/*------------------------------------------------------------
 *                              CACTI 4.0
 *         Copyright 2005 Hewlett-Packard Development Corporation
 *                         All Rights Reserved
 *
 * Permission to use, copy, and modify this software and its documentation is
 * hereby granted only under the following terms and conditions.  Both the
 * above copyright notice and this permission notice must appear in all copies
 * of the software, derivative works or modified versions, and any portions
 * thereof, and both notices must appear in supporting documentation.
 *
 * Users of this software agree to the terms and conditions set forth herein, and
 * hereby grant back to Hewlett-Packard Company and its affiliated companies ("HP")
 * a non-exclusive, unrestricted, royalty-free right and license under any changes, 
 * enhancements or extensions  made to the core functions of the software, including 
 * but not limited to those affording compatibility with other hardware or software
 * environments, but excluding applications which incorporate this software.
 * Users further agree to use their best efforts to return to HP any such changes,
 * enhancements or extensions that they make and inform HP of noteworthy uses of
 * this software.  Correspondence should be provided to HP at:
 *
 *                       Director of Intellectual Property Licensing
 *                       Office of Strategy and Technology
 *                       Hewlett-Packard Company
 *                       1501 Page Mill Road
 *                       Palo Alto, California  94304
 *
 * This software may be distributed (but not offered for sale or transferred
 * for compensation) to third parties, provided such third parties agree to
 * abide by the terms and conditions of this notice.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND HP DISCLAIMS ALL
 * WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS.   IN NO EVENT SHALL HP 
 * CORPORATION BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
 * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
 * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
 * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
 * SOFTWARE.
 *------------------------------------------------------------*/

#include <math.h>
#include "def.h"
#include "areadef.h"
#include "leakage.h"
#include "basic_circuit.h"
#include "stdio.h"

//extern double calculate_area (area_type, double);

//v4.1: Earlier all the dimensions (length/width/thickness) of transistors and wires and
//all delays were calculated for the 0.8 micron process and then scaled to the input techology. 
//Now all dimensions and delays are calculated directly for the input technology. 

int force_tag, force_tag_size;

void reset_tag_device_widths() {

  Wtdecdrivep_second= 0.0;
  Wtdecdriven_second= 0.0;
  Wtdecdrivep_first= 0.0;
  Wtdecdriven_first= 0.0;
  Wtdec3to8n = 0.0;
  Wtdec3to8p = 0.0;
  WtdecNORn  = 0.0;
  WtdecNORp  = 0.0;
  Wtdecinvn  = 0.0;
  Wtdecinvp  = 0.0;
  WtwlDrvn = 0.0;
  WtwlDrvp = 0.0;

  Wtbitpreequ= 0.0;
  Wtiso= 0.0;
  Wtpch= 0.0;
  Wtiso= 0.0;
  WtsenseEn= 0.0;
  WtsenseN= 0.0;
  WtsenseP= 0.0;
  WtoBufN = 0.0;
  WtoBufP = 0.0;
  WtsPch= 0.0;

  WtpchDrvp= 0.0; WtpchDrvn= 0.0;
  WtisoDrvp= 0.0; WtisoDrvn= 0.0;
  WtspchDrvp= 0.0; WtspchDrvn= 0.0;
  WtsenseEnDrvp= 0.0; WtsenseEnDrvn= 0.0;

  WtwrtMuxSelDrvn= 0.0;
  WtwrtMuxSelDrvp= 0.0;
}
void reset_data_device_widths()
{
  Waddrdrvn1 = 0.0;
  Waddrdrvp1= 0.0;
  Waddrdrvn2= 0.0;
  Waddrdrvp2= 0.0;

  Wdecdrivep_second = 0.0;
  Wdecdriven_second = 0.0;
  Wdecdrivep_first = 0.0;
  Wdecdriven_first = 0.0;
  Wdec3to8n = 0.0;
  Wdec3to8p = 0.0;
  WdecNORn  = 0.0;
  WdecNORp  = 0.0;
  Wdecinvn  = 0.0;
  Wdecinvp  = 0.0;
  WwlDrvn = 0.0;
  WwlDrvp = 0.0;



  Wbitpreequ= 0.0;

  Wiso = 0.0;
  Wpch= 0.0;
  Wiso= 0.0;
  WsenseEn= 0.0;
  WsenseN= 0.0;
  WsenseP= 0.0;
  WoBufN = 0.0;
  WoBufP = 0.0;
  WsPch= 0.0;

  WpchDrvp= 0.0; WpchDrvn= 0.0;
  WisoDrvp= 0.0; WisoDrvn= 0.0;
  WspchDrvp= 0.0; WspchDrvn= 0.0;
  WsenseEnDrvp= 0.0; WsenseEnDrvn= 0.0;

