RPP Tensor Support - Fisheye on HOST and HIP #346

sampath1117 · 2024-09-25T03:52:49Z

Adds tensor support for Fog function optimized on AVX2 and HIP
Adds test suite support

consolidated the code

…-PLN3 variant

Srihari-mcw · 2025-01-03T10:14:01Z

CHANGELOG.md

@@ -2,12 +2,19 @@

 Full documentation for RPP is available at [https://rocm.docs.amd.com/projects/rpp/en/latest](https://rocm.docs.amd.com/projects/rpp/en/latest)

+## RPP 1.18.0 (unreleased)


Add the change under 1.9.10

Srihari-mcw · 2025-01-03T10:15:02Z

CMakeLists.txt

@@ -43,7 +43,7 @@ endif()
 set(CMAKE_CXX_STANDARD 17)

 # RPP Version
-set(VERSION "1.9.10")
+set(VERSION "1.18.0")


Set version here to 1.9.10

Srihari-mcw · 2025-01-03T12:40:02Z

include/rpp_version.h

@@ -39,8 +39,8 @@ extern "C" {
 #endif
 // NOTE: IMPORTANT: Match the version with CMakelists.txt version
 #define RPP_VERSION_MAJOR 1
-#define RPP_VERSION_MINOR 9


Here also retain version to 1.9.10

Srihari-mcw · 2025-01-05T14:41:13Z

src/include/cpu/rpp_cpu_simd.hpp

@@ -120,6 +120,8 @@ const __m256 avx_p255 = _mm256_set1_ps(255.0f);
 const __m256 avx_p1op255 = _mm256_set1_ps(1.0f / 255.0f);
 const __m256 avx_p1op3 = _mm256_set1_ps(1.0f / 3.0f);
 const __m256 avx_p2op3 = _mm256_set1_ps(2.0f / 3.0f);
+const __m256 avx_pMinus1 = _mm256_set1_ps(-1);


Can we make this -1.0f?

Srihari-mcw · 2025-01-05T14:48:06Z

src/modules/cpu/kernel/fisheye.hpp

+
+inline void compute_fisheye_src_loc_avx(__m256 &pDstY, __m256 &pDstX, __m256 &pSrcY, __m256 &pSrcX, __m256 &pHeight, __m256 &pWidth)
+{
+    __m256 pNormX, pNormY, pDist;


Add the AVX helper functions overall inside #if AVX2

Srihari-mcw · 2025-01-05T15:28:02Z

src/modules/cpu/kernel/fisheye.hpp

+        }
+
+        // fisheye without fused output-layout toggle (NHWC -> NHWC)
+        if ((srcDescPtr->c == 3) && (srcDescPtr->layout == RpptLayout::NHWC) && (dstDescPtr->layout == RpptLayout::NHWC))


this must be else if

Srihari-mcw · 2025-01-05T15:37:40Z

src/modules/cpu/kernel/fisheye.hpp

+                    {
+                        __m256i pRow;
+                        rpp_simd_load(rpp_generic_nn_load_i8pln1_avx, srcPtrTempChn, srcLocArray, invalidLoad, pRow);
+                        rpp_storeu_si64((__m128i *)(dstPtrTempChn), _mm256_castsi256_si128(pRow));


Can probably use reinterpret_cast<__m128i *>(dstPtrTempChn) here

Srihari-mcw · 2025-01-05T16:25:24Z

src/modules/hip/kernel/fisheye.hpp

+    int id_z = hipBlockIdx_z * hipBlockDim_z + hipThreadIdx_z;
+    int4 srcRoi_i4 = *(int4 *)&roiTensorPtrSrc[id_z];
+    int width = (srcRoi_i4.z - srcRoi_i4.x) + 1;
+    int height =  (srcRoi_i4.w - srcRoi_i4.y) + 1;


