Question on Voxel size definition

[Weizhe Fan]

Hi Dr. Fang,

Thank you for making such a powerful tool. I have a question below and would appreciate your help.

I'd like to simulate a very thin layer in MCX, so I first tried defining the voxel size as 0.02mm and see whether the simulation can run. I also change values the tissue region, detector, source and so on, to corresponding voxel unit. Then I ran the simulation, but got the error 'MCX ERROR(-1):invalid argument in unit mcx_core.cu:1981'. 

Can I ask is it because MCX does not support such small voxel size, or is it because there are just too many voxels in total (I have 3000^3=2.7e10)?

I have also attached my setting here. Thank you for your time!

Regards,

Weizhe

[Qianqian Fang]

just for the purpose of storing the fluence output, mcx needs to allocates 8-byte per voxel per time gate. based on your domain size, you will need at least 54GB of memory for storing fluence alone, not mention other buffers needed for the simulation. 

the question is - does your GPU have 54GB of graphics memory? similarly, what is the size of your host/CPU memory?

please also read the following FAQ item for new mcx users

https://mcx.space/wiki/index.cgi?Doc/FAQ#I_am_a_beginner_to_MCX_what_is_your_advice_before_I_start

especially the section "Brute-force is never a good solution."

for the output memory requirements, please also read this section in the README

https://github.com/fangq/mcx?tab=readme-ov-file#requirement-and-installation

On 3/22/26 16:40, Weizhe Fan wrote:

You don't often get email from fweiz...@gmail.com. Learn why this is important

--
You received this message because you are subscribed to the Google Groups "mcx-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mcx-users+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/mcx-users/84074911-3375-44d5-b922-482f423c9cebn%40googlegroups.com.

[Qianqian Fang]

sorry, I miscalculated - you will need 216 GB GPU memory = 3000^3*8 bytes, which is not physically possible.

[Weizhe Fan]

Thank you for the information. I need to design my simulation in a better way.

Qianqian Fang <q.f...@neu.edu> 于2026年3月23日周一 17:48写道：