Squashed 'solver' changes from d0d987d..8fbf3fc (#995)

8fbf3fc add detail explaination on svd (#543)
0757518 create main branch from next branch

Co-authored-by: sdausr <sdausr@xilinx.com>

solver/Jenkinsfile

@@ -1,5 +1,5 @@
 @Library('pipeline-library')_
 
-VitisLibPipeline (branch: 'next', libname: 'xf_solver', TARGETS: 'hls_csim:hls_csynth:hls_cosim:vitis_sw_emu:vitis_hw_emu:vitis_hw_build:vitis_aie_sim:vitis_aie_x86sim',
-                  upstream_dependencies: 'xf_utils_hw,next,../utils;xf_dsp,next,../dsp',
-                  devtest: 'RunDeploy.sh', TOOLVERSION: '2023.2_stable_latest', mail_on:'daily:PR')
+VitisLibPipeline (branch: 'main', libname: 'xf_solver', TARGETS: 'hls_csim:hls_csynth:hls_cosim:vitis_sw_emu:vitis_hw_emu:vitis_hw_build:vitis_aie_sim:vitis_aie_x86sim',
+                  upstream_dependencies: 'xf_utils_hw,main,../utils;xf_dsp,main,../dsp',
+                  devtest: 'RunDeploy.sh', TOOLVERSION: '2023.2_released', mail_on:'daily:PR')

solver/docs/src/guide_L2_AIE/MatrixDecomposition/qr_decomposition/qr_decomposition.rst

@@ -48,7 +48,7 @@ Template Parameters
 ---------------------
 * `column_num`: the number of columns;
 * `row_num`: the number of rows;
-* `k_rep`: the number of input matrix;
+* `k_rep`: the number of input matrixs that are processed by one graph call.
 
 To access more details, see :ref:`AIE APIs Overview`.
 

solver/docs/src/guide_L2_AIE/MatrixDecomposition/svd/svd.rst

@@ -27,7 +27,7 @@ Singular value Decomposition
 Introduction
 ==============
 
-This function computes the Singular value decomposition of matrix :math:`A`
+This function computes the Singular value decomposition of matrix :math:`A`. Current implementation takes one-sided Jacobi algorithm to solve. One-sided Jacobi is an iterative convergence algorithm, and each graph call will perform one "column-sweep" on a batch of input matrixs. To achieve desired precision, you need to store the output of one graph call and feed them as input to the graph again. Such "output-input" needs to be executed multiple times and the number of repetitions depends on precision requirements. Due to "one-way" through character of current implementation, it will need PL data-movers to servers as the mechanism to loop back last iteration's output into next iteration's input. We will release such PL data-movers in the future.
 
 .. math::
     A = U M V*
@@ -47,7 +47,7 @@ Template Parameters
 ---------------------
 * `column_num`: the number of columns;
 * `row_num`: the number of rows;
-* `k_rep`: the number of input matrix;
+* `k_rep`: Each graph call will perform one "column-sweep" on all `k_rep` input matrixs and make columns.
 
 To access more details, see :ref:`AIE APIs Overview`.
 
@@ -79,7 +79,7 @@ Kernel Interfaces
 
 .. note::
 
-   * To utilize bandwidth of input / output stream, the input matrix and output result are transfered in such way: Elem[N*4] and Elem[N*4+1] are transferred with in_0 / out_0, Elem[N*4 + 2] and Elem[N*4 + 3] are transferred with in_1 / out_1.
+   * To utilize bandwidth of input / output stream, the input matrix and output result are transfered in such way: Elem[N*4] and Elem[N*4+1] are transferred with in_0 / out_0, Elem[N*4 + 2] and Elem[N*4 + 3] are transferred with in_1 / out_1. Input matrix contains :math: `A[M*N]` and :math: `U[M*M]` (original input matrix and zero matrix as place holder if it's first graph call, or output matrix from last graph call) and comes in column-major order, like A[:,0], U[:,0], A[:,1], U[:,1], ect.
 
 
 * Input: