Skip to content

RPP Tensor Fisheye on HOST and HIP#654

Merged
kiritigowda merged 41 commits intoROCm:developfrom
Srihari-mcw:fisheye_dev_branch
Jan 22, 2026
Merged

RPP Tensor Fisheye on HOST and HIP#654
kiritigowda merged 41 commits intoROCm:developfrom
Srihari-mcw:fisheye_dev_branch

Conversation

@Srihari-mcw
Copy link
Copy Markdown
Contributor

  • Adds Fisheye augmentation on HIP and HOST
  • Adds support for U8/F32/F16/I8 bit depths and NCHW/NHWC variants with toggle support
  • Adds relevant unit / qa / performance tests

Copy link
Copy Markdown
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull request overview

This pull request adds fisheye lens distortion augmentation support for both HIP (GPU) and HOST (CPU) backends. The implementation supports multiple data types (U8, F16, F32, I8) and layout formats (NCHW, NHWC) with layout conversion capabilities.

Key Changes

  • Added fisheye augmentation kernel implementations for HIP and HOST
  • Integrated fisheye into test suites with appropriate test case ID (25)
  • Added API documentation with sample input/output images
  • Implemented AVX2-optimized SIMD code paths for CPU execution

Reviewed changes

Copilot reviewed 11 out of 14 changed files in this pull request and generated 1 comment.

Show a summary per file
File Description
utilities/test_suite/rpp_test_suite_image.h Added fisheye enum entry (FISHEYE = 25) and augmentation map entry
utilities/test_suite/common.py Added fisheye to test configuration with HOST and HIP backend support, categorized as geometric augmentation
utilities/test_suite/HOST/Tensor_image_host.cpp Added HOST test case for fisheye supporting U8/F16/F32/I8 data types
utilities/test_suite/HIP/Tensor_image_hip.cpp Added HIP test case for fisheye supporting U8/F16/F32/I8 data types
src/modules/tensor/rppt_tensor_geometric_augmentations.cpp Implemented rppt_fisheye_host and rppt_fisheye_gpu API functions with data type dispatching
src/modules/tensor/hip/kernel/fisheye.cpp New HIP kernel implementation with device helpers and layout-specific kernels
src/modules/tensor/cpu/kernel/fisheye.cpp New CPU kernel implementation with AVX2 optimizations and scalar fallback
src/include/tensor/host_tensor_executors.hpp Added function declarations for fisheye HOST implementations
src/include/tensor/hip_tensor_executors.hpp Added template function declaration for fisheye HIP implementation
src/include/common/cpu/rpp_cpu_simd_load_store.hpp Added new AVX constants (avx_pMinus1, avx_p1op2) for fisheye computations
api/rppt_tensor_geometric_augmentations.h Added comprehensive API documentation for rppt_fisheye_host and rppt_fisheye_gpu
docs/data/doxygenOutputs/geometric_augmentations_fisheye_img150x150.png Added sample output image for documentation

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

@kiritigowda kiritigowda merged commit cd9ba5f into ROCm:develop Jan 22, 2026
4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants