What hardware configuration advice can you give for optimal performance?
What hardware configuration is required for optimal performance?
- 64 bit operating system
- Microsoft® Windows Vista®, Windows 7, Windows 8 or higher
- If you encounter problems installing from a network drive please contact INTEGRATED Technical Support
- Installation requires approximately 110 MB disk space
- A minimum of 4 GB of RAM is required
- Although the software runs on single processor machines, users may wish to install on a multi processor system in preparation for the additional power that will be available in the next release of the software
- The programs will run with a minimum of 4 GB of RAM but this is not recommended for larger problems in which 12 GB of RAM or more should be used. The more RAM used, the faster larger problems will be solved
- Multi-core processors are strongly recommended as the 3D programs are multi-threaded
INTEGRATED supports software products for 64 bit operating systems
For most of our customers, this announcement has no relevance as most companies have standardized hardware requirements that fit or exceed the 64 bit operating systems required by our company. INTEGRATED’s development tools used for advancements require the installation of 64 bit systems*.
Experience the full power of our simulation tools. Automatically reduce the solution time.
By running the software in a 64 bit system, the system is able to make more effective use of available RAM. The performance of the simulation is greatly affected by the power of the computer in use.
*The suppliers of the software tools we use have this requirement for us as well.
Available RAM versus Problem Size
For small problems the processor speed is the biggest consideration for calculations. If your processor works at twice the speed the problem will be solved in half the time. For larger problems, however, memory management progressively becomes a bigger and bigger consideration. If the memory needed to solve is larger than available RAM - then most of the problem is being swapped back and forth between RAM and the hard disk as the problem proceeds. The efficiency of this process becomes the biggest single factor in the speed of solving large problems. Since this is managed by Windows itself - taking account of other processes also running - we can do very little to help you optimize further from within our software, but can offer the following advice regarding the system setup:
Determine the size of problems you will be solving. This is reported in the Message Area as required disk space when the BEM solver begins. It is also reported for the existing element distribution from the menu Solution>Elements>Problem Size.
The importance of getting as much RAM as needed on a 64 bit system is illustrated by the benchmark results below for a challenging magnetic problem run on 4 different computers:
Comparising of solution times for a nonlinear 3D magnetic model requiring 6 GB memory
The model took 6 hours to solve on a basic system and 3/4 hour to solve on a good system. There are many differences between the 4 systems used, leading to some noise in the plot. However, it is clear that the optimal solution is to use a 64 bit version of the software with more RAM available than the reported memory requirement.
2 hard drives: When choosing hard disk features access time is clearly important. You can set up the locations of the scratch files from Utilities>Settings. Out of various configurations we tested, this was the single most important factor in performing faster analyses when the memory required exceeded available RAM.
RAID ARRAY: using a RAID array lets you use multiple disks as a single drive letter, but will manage the access very efficiently. We configure our own systems such that IES software is installed on d: (a RAID array) with the program and scratch files using d:. For more generic information about configuring a RAID array on your computer, check HOW TO: Establish a Striped Volume (RAID 0) in Windows Server 2003 (Microsoft Knowledge Base).
Last updated: December, 2012