Fix indentation

in all the functions

Srihari-mcw

Please address review comments. Thanks

…ocs/sphinx (ROCm#515) Bumps [rocm-docs-core[api_reference]](https://github.com/ROCm/rocm-docs-core) from 1.14.1 to 1.15.0. - [Release notes](https://github.com/ROCm/rocm-docs-core/releases) - [Changelog](https://github.com/ROCm/rocm-docs-core/blob/develop/CHANGELOG.md) - [Commits](ROCm/rocm-docs-core@v1.14.1...v1.15.0) --- updated-dependencies: - dependency-name: rocm-docs-core[api_reference] dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

sampath1117 added 11 commits September 10, 2024 06:08

added initial api support for fisheye HIP kernel

12bc18f

added initial HIP support for PKD3-PKD3 variant

1b764d2

added ROI support

f88165e

consolidated the code

added support for remaining layout variants

1857a7f

added few comments in the code

1bd58c7

added initial api support for fisheye HOST kernel

1518df5

added initial code for fishey src loc compute and support for U8 PKD3…

50859a9

…-PLN3 variant

added support for reamining layout variants in U8

cd85784

added support for F16 bitdepth

df9cb87

added support for F32 and I8 bitdepths in HOST

88e253e

Merge branch 'develop' into sr/opt_fisheye

e34e0e4

sampath1117 changed the title ~~RPP Tensor Support - Fisheye on HOST and HIP~~ WIP - RPP Tensor Support - Fisheye on HOST and HIP Sep 25, 2024

sampath1117 changed the base branch from develop to master September 26, 2024 11:15

sampath1117 changed the base branch from master to develop September 26, 2024 11:15

sampath1117 added 3 commits September 26, 2024 12:08

added the versioning changes

13fdc30

added fisheye output image in doxygen

1745580

added golden outputs

287ae33

sampath1117 changed the title ~~WIP - RPP Tensor Support - Fisheye on HOST and HIP~~ RPP Tensor Support - Fisheye on HOST and HIP Sep 27, 2024

HazarathKumarM added 3 commits November 19, 2024 07:02

Merge with develop

49282b8

Merge remote-tracking branch 'develop' into sr/opt_fisheye

4f68499

Fixes on the QA Failure on the HOST side

2113c51

HazarathKumarM force-pushed the sr/opt_fisheye branch from 01e5007 to 2113c51 Compare November 25, 2024 04:19

HazarathKumarM added 3 commits December 13, 2024 10:21

Merge branch 'develop' into opt_fisheye

f7ab4dd

Changes in Test suite

c4014c4

Minor changes due to merge conflits

27bddad

Srihari-mcw reviewed Jan 3, 2025

View reviewed changes

Merge branch 'develop' into sr/opt_fisheye

8b40e87

Srihari-mcw reviewed Jan 3, 2025

View reviewed changes

Srihari-mcw reviewed Jan 5, 2025

View reviewed changes

Srihari-mcw suggested changes Jan 5, 2025

View reviewed changes

HazarathKumarM force-pushed the sr/opt_fisheye branch from b25e8ec to 8b40e87 Compare January 6, 2025 07:02

HazarathKumarM and others added 4 commits January 6, 2025 07:03

Minor changes based on the review commands

d5d9095

Merge branch 'develop' into sr/opt_fisheye

9e1e777

Changes due to merge conflits

3cd9cb5

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

RPP Tensor Support - Fisheye on HOST and HIP #346

RPP Tensor Support - Fisheye on HOST and HIP #346

sampath1117 commented Sep 25, 2024

Srihari-mcw Jan 3, 2025

Srihari-mcw Jan 3, 2025

Srihari-mcw Jan 3, 2025

Srihari-mcw Jan 5, 2025

Srihari-mcw Jan 5, 2025

Srihari-mcw Jan 5, 2025

Srihari-mcw Jan 5, 2025

Srihari-mcw Jan 5, 2025

Srihari-mcw Jan 5, 2025

Srihari-mcw left a comment

		@@ -2,12 +2,19 @@

		Full documentation for RPP is available at [https://rocm.docs.amd.com/projects/rpp/en/latest](https://rocm.docs.amd.com/projects/rpp/en/latest)

		## RPP 1.18.0 (unreleased)

RPP Tensor Support - Fisheye on HOST and HIP #346

Are you sure you want to change the base?

RPP Tensor Support - Fisheye on HOST and HIP #346

Conversation

sampath1117 commented Sep 25, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Srihari-mcw left a comment

Choose a reason for hiding this comment