# Linux fs recommendations

**URL:** <https://community.mrtrix.org/t/linux-fs-recommendations/891>\
**Category:** Uncategorized\
**Tags:** linux, performance\
**Created:** [May 12, 2017, 9:10am UTC](https://community.mrtrix.org/t/linux-fs-recommendations/891 "2017-05-12T09:10:47Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![maedoc](https://community.mrtrix.org/user_avatar/community.mrtrix.org/maedoc/32/53_2.png) [@maedoc](https://community.mrtrix.org/u/maedoc)\
**Post date:** [May 12, 2017, 9:10am UTC](https://community.mrtrix.org/t/linux-fs-recommendations/891/1 "2017-05-12T09:10:50Z")

</div>

Hi

I’m puzzling through a filesystem related performance issue on a new Ubuntu Dell SSD workstation, which is showing abysmal performance for some I/O intensive steps. For example, using the `-stride` option on `mrconvert` on an HCP diffusion dataset takes half an hour, while less than a minute on a MacBookPro (the latter has an nvme disk, but that doesn’t account for a 30x difference).

`mrconvert` reports (with `-debug`) that it’s memory-mapping the output file, and `using indirect IO`. `iostat` reports many threads saturating write bandwidth on the disk.

So, I’m left poking around in filesystem options (I’m using ext4 mounted with `noatime,date=writeback`), and I’m wondering if you have some recommendations here. OTOH if this sounds strange, maybe I need to try a newer or non-Dell kernel.

Thanks,  
Marmaduke

---

<div class="post-metadata">

**Author:** ![maedoc](https://community.mrtrix.org/user_avatar/community.mrtrix.org/maedoc/32/53_2.png) [@maedoc](https://community.mrtrix.org/u/maedoc)\
**Post date:** [May 12, 2017, 10:21am UTC](https://community.mrtrix.org/t/linux-fs-recommendations/891/2 "2017-05-12T10:21:21Z")

</div>

I guess I’m answering my question, but it looks like an aberrant kernel or linux thing: I popped in a vanilla CentOS 7 Live CD and the strided conversion runs faster than the MacBookPro now.

Hopefully this thread will at least be useful if others run into similar issues.

---

<div class="post-metadata">

**Author:** ![maxpietsch](https://community.mrtrix.org/user_avatar/community.mrtrix.org/maxpietsch/32/25_2.png) [@maxpietsch](https://community.mrtrix.org/u/maxpietsch)\
**Post date:** [May 12, 2017, 10:42am UTC](https://community.mrtrix.org/t/linux-fs-recommendations/891/3 "2017-05-12T10:42:14Z")

</div>

Thanks for the info! We have had issues on Ubuntu 14.04 which can be mostly solved by tweaking the [buffer cache](https://mrtrix.readthedocs.io/en/latest/troubleshooting/FAQ.html#linux-very-slow-performance-when-writing-large-images) settings.

---

<div class="post-metadata">

**Author:** ![jdtournier](https://community.mrtrix.org/user_avatar/community.mrtrix.org/jdtournier/32/2594_2.png) [@jdtournier](https://community.mrtrix.org/u/jdtournier)\
**Post date:** [May 16, 2017, 12:27pm UTC](https://community.mrtrix.org/t/linux-fs-recommendations/891/4 "2017-05-16T12:27:41Z")

</div>

Just thought I’d add that I’ve been looking into these issues recently, there are ways to avoid these issues. But it’ll require a fair bit of work to implement, it may take a while for me to find the time to work on this…
