Skip to content

Conversation

@vlad-perevezentsev
Copy link
Contributor

@vlad-perevezentsev vlad-perevezentsev commented Jun 3, 2024

This PR suggests adding syevd_batch and heevd_batch implementation as a separate function and provide iteration of the batch array within it to avoid allocation of each 2D array before calling syevd and heevd to improve performance and avoid serialization of all host tasks in the queue in MKL call;

Performance of dpnp.eigh on Iris Xe
image

  • Have you provided a meaningful PR description?
  • Have you added a test, reproducer or referred to issue with a reproducer?
  • Have you tested your changes locally for CPU and GPU devices?
  • Have you made sure that new changes do not introduce compiler warnings?
  • Have you checked performance impact of proposed changes?
  • If this PR is a work in progress, are you filing the PR as a draft?

@github-actions
Copy link
Contributor

github-actions bot commented Jun 3, 2024

View rendered docs @ https://intelpython.github.io/dpnp/pull/1867/index.html

@vlad-perevezentsev vlad-perevezentsev force-pushed the impl_syevd_batch_via_syevd branch from 0a3d67b to 671bb2c Compare June 4, 2024 14:31
@vlad-perevezentsev vlad-perevezentsev changed the title Implement syevd_batch via syevd call Implement syevd_batch/heevd_batch via syevd/heevd call Jun 4, 2024
@vlad-perevezentsev vlad-perevezentsev marked this pull request as ready for review June 4, 2024 18:00
@vlad-perevezentsev vlad-perevezentsev self-assigned this Jun 4, 2024
@vlad-perevezentsev vlad-perevezentsev changed the title Implement syevd_batch/heevd_batch via syevd/heevd call dublicate: Implement syevd_batch/heevd_batch via syevd/heevd call Jul 19, 2024
@vlad-perevezentsev
Copy link
Contributor Author

Work continued in #1936

@antonwolfy antonwolfy deleted the impl_syevd_batch_via_syevd branch May 14, 2025 16:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants