Recommended hardware configurations
Mistika Workflows is designed to run on even the smallest computers, but depending on the usage it may also use all the resources available to it (especially when used for heavy transcoding). The following are the important considerations:
RAM: 8GB minimum. When transcoding complicated formats at high resolutions 32GB is recommended, or even more RAM if the server has many cores and you plan to run more processes in parallel.
Transcoding uncompressed formats (such as DPX, EXR uncompressed, or Tiff16): use storage devices which are as fast as possible, and try to use local intermediate copies rather than cloud storage directly.
Compressed formats (such as EXR compressed, H264/H265, R3D, or ProRes): these formats depend mainly on the system's CPUs, although in some cases the GPU is also used. We recommend using the OS Task Manager or similar tools to discover where hardware bottlenecks exist for your particular case.
Minimum Configuration Required for Mistika Workflows:
This is the absolute minimum:
- Number of processors: 1
- Number of cores: 2
- RAM: 8 GB
- Display resolution: A minimum of 1920x1080 is recommended for the GUI. 1920x1200 is a better option.
- Graphics card model: NVIDIA is recommended as it is the certified manufacturer for Mistika Workflows. However, modern AMD boards and Intel Iris are also working without presenting problems at the time of writing (except for encoding to h264/h265 as explained below). Please note that many Mistika Workflows tasks are image processing oriented , and for this reason Mistika Workflows will not run without having a discreete GPU installed.
Supported OS distributions
- Windows: Windows 10 (64bit).
- Mac: Sierra, High Sierra, and Mojave (Operating system versions prior to macOS Sierra are known to crash the application).
- Linux: CentOS 7.xxx / RedHat 7.xxx. An automatic installer is provided for CentOS 7.xxx and compatible distributions. Other distributions may also work, but the installation process can be complicated and they are currently unsupported. (You may need to solve package dependencies, daemon differences and typical linux issues...). However this document is intended for the first version of Mistika Workflows, and we expect to add official support for more linux distributions in forthcoming versions.
Extended recommendations for selecting optimal hardware:
- GPU: Part of the Mistika Workflows's processing is cARRIed out by the GPU: Geometry adjustments, Color adjustments, NVidia transcoder, etc., so this is an important component. A modern NVidia board (GeForce or Quadro) with high specs is recommended.
- CPUs: The processes that are cARRIed out by CPUs are the decoding of input codecs and encoding of rendered codecs, but there are important exceptions that use the GPU. The GPU plays an important role in the decoding of R3D formats, and the encoding to NVidia H264/H265 codec. (For those cases the GPU must be NVidia, and the computer must be Windows or Linux, as Apple does not currently support NVidia encoder). The GPU is also used for debayering some camera raw codecs like ARRI, although for these cases it does not need to be NVidia.
If you want to take advantage of the NVidia transcoder we recommend a Pascal generation or later NVidia model, as older models can not be used for hardware encoding or only at low resolutions (For more details see NVIDIA VIDEO CODEC SDK).
CPU based tasks of common usage: Apple ProRes. EXR compressed (DWA, ZIP, PIZ). J2K (JPEG2000). JPEG. Compressed tar (zip/tgz). MD5 checksums. H264/H25 decoding.
Storage and network bandwidth: Uncompressed formats do not require either significant GPU or CPU, but they do need a great deal of disk speed and fast networks!
Therefore, each case depends on its particular circumstances. Workflows with lots of inputs based on compressed codecs and heavy CPU encoding to ProRes may take advantage of having many CPU cores at high clock speeds. While other cases can be well served with just a decent GeForce.
Mistika Workflows is not multGPU, so extra speed is cannot be expected by having two or more GPUs.
The following is another recommendation: in general, the highest end models of CPUs and GPUs offer only a little more speed than marginally inferior models, but often cost much more. At the highest end, a typical figure might be to pay up to 100% more for just 25% more speed, compared with cost-efficient hardware in the sweet spot. For this reason, in general, it may be better to use two computers with good, but not necessarily top of the range, specs rather than one only computer with top specs. Workflows processing can be easily distributed between several computers by using “Watch folder” tasks or other communication tools.