Support device awareness of result_type and promote_types

[Marc-Jindra]

Due Diligence

• General:
  • title of the PR is suitable to appear in the Release Notes
• Implementation:
  • unit tests: all split configurations tested
  • unit tests: multiple dtypes tested
  • benchmarks: created for new functionality
  • benchmarks: performance improved or maintained
  • documentation updated where needed

Description

Apple MPS don't support double precision floats or complex numbers, so this PR introduces some checks for that in result_type and promote_types.

Issue/s resolved: #1820

Changes proposed:

• result_type now checks whether an arrays device is mps and changes the dtype accordingly (float64 -> float32, complex128 -> complex64)
• promote_types now takes an optional device argument and checks if that device is mps

Type of change

Bug fix (non-breaking change which fixes an issue)

Memory requirements

Performance

Does this change modify the behaviour of other functions? If so, which?

no

[github-actions]

Thank you for the PR!

[codecov]

Codecov Report

Attention: Patch coverage is 55.55556% with 12 lines in your changes missing coverage. Please review.

Project coverage is 88.95%. Comparing base (e12b24b) to head (fe3b7c3).

Files with missing lines	Patch %	Lines
heat/core/types.py	53.84%	12 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1836      +/-   ##
==========================================
- Coverage   92.10%   88.95%   -3.16%     
==========================================
  Files          86       86              
  Lines       13144    13159      +15     
==========================================
- Hits        12106    11705     -401     
- Misses       1038     1454     +416     

Flag	Coverage Δ
unit	88.95% <55.55%> (-3.16%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

Thank you for the PR!

[JuanPedroGHM]

There are still a couple of instances where references to MPS need to be removed from the codebase. Some quick examples that I found:

• _operations.py: _binary_op and _cum_op
• arithmetics.py: hypot and hypot_

There might be more, but those are some I could find quickly that already use promote_type or result_type. The references to MPS should be removable.

Otherwise, great code. Thanks!