# Best pipeline for large MultiBlockDataset

**URL:** https://discourse.vtk.org/t/best-pipeline-for-large-multiblockdataset/8998
**Category:** Support
**Created:** [July 23, 2022, 7:51am UTC](https://discourse.vtk.org/t/best-pipeline-for-large-multiblockdataset/8998 "2022-07-23T07:51:53Z")
**Posts on this page:** 3
**Page:** 1

<div class="post-metadata">

### Author: ![BotellaA](https://discourse.vtk.org/user_avatar/discourse.vtk.org/botellaa/32/1243_2.png) [@BotellaA](https://discourse.vtk.org/u/BotellaA)
#### Post date: [July 23, 2022, 7:51am UTC](https://discourse.vtk.org/t/best-pipeline-for-large-multiblockdataset/8998/1 "2022-07-23T07:51:53Z")

</div>

Hello,

I am starting to use a MultuBlockDataset. It can be composed of hundreds of point sets, lines, surfaces, and unstructured grids. I struggle to understand what pipeline I should use for this kind of data. Should I merge everything into a single PolyData, to avoid too many calls to the GPU? Should I use CompositeMapper directly?

Thanks a lot for your help

---

<div class="post-metadata">

### Author: ![Paulo\_Carvalho](https://discourse.vtk.org/user_avatar/discourse.vtk.org/paulo_carvalho/32/370_2.png) [@Paulo\_Carvalho](https://discourse.vtk.org/u/Paulo_Carvalho)
#### Post date: [July 25, 2022, 11:43am UTC](https://discourse.vtk.org/t/best-pipeline-for-large-multiblockdataset/8998/2 "2022-07-25T11:43:48Z")

</div>

Hello,

My two cents: based on what people said throughout the years, having too many actors is a bad idea. So, yes, those approaches of yours should do.

regards,

Paulo

---

<div class="post-metadata">

### Author: ![BotellaA](https://discourse.vtk.org/user_avatar/discourse.vtk.org/botellaa/32/1243_2.png) [@BotellaA](https://discourse.vtk.org/u/BotellaA)
#### Post date: [July 27, 2022, 2:44pm UTC](https://discourse.vtk.org/t/best-pipeline-for-large-multiblockdataset/8998/3 "2022-07-27T14:44:11Z")

</div>

Ok, so it’s more about the number of actors than the number of mappers. I was wrong about that.

Thanks