# Set dtype to np.float32 and memory conservation

**URL:** <https://discourse.scverse.org/t/set-dtype-to-np-float32-and-memory-conservation/1734>\
**Category:** scanpy\
**Tags:** scvi\
**Created:** [September 1, 2023, 1:34am UTC](https://discourse.scverse.org/t/set-dtype-to-np-float32-and-memory-conservation/1734 "2023-09-01T01:34:21Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![malonzm1](https://avatars.discourse-cdn.com/v4/letter/m/9de0a6/32.png) [@malonzm1](https://discourse.scverse.org/u/malonzm1)\
**Post date:** [September 1, 2023, 1:34am UTC](https://discourse.scverse.org/t/set-dtype-to-np-float32-and-memory-conservation/1734/1 "2023-09-01T01:34:21Z")

</div>

Hi,

I am trying to analyze a huge scRNA-seq dataset. When I concatenate the datasets I get a memory low error even though I’m already using 1024Gb. As a way of conserving memory, I was wondering if I can set the datatype to float32 instead of float64. How can I do this with sc.read and concatenate functions? Also, are there other memory conservation steps I can use?

Thanks and good day.