  WwrtMuxSelDrvn= 0.0;
  WwrtMuxSelDrvp= 0.0;

  WmuxdrvNANDn    = 0.0;
  WmuxdrvNANDp    = 0.0;
  WmuxdrvNORn	= 0.0;
  WmuxdrvNORp	= 0.0;
  Wmuxdrv3n	= 0.0;
  Wmuxdrv3p	= 0.0;
  Woutdrvseln	= 0.0;
  Woutdrvselp	= 0.0;

  Woutdrvnandn= 0.0;
  Woutdrvnandp= 0.0;
  Woutdrvnorn	= 0.0;
  Woutdrvnorp	= 0.0;
  Woutdrivern	= 0.0;
  Woutdriverp	= 0.0;
}

void compute_device_widths(int C,int B,int A,int fullyassoc, int Ndwl,int Ndbl,double Nspd)
{
  int rows, cols, numstack,l_predec_nor_v,l_predec_nor_h;
  double desiredrisetime, Rpdrive, Cline, Cload;
  double effWdecNORn,effWdecNORp,effWdec3to8n,effWdec3to8p, wire_res, wire_cap;
  int l_outdrv_v,l_outdrv_h;
  //int rows_fa_subarray,cols_fa_subarray, tagbits;
  int tagbits;
  int horizontal_edge = 0;
  int nr_subarrays_left = 0, v_or_h = 0;
  int horizontal_step = 0, vertical_step = 0;
  int h_inv_predecode = 0, v_inv_predecode = 0;

  double previous_ndriveW = 0, previous_pdriveW = 0, current_ndriveW = 0, current_pdriveW = 0;
  int i;

  //v4.1: Fixing double->int type conversion problems. EPSILON is added below to make sure
  //the final int value is the correct one 
  //rows = C/(CHUNKSIZE*B*A*Ndbl*Nspd);
  rows = (int) (C/(CHUNKSIZE*B*A*Ndbl*Nspd) + EPSILON);
  //cols = CHUNKSIZE*B*A*Nspd/Ndwl;
  cols = (int) (CHUNKSIZE*B*A*Nspd/Ndwl + EPSILON);

  // Wordline capacitance to determine the wordline driver size
  /* Use a first-order approx */
  desiredrisetime = krise*log((double)(cols))/2.0;
  /*dt: I'm changing this back to what's in CACTI (as opposed to eCacti), i.e. counting the short poly connection to the pass transistors*/
  Cline = (gatecappass(Wmemcella,(BitWidth-2*Wmemcella)/2.0)+ gatecappass(Wmemcella,(BitWidth-2*Wmemcella)/2.0)+ Cwordmetal)*cols;
  Rpdrive = desiredrisetime/(Cline*log(VSINV)*-1.0);
  WwlDrvp = restowidth(Rpdrive,PCH);
  if (WwlDrvp > Wworddrivemax) {
    WwlDrvp = Wworddrivemax;
  }

  /* Now that we have a reasonable psize, do the rest as before */
  /* If we keep the ratio the same as the tag wordline driver,
     the threshold voltage will be close to VSINV */

  /* assuming that nsize is half the psize */
  WwlDrvn = WwlDrvp/2;

  // Size of wordline
  // Sizing ratio for optimal delay is 3-4.
  Wdecinvn = (WwlDrvp + WwlDrvn) * SizingRatio * 1/3;
  Wdecinvp = (WwlDrvp + WwlDrvn) * SizingRatio * 2/3;

  // determine size of nor and nand gates in the decoder

  // width of NOR driving decInv -
  // effective width (NORp + NORn = Cout/SizingRatio( FANOUT))
  // Cout = Wdecinvn + Wdecinvp; SizingRatio = 3;
  // nsize = effWidth/3; psize = 2*effWidth/3;

  numstack =
    (int)ceil((1.0/3.0)*logtwo( (double)((double)C/(double)(B*A*Ndbl*Nspd))));
  if (numstack==0) numstack = 1;

  if (numstack>5) numstack = 5;

  effWdecNORn = (Wdecinvn + Wdecinvp)*SizingRatio/3;
  effWdecNORp = 2*(Wdecinvn + Wdecinvp)*SizingRatio/3;
  WdecNORn = effWdecNORn;
  WdecNORp = effWdecNORp * numstack;

  /* second stage: driving a bunch of nor gates with a nand */

  /*dt: The *8 is there because above we mysteriously divide the capacity in BYTES by the number of BITS per wordline */
  l_predec_nor_v = rows*CHUNKSIZE; 
  /*dt: If we follow the original drawings from the TR's, then there is almost no horizontal wires, only the poly for contacting
    the nor gates. The poly part we don't model right now */
  l_predec_nor_h = 0;

