Bugfix: _like factories now default to same device as input DNDarray

[mrfh92]

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
  • documentation updated where needed

Description

Issue/s resolved: #1464

Changes proposed:

Edited by @ClaudiaComito on Apr 30

• changed default device in factories.__factory_like(a), now device defaults to a.device unless explicitly set
• addressed pre-commit complaints
• removed unnecessary/legacy numpy() calls from test_random to facilitate testing on AMD runner

Type of change

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

Memory requirements

NA

Performance

test_random now takes a bit longer as it's running statistical functions on distributed DNDarrays instead of on the gathered numpy arrays

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

no

[mrfh92]

First bug fix for v1.4 (unfortunately we need to release this as it makes problems in one of our benchmarks)

background: this can't be found in our tests as HEAT_TEST_USE_DEVICE=gpu seems to set the default for devices to "gpu", whereas in the standard mode we have default = "cpu" at many places ... therefore, errors of this kind are only found when working with device="gpu" manually in a code

[github-actions]

Thank you for the PR!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 91.58%. Comparing base (f465ef0) to head (91f0b4b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1443      +/-   ##
==========================================
- Coverage   91.58%   91.58%   -0.01%     
==========================================
  Files          80       80              
  Lines       11647    11652       +5     
==========================================
+ Hits        10667    10671       +4     
- Misses        980      981       +1     

Flag	Coverage Δ
unit	91.58% <100.00%> (-0.01%)	⬇️

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.

[github-actions]

Thank you for the PR!

[github-actions]

Thank you for the PR!

[ClaudiaComito]

Error in random module in AMD runner, similar but not same as #1289

[github-actions]

Thank you for the PR!

[ClaudiaComito]

@mrfh92 it should work now. I don't know why pre-commit fails.

[github-actions]

Thank you for the PR!

[github-actions]

Thank you for the PR!

[github-actions]

Thank you for the PR!

[github-actions]

Thank you for the PR!

[github-actions]

Thank you for the PR!

[github-actions]

Thank you for the PR!

[github-actions]

Successfully created backport PR for release/1.4.x:

• [Backport release/1.4.x] Bugfix: _like factories now default to same device as input DNDarray  #1472