Section 2 - Hardware and Platform Configurations

Linux vs. NT

Q) How does OrthoVista performance compare between Linux and NT on the same PC hardware?
A) The following was reported by an OrthoVista customer "We are running OrthoVista on Linux. (This is because when we did some initial performance testing we found a quite significant performance increase over NT on exactly the same hardware with the same data set.)"

Recommended Configuration

Q: What is a hardware configuration should I consider for running OrthoVista?
A: In general, the hardware configuration should include: 1) enough (preferably local) disk space to store input and output data; 2) enough memory to run typically sized projects; 3) fast enough processor.
To determine the disk space approximately, multiply the project input image data size (number of images times the size of each image) by a factor of 2 for typical mosaic production. If you also intend to process and output individual images in addition to mosaics, use a factor of 3. In general, it is a good idea to budget for 10-20% free disk space (e.g. for special projects, etc.). For example, if you routinely work with project areas consisting of 100 images of 250 MB images (e.g. 1000 dpi color scan), you should consider a 50-75 GB storage system. For best performance, a local RAID system is recommended. In general locally connected storage systems provide better I/O Performance than networked storage. See also Hardware I/O Speed.
The amount of memory should be "enough" - in particular, it must be large enough to hold the OrthoVista cache (without swapping) and also hold dynamically allocated data structures (which can be on the order of a couple MB per input image). See Setting Cache Size for discussion on caching requirements. Experience has shown that reasonably large production projects (e.g. order of 10s to 100 images of 100-300 MB each) can be performed with about 512 MB of RAM. However, for larger projects (or larger images), or if you are working at the higher end of this example range, you may want to consider using 768 MB or 1 GB of RAM.
In general, OrthoVista is CPU bound so that you can almost always benefit from faster processor speed. Therefore, the relevant CPU recommendation is "as fast as affordable" - after stuffing the system with memory. Note that increasing CPU speed is a "performance optimization" whereas memory size is an "enabling resource" for processing large project data sets. See also Hardware CPU Speed.
Q: Which Operating System is best for running OrthoVista
A: OrthoVista is available for several hardware/operating system combinations. In general, the "best" choice for the operating system is the one with which you are most comfortable.

SGI/IRIX

Q) The IRIX R4000 version is not running after installation. We get error messages like the following when we try to start OrthoVista.
Fatal Error: Cannot Successfully Map soname 'libm.so' under any of the filenames ...(list of library files)...
A) Likely cause is that system does not have Irix N32 libraries installed. Install the N32 dynamic link libraries.

Multi-Processor Support

Q) Is OrthoVista optimized for multiple processors?
A) No. However, multiple copies of OrthoVista may be run simultaneously on multiprocessor systems. To do this, note that each running version must have enough system resources (e.g. memory) available to it independent of the other programs.
Q) How is the OrthoVista software licensing affected by running it on a dual CPU machine?
A) Licensing is tied to the network card hardware address. Therefore, the same license supports all processors on that machine.
Q) If running two sessions of OrthoVista, is it advisable to cut the amount of image cache in half for each so that they each get enough system resources?
A) NO. Memory use must be considered per process - independent of any other process. Each process must have enough cache space to hold its own data (e.g. multiple copies do NOT share cache space). Each process must also have enough available heap space (See Setting Cache Size) for the job it is doing.