  //v4.1: Scaling the poly length to the input tech node.
  //Cline = gatecap(WdecNORn+WdecNORp,((numstack*40)+20.0))*rows/8 +
  //GlobalCbitmetal*(l_predec_nor_v)+GlobalCwordmetal*(l_predec_nor_h);

  Cline = gatecap(WdecNORn+WdecNORp,((numstack*40 / FUDGEFACTOR)+20.0 / FUDGEFACTOR))*rows/8 +
    GlobalCbitmetal*(l_predec_nor_v)+GlobalCwordmetal*(l_predec_nor_h);

  Cload = Cline / gatecap(1.0,0.0);

  effWdec3to8n = Cload*SizingRatio/3;
  effWdec3to8p = 2*Cload*SizingRatio/3;

  Wdec3to8n = effWdec3to8n * 3; // nand3 gate
  Wdec3to8p = effWdec3to8p;

  // size of address drivers before decoders
  /* First stage: driving the decoders */
  //v4.1: Fixing double->int type conversion problems. EPSILON is added below to make sure
  //the final int value is the correct one 
  //horizontal_edge = CHUNKSIZE*B*A*Nspd;
  horizontal_edge = (int) (CHUNKSIZE*B*A*Nspd + EPSILON);

  previous_ndriveW = Wdec3to8n;
  previous_pdriveW = Wdec3to8p;

  if(Ndwl*Ndbl==1 ) {
    wire_cap = GlobalCwordmetal*horizontal_edge;
    wire_res = 0.5*GlobalRwordmetal*horizontal_edge;
    Cdectreesegments[0] = GlobalCwordmetal*horizontal_edge;
    Rdectreesegments[0] = 0.5*GlobalRwordmetal*horizontal_edge;

    Cline = 4*gatecap(previous_ndriveW+previous_pdriveW,10.0 / FUDGEFACTOR)+GlobalCwordmetal*horizontal_edge;
    Cload = Cline / gatecap(1.0,0.0);

    current_ndriveW = Cload*SizingRatio/3;
    current_pdriveW = 2*Cload*SizingRatio/3;

    nr_dectreesegments = 0;
  }
  else if(Ndwl*Ndbl==2 || Ndwl*Ndbl==4) {
    wire_cap = GlobalCwordmetal*horizontal_edge;
    wire_res = 0.5*GlobalRwordmetal*horizontal_edge;
    Cdectreesegments[0] = GlobalCwordmetal*horizontal_edge;
    Rdectreesegments[0] = 0.5*GlobalRwordmetal*horizontal_edge;

    Cline = 4*gatecap(previous_ndriveW+previous_pdriveW,10.0 / FUDGEFACTOR)+GlobalCwordmetal*horizontal_edge;
    Cload = Cline / gatecap(1.0,0.0);

    current_ndriveW = Cload*SizingRatio/3;
    current_pdriveW = 2*Cload*SizingRatio/3;

    nr_dectreesegments = 0;
  }
  else {
    /*dt: For the critical path  in an H-Tree, the metal */

    nr_subarrays_left = Ndwl* Ndbl;
    /*all the wires go to quads of subarrays where they get predecoded*/
    nr_subarrays_left /= 4;
    //v4.1: Fixing double->int type conversion problems. EPSILON is added below to make sure
    //the final int value is the correct one 
    //horizontal_step = CHUNKSIZE*B*A*Nspd/Ndwl;
    horizontal_step = (int) (CHUNKSIZE*B*A*Nspd/Ndwl + EPSILON);
    //vertical_step = C/(B*A*Ndbl*Nspd);
    vertical_step = (int) (C/(B*A*Ndbl*Nspd) + EPSILON);
    h_inv_predecode = horizontal_step;

    Cdectreesegments[0] = GlobalCwordmetal*horizontal_step;
    Rdectreesegments[0] = 0.5*GlobalRwordmetal*horizontal_step;
    Cline = 4*gatecap(previous_ndriveW+previous_pdriveW,10.0 / FUDGEFACTOR)+GlobalCwordmetal*horizontal_step;
    Cload = Cline / gatecap(1.0,0.0);

    current_ndriveW = Cload*SizingRatio/3;
    current_pdriveW = 2*Cload*SizingRatio/3;
    WdecdrivetreeN[0] = current_ndriveW;
    nr_dectreesegments = 1;

    horizontal_step *= 2;
    v_or_h = 1; // next step is vertical

    while(nr_subarrays_left > 1) {
      previous_ndriveW = current_ndriveW;
      previous_pdriveW = current