# MPS working group?

**URL:** <https://dev-discuss.pytorch.org/t/mps-working-group/1910>\
**Category:** hardware-backends\
**Created:** [February 29, 2024, 8:00am UTC](https://dev-discuss.pytorch.org/t/mps-working-group/1910 "2024-02-29T08:00:28Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![lamroger](https://yyz2.discourse-cdn.com/flex036/user_avatar/dev-discuss.pytorch.org/lamroger/32/1688_2.png) [@lamroger](https://dev-discuss.pytorch.org/u/lamroger)\
**Post date:** [February 29, 2024, 8:00am UTC](https://dev-discuss.pytorch.org/t/mps-working-group/1910/1 "2024-02-29T08:00:28Z")

</div>

Hi,

Very exciting developments going on in the mps world. I was wondering if there a working group or something like that to get involved in the latest efforts.

I found this support matrix: [MPS Support Matrix](https://qqaatw.dev/pytorch-mps-ops-coverage/)  
and this README: [MPS Backend · pytorch/pytorch Wiki · GitHub](https://github.com/pytorch/pytorch/wiki/MPS-Backend)

but wondering if there any monthly meeting or anything like that. Would like to get knowledgeable enough to squash bugs

---

<div class="post-metadata">

**Author:** ![arunppsg](https://yyz2.discourse-cdn.com/flex036/user_avatar/dev-discuss.pytorch.org/arunppsg/32/1146_2.png) [@arunppsg](https://dev-discuss.pytorch.org/u/arunppsg)\
**Post date:** [March 1, 2024, 8:20am UTC](https://dev-discuss.pytorch.org/t/mps-working-group/1910/2 "2024-03-01T08:20:57Z")

</div>

Yeah, I am also interested in such a working group for MPS.

---

<div class="post-metadata">

**Author:** ![lamroger](https://yyz2.discourse-cdn.com/flex036/user_avatar/dev-discuss.pytorch.org/lamroger/32/1688_2.png) [@lamroger](https://dev-discuss.pytorch.org/u/lamroger)\
**Post date:** [March 12, 2024, 10:31pm UTC](https://dev-discuss.pytorch.org/t/mps-working-group/1910/3 "2024-03-12T22:31:59Z")

</div>

Wooo! One method I’ve been using is looking at the “good first issues” and “mps” tags to see if there are any approachable bugs to work on. Are you looking at any bugs in particular?

I’m working my way up to understand [addcdiv computes incorrect results on MPS with noncontiguous tensors · Issue #118115 · pytorch/pytorch · GitHub](https://github.com/pytorch/pytorch/issues/118115) and how the striding works to see if it’s something I can fix