fp_accum.cpp

Go to the documentation of this file.

/************************************************
Copyright (c) 2016, Xilinx, Inc.
All rights reserved.

Redistribution and use in source and binary forms, with or without modification, 
are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright notice, 
this list of conditions and the following disclaimer.

2. Redistributions in binary form must reproduce the above copyright notice, 
this list of conditions and the following disclaimer in the documentation 
and/or other materials provided with the distribution.

3. Neither the name of the copyright holder nor the names of its contributors 
may be used to endorse or promote products derived from this software 
without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND 
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, 
THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 
IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 
HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, 
EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
************************************************/

#include <stdlib.h>
#include <stdio.h>
#include <ap_int.h>
#include <ap_fixed.h>

#define NUM_ELEM 128

#define DB_OPTIMIZED
#ifndef DB_OPTIMIZED


float hls_fp_accumulator(float window[NUM_ELEM])
{

        float result = 0.0;

        L1:for(unsigned char x=0; x<NUM_ELEM;x++)
        {
                result = result + window[x];
        }

        return result;
}

#else

float hls_fp_accumulator(float window0[NUM_ELEM])
{

        float window1[NUM_ELEM/2] = {0.0};
        float window2[NUM_ELEM/4] = {0.0};
        float window3[NUM_ELEM/8] = {0.0};
        float window4[NUM_ELEM/16]= {0.0};
        float window5[NUM_ELEM/32]= {0.0};
        float window6[NUM_ELEM/64]= {0.0};

        float result = 0.0;

        L1: for(ap_uint<7> x=0; x.to_uint()<NUM_ELEM/2; x++)
    {
#pragma HLS PIPELINE
         window1[x] = window0[x] +  window0[NUM_ELEM/2+x];
        }
        L2: for(ap_uint<7> x=0; x.to_uint()<NUM_ELEM/4; x++)
    {
#pragma HLS PIPELINE

         window2[x] = window1[x] +  window1[NUM_ELEM/4+x];
        }
        L3: for(ap_uint<7> x=0; x.to_uint()<NUM_ELEM/8; x++)
    {
#pragma HLS PIPELINE

         window3[x] = window2[x] +  window2[NUM_ELEM/8+x];
        }
        L4: for(ap_uint<7> x=0; x.to_uint()<NUM_ELEM/16; x++)
    {
#pragma HLS PIPELINE

         window4[x] = window3[x] +  window3[NUM_ELEM/16+x];
        }
        L5: for(ap_uint<7> x=0; x.to_uint()<NUM_ELEM/32; x++)
    {
#pragma HLS PIPELINE

         window5[x] = window4[x] +  window4[NUM_ELEM/32+x];
        }
        L6: for(ap_uint<7> x=0; x.to_uint()<NUM_ELEM/64; x++)
    {
#pragma HLS PIPELINE

         window6[x] = window5[x] +  window5[NUM_ELEM/64+x];
        }

        result = window6[0] + window6[1];

        return result;
}

#endif




float ref_fp_accumulator(float window[NUM_ELEM]) 
{

        float result = 0.0;

        for(unsigned char x=0; x<NUM_ELEM;x++)
        {
                result = result + window[x];
        }
        
        return result;
}



// DB
//random = pseudoCasual2(3,  7^5, 1, 2^32-1, 1920*1080);
float PseudoCasual(void)
{

        static unsigned long int k = 0;
        static unsigned long int series;

        float result;
        unsigned long long mult;

        const unsigned long long int x0 = 3;
        const unsigned long long int  a = 16807;
        const unsigned int  c = 1;
        const unsigned int module =  4294967295ULL;
        const float recipr_module = (float) 2.3283064e-10f;

        if (k==0)
        {
                mult = a*x0;
        }
        else
        {
                mult = a*series;
        }

        series = (mult+c) % module;

        k++;

        //result = (float) series / (float) module;

        result = (float) series * recipr_module;

        return result;

}


int main(void)
{
        int x, y;
        int ret_val = 0;

        float ref_window[NUM_ELEM];
        float hls_window[NUM_ELEM];
        float threshold = ((float)1.0)/1024;

        for (x=0; x < NUM_ELEM; x++)
        {
                        ref_window[x] = (65536)*PseudoCasual();
                        hls_window[x] = ref_window[x] ;
        }

        // REF
        float ref_res  = ref_fp_accumulator(ref_window);

        // DUT
        float hls_res  = hls_fp_accumulator(hls_window);


        // check results
        float total_error = 9.5367e-07f;

        float diff = ref_res - hls_res;
        if (diff < threshold) diff = 0-diff; // take absolute value
        if (diff > threshold)
        {
                total_error += (float) diff;
        }
        
        printf("\n%010.4f\t%010.4f\t%010.4f\n", ref_res, hls_res, total_error);


        if (total_error < 1.0)
        {
                ret_val=0;
                printf("TEST OK!\n");
        }
        else
        {
                ret_val=1;
                printf("TEST FAILED!\n");
        }


        return ret_val;

}