no longer officially supported by Polyspace Verifier?
Red Hat 9 is no longer supported principally due to the incompatibility of drivers. Officially this indicates that we do not perform strong Verification and Validation on this OS and do not guarantee performance. However, it is still possible to install and run Polyspace on Red Hat 9 by launching Polyspace with the allow-unsupported-linux tag.
6. Why does the installation manual recommend a Linux platform for the Polyspace server?
The installation manual for Polyspace recommends a Linux system for the Server because the software is native under Linux. Other advantages include better memory management and faster speed.
7. Would there be any performance gains by using a 64-bit version of Linux?
Polyspace can be installed and is supported on certain 64 bit Linux platforms. If the machine has 6 GB of memory or less, it will run in 32 bit mode by default. In this case, the performance gain from running Polyspace on a 64 bit system is the performance gain given by the OS. The 64 bit mode can be enabled by specifying the following option:
-machine-architecture 64
Running Polyspace in 64 bit mode is mostly recommended for situations in which a 32 bit mode verification fails due to insufficient memory. In a 64 bit architecture, the CPU requires twice as much memory to manipulate data as a 32 bit counterpart; the performance gain may thus become overshadowed by the higher memory usage.
If the amount of memory is greater than 6 GB, PolySpace will automatically run in 64 bit mode. The option '-machine-architecture 32' can be used to force the 32 bit mode.
Generally speaking, the priority for the different things to consider is:
1) High frequency CPU (4 cores) and fast RAM with the latest chipset
2) Latest OS that will fully benefit of the CPU
3) Amount of RAM (useful to prevent the usage of the hard disk/SSD)
4) Speed of disk
Example of configuration:
  • CPU : 4 cores, x64, Chipset Intel i7-2600, 3.4 Ghz
  • RAM : 8 Gb
  • OS : Linux RHEL 6.0

More Answers (0)

Products


Release

No release entered yet.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!