Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

RPP Warp Perspective on HOST and HIP #451

Open
wants to merge 98 commits into
base: develop
Choose a base branch
from

Conversation

r-abishek
Copy link
Member

@r-abishek r-abishek commented Sep 18, 2024

  • Updates version to RPP 1.9.6
  • Adds tensor implementation for Warp Perspective on HOST and HIP
  • Adds support for U8/F16/F32/I8 bit-depths with NHWC/NCHW and layout toggle support
  • Adds relevant unit/perf tests

Srihari-mcw and others added 30 commits August 8, 2024 08:31
Copy link
Contributor

@rrawther rrawther left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added review comments

include/rppdefs.h Show resolved Hide resolved
src/include/cpu/rpp_cpu_simd.hpp Show resolved Hide resolved
bool kernelSizeCase = (testCase == 49);
bool dualInputCase = (testCase == 2 || testCase == 30 || testCase == 33 || testCase == 61 || testCase == 63 || testCase == 65 || testCase == 68);
bool randomOutputCase = (testCase == 6 || testCase == 8 || testCase == 84);
bool nonQACase = (testCase == 24);
bool interpolationTypeCase = (testCase == 21 || testCase == 23 || testCase == 24 || testCase == 79);
bool nonQACase = (testCase == 24 || testCase == 28);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why this is added to nonQA

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Warp Affine and Perspective alone have some very minor precision related differences that don't match exactly. We's spent considerable cycles trying to match that minor difference, but its something to do with the _mm256_setr_ps instruction on computed float location indexes that differs from rawC. @HazarathKumarM You can put the sertr vs rawC values here.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am confused. Is this going to be fixed?

Copy link
Contributor

@LakshmiKumar23 LakshmiKumar23 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Tested on HIP and HOST backend

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ci:precheckin enhancement New feature or request
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants