Implementing Arnoldi Block and linear algebra function [cBLAS]

[fvella]

• Arnoldi Block implementation
• Take the time to perform the extraction of dense components. Put the profiling function in utils.
• Check if it works by performing GEMM (use a fake matrix as the second operand).
• Provide instructions (compilation, download of the matrix and execution) in the README to reproduce the result.

Matrix instance

[seharsyed]

This information is solely meant for dividing the tasks into sub-tasks and then to check the progress at each step

1. (a) Before implementing Arnoldi, C should be able to read the matrix instance from mtx format and assign the non-zero values to matrix A.
   1(b) Do the conversion of format if necessary

2. Generate a random B matrix and use the same instance for MATLAB readings

3. Arnoldi Implementation

Arnoldi block is further divided into these steps:

a) Factorize the Residual into QR Components
b) Assign Q Components to V
c) Block Matrix Vector Multiplication (would need cblas routine)
d) Element by Element Multiplication
e) Computing Norm

5. Instruction file for compilation, downloading and execution

[seharsyed]

The specified matrix instance was corrupted as there were no values for the array 'val'. Used the following instance for input of matrix A

https://math.nist.gov/MatrixMarket/data/Harwell-Boeing/bcsstruc1/bcsstk01.html

[seharsyed]

Can we close this issue and open the issue for Arnoldi implementation separately by further dividing it into components as I mentioned in my comment? So that the real progress could be followed and if I get stuck somewhere we could trace it down. The basic data and file structure is created. GEMM is applied and implemented the clock as well.

[fvella]

Well done!
Please fix makefile by removing cblas from dependency and close the issue :)