C-DAC Tech Workshop : hyPACK-2013
                October 15-18, 2013

---------------------------------------------------------------------------
                        README for  hypack-hpc-opencl-v0.1
---------------------------------------------------------------------------


	The following are the details for using hypack-hpc-opencl-v0.1.
              
	Section 1 : Installation of hypack-hpc-opencl-v0.1.
	Section 2 : Cleanup.
   	Section 3 : Limitations.



------------------------------------------------------------------------------
Section 1 : Installation, Compilation and Execution of hypack-hpc-opencl-v0.1
------------------------------------------------------------------------------

  1.1.How to install hypack-hpc-opencl-v0.1
       ---------------------

       Untar the tar hypack-hpc-opencl-v0.1.tar using the following command

              tar -xvf  hypack-hpc-opencl-v0.1

  1.2. How to compile & Create Executable in  hypack-hpc-opencl-v0.1
        --------------------------------------------------

       Use the "Makefile" in the directory and use the following command
       for compilation and creation of a executable same as of benchmark
       in hypack-hpc-opencl-v0.1
            	
	make 

  1.3. How to Execute in hypack-hpc-opencl-v0.1
	---------------------------------------

	The executable in  hypack-hpc-opencl-v0.1 is executed on target machine 
	using the following command.

	./executable name
	
	for example: ./MatMatMultLocalMemSP

-------------------------------------------------------------------------
Section 2 : Cleanup
-----------------------------------------------------------------------

To clean up the executable file, type
                make clean

---------------------------------------------------------------------------
Section 3 :  Limitations
--------------------------------------------------------------------------

    1. Codes are designed for square matrices.
    2. Codes are designed for single platform and single device.
    3. The codes with local memory implementation Data sizes should be multiple of BLOCK_SIZE.

Note:- SP: single precision , DP: double precision 
       GM: global memory , LM: local memory

-----------------------------------------------------------------------------------------------------------