Hi Tomas,
I’ll just give my opinion (someone with more hardware knowledge could still respond). A ZRD file is a binary file that has to be read sequentially, segment by segment. This is because to efficiently pack the ZRD file, after the header section of the ZRD file with version number and maximum allowed segments, each individual segment is defined by a 4-byte integer storing the number of segments for the current ray and then a 232-byte struct. You don’t know where the next ray starts in the binary file until all the segments are read from the previous ray.
With that said, the 3 main impacts for ZRD speed will be:
- CPU clock speed (faster speed is quicker read/write of ZRD)
- Amount of RAM (should make sure RAM is 2x-4x larger than maximum ZRD file size, otherwise memory paging and writing to disk will happen a lot)
The things that would not affect the speed of reading ZRD file:
*The Number of CPUs will affect the NSC Ray Trace, so it will drastically speed up the creation of a ZRD file since OpticStudio can chunk the number of rays each CPU/thread will trace. But once the ZRD is created, it has to be read sequentially on a single thread.
There are additional recommendations in the following articles:
https://support.zemax.com/hc/en-us/articles/1500005575042-What-computer-should-I-buy-to-run-OpticStudio
https://community.zemax.com/got-a-question-7/new-computer-performance-advice-3550