RPP RICAP Tensor for HOST and HIP

[r-abishek]

Introduces RICAP functionality for HOST and HIP in RPP as per https://arxiv.org/pdf/1811.09030.pdf Adds corresponding unit/qa/performance tests

[kiritigowda]

@r-abishek this pr has conflicts.

[kiritigowda]

@r-abishek this pr has few issues - https://app.codacy.com/gh/GPUOpen-ProfessionalCompute-Libraries/rpp/pullRequest?prid=12598993

[r-abishek]

@kiritigowda Codacy and conflicts resolved

[paveltc]

@r-abishek This test passes both HOST and HIP QA tests. I need to talk to you about the unit tests and where to get the inputs for them.

[paveltc]

@r-abishek I ran the unit tests for this and sent you an email regarding what I found. If the behavior I saw is expected, this PR should be good to merge.

[r-abishek]

@paveltc RICAP is good, its just that the test-suite inputs given to the function are randomized, so creates black areas if inputs are not range bound. A bound has been added on the test suite to guarantee a better output, when the demo / unit test is run by users.

[paveltc]

@rrawther @kiritigowda @r-abishek I am now getting the following error when running the latest version of this PR: /home/svcbuild/rpp-pr213/rpp/utilities/test_suite/HIP/../rpp_test_suite_common.h:1092:20: error: use of undeclared identifier 'boost' randFromDist = boost::math::ibeta_inv(betaParam, betaParam, p); // Computes the inverse of the incomplete beta function on parameter ^ /home/svcbuild/rpp-pr213/rpp/utilities/test_suite/HIP/../rpp_test_suite_common.h:1098:21: error: use of undeclared identifier 'boost' randFromDist1 = boost::math::ibeta_inv(betaParam, betaParam, p1);

[r-abishek]

@paveltc This is due to the recent removal of boost from RPP. Added that compatibility for test suite now.

Please try running python3 runTests.py --case_list 82 --qa_mode 0 --test_type 0 --batch_size 3

[paveltc]

@r-abishek I was able to successfully run both the HIP and HOST tests and check their output with the original command line we used, the one with the absolute path to the input images.

[r-abishek]

@paveltc Sounds good. @kiritigowda I believe RICAP could now be merged.

[kiritigowda]

@r-abishek - please resolve conflicts - @paveltc

[r-abishek]

@kiritigowda I think it needs a manual trigger? Could you trigger the CI?

[r-abishek]

HIP:

HOST:

@kiritigowda CI with RICAP seems to be working now. Passes HIP and HOST

[r-abishek]

@sampath1117 @snehaa8 We will need QA passes on tests henceforth. Pls take a look at above images and the CI link - http://math-ci.amd.com/blue/organizations/jenkins/main%2FShort-GPUOpen%2Frpp/detail/PR-213/12/pipeline

RICAP PR merged into develop is the initial test for this. Audio and Voxel PRs will need to have this.