# Estimating memory required for NNinteractive

**URL:** <https://discourse.slicer.org/t/estimating-memory-required-for-nninteractive/43239>\
**Category:** Support\
**Created:** [June 5, 2025, 5:54pm UTC](https://discourse.slicer.org/t/estimating-memory-required-for-nninteractive/43239 "2025-06-05T17:54:50Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![muratmaga](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/muratmaga/32/3622_2.png) [@muratmaga](https://discourse.slicer.org/u/muratmaga)\
**Post date:** [June 5, 2025, 5:54pm UTC](https://discourse.slicer.org/t/estimating-memory-required-for-nninteractive/43239/1 "2025-06-05T17:54:50Z")

</div>

Is there a way to predictably calculate the memory required to segment a volume using NNinteractive? For example given the 400x400x400 16 bit volume how much GPU memory would I need. I

Trying to figure out via trial and experiment is a fairly frustrating experience, as you have to wait until it fails…

---

<div class="post-metadata">

**Author:** ![pieper](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/pieper/32/8_2.png) [@pieper](https://discourse.slicer.org/u/pieper)\
**Post date:** [June 5, 2025, 6:12pm UTC](https://discourse.slicer.org/t/estimating-memory-required-for-nninteractive/43239/2 "2025-06-05T18:12:34Z")

</div>

I was running with [nvtop](https://github.com/Syllo/nvtop) installed and seeing how high the bar gets for different scenarios. This could give you a feel for the size vs memory tradeoffs but I’m not sure if there are other factors involved (like the autozoom).

---

<div class="post-metadata">

**Author:** ![muratmaga](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/muratmaga/32/3622_2.png) [@muratmaga](https://discourse.slicer.org/u/muratmaga)\
**Post date:** [June 5, 2025, 9:19pm UTC](https://discourse.slicer.org/t/estimating-memory-required-for-nninteractive/43239/3 "2025-06-05T21:19:02Z")

</div>

I do that but I was hoping some sort of a deterministic method I can use to decide what’s the largest volume chunk I can fit in, in one go. Our segmentations are too big as is, so I am trying to figure out how to crop them into managable regions (4 or 8 or 16 ROIs, the fewer the better from data management purpose).