Choosing an OpenSPIM Computer

Greetings all,

To preface, I am pretty new to microscopy and optics. I am currently working on an OpenSPIM setup and am curious what the key considerations are for selecting a computer to run this kind of setup (minimum processor specs, storage recommendations, cable connections, operating system, etc.).

I would greatly appreciate any guidance and wisdom you can share in this area.

Thank you in advance for your help.

Best,
Joseph

3 Likes

Hi Joseph

I am not knowledgeable enough about acquisition to make any recommendations. Though I assume transfer and storage speed will be important. Also some acquisition software only works on specific operating systems. Hopefully someone knowledgeable about acquisition will respond.

If you are running post processing software you will want to understand if the software runs mostly on the GPU or CPU or both. Personally I’ve found a fairly new i7 based processor and a new NVIDIA (RTX) card are impressive and “somewhat” cost efficient (you can probably get a good gaming machine with both for apr. $2000 USD). Of course “cost efficient” is relative.

If I need more firepower than a gaming machine can provide I use AWS virtual machines. Though there is a cost to this, and it takes time to transfer the data.

I concede my answer is a bit vague and fuzzy… Hopefully others respond. I’m interested myself in hearing people’s strategies for putting together a good Microscopy setup.

Brian

1 Like

Hi there!

Minimum requirements is an entry level desktop. Acquisition is fast and saved for each plane, last I remember, so the acquisition machine needs very little processing power as long as it has a good usb3 for the camera(s).
But it will really all depend on processing if you want to work on SPIM data on that machine, and also storage if you’re planning many many experiments.

I’d suggest you answer some of the questions below so that me and others can help guide you better.

How large are the samples you want to acquire? How often will you be imaging? Where do you wish to store the acquired data?

Depending on the answers, it could be that a good high end desktop PC with some extra storage is enough, or perhaps you need a beast of a machine linked to a NAS via 10Gb ethernet.

Looking forward to hearing from you!

5 Likes

I apologize for the long delay in my response. Here are my answers to the clarifying questions asked. These are pretty rough estimates.

On the large/fast end of things, I could see us collecting 5-10 images per second for perhaps 5 minutes. More likely we will be taking 1 image every couple of seconds over 5 to 10 minutes. At the fastest, 15~20 images per second for around a minute.

We will probably be running an experiment about once a week.

It would be most convenient to have all data storage and processing done on the computer that is with the scope. However, it would not be difficult to add an external storage device for additional space or backup.

Budget wise, I am hoping that a high end desktop PC with extra storage will be enough. We do not need anything crazy fast as far as data processing speeds go, but having it take an afternoon or a day instead of a week would be nice.

Again, these are rough estimates from someone who is totally new to this field. Please let me know if there is anything else I can clarify here.

Your patience and expertise is much appreciated.

Thanks,
Joseph

OK so 10 images per second over 10 minutes for 3 channels (in order to get some upper bounds)

10 \times 60 \times 10 \times 3 = 18000 images

So 18000 images at with a 2048x2048 dexel sCMOS camera (~ 8MB) per image is about 18GB of raw per dataset (really upper bound).

If you want to store (permanently?) and process on the machine then you would benefit from having a fast SSD (2TB) for the processing and a storage solution with some redundancy (RAID 5 for example made of 3x 8TB disks).

The processing part will benefit from having a good amount of RAM, so I’d say nothing below 32GB (with potential to expand) seems reasonable.

Hope this helps, these are my two cents and I look forward to comments from others :slight_smile:

Oli