Avizo and Amira are two separate packages that are used for similar but distinct purposes.
Avizo is a commercial data visualization and analysis tool that supports 2D and 3D visualizations in areas such as materials research, tomography, non-destructive testing and more.
Amira, on the other hand, works in the same domain but is one of the most popular packages to deal with medical imaging data in the world. It’s application also extend to fields such as quantum physics and materials science.
The main determinant of how much horsepower you’ll need to run either of these packages is highly dependent on the size and complexity of your data set.
The software can be hard on all four main computer components: GPU, CPU, RAM and secondary storage such as HDD or SSD.
Of these four components, the GPU is by far the most important. Due to the nature of the sorts of computing the software performs it’s ideally suited to GPU parallel processing.
Both packages will run on any GPU that supports OpenGL 2.1 and up, but the memory size is critical and may need to be very large for professional workloads.
The good news is that the latest generation of consumer cards such as the GTX 1080 and Titan X (Pascal) have significantly more VRAM than previous generations. It’s now quite normal to have more than six gigabytes of VRAM and as much as twelve GB on consumer cards. When we look at professional GPUs such as the Quadro range we see VRAM counts over 20GB, but unless you have a specific reason to believe your dataset will require this much it’s unlikely you’ll need that amount of RAM.
Quadro cards give a better assurance of reliability and driver compatibility, but for a much lower asking price one can have the same performance. Therefore, unless reliability is absolutely mission critical, we recommend a high-end consumer card such as the GTX 1080 or the Titan X
For certain types of processing multiple-GPUs can be employed as well, which may also support a decision to have, for example, two Titan X cards as opposed to a single Quadro card.
These applications are true RAM hogs. Luckily the developer has provided us with a simple rule of thumb. Complex data sets require RAM six to eight times as large. So if you have a 4GB data set 32GB of RAM would be a good idea.
This is pretty straightforward. These are massive data streams and the faster the drive channeling all that info, the less likely you are to have a system-slowing bottleneck. We recommend a fast NVMe SSD to really give the other components the room to stretch their legs.
We also have great guidelines from the developer on CPU requirements. The CPU-dependent components of the software is generally pretty well multi-threaded. However returns diminish once you go past eight cores. At this point clock frequency and cache have a bigger effect. In other words having a dual-socket system with two fat Xeon quad cores or one eight core Xeon with lots of cache and as much frequency as you can squeeze out of it is optimal.
Basically AMIRA / AVIZO needs the fastest clock speed in about 8 cores. As far as a dual CPU system, the only recommended CPU option is 2x Xeon Gold 6136 or 6154 which can reach a top speed of 3.7Ghz.
Operating systems supported by AVIZO / AMIRA
Linux x86 64 (64-bit). Supported 64-bit architecture is Intel64/AMD64 architecture. Supported Linux distribution is
Red Hat Enterprise Linux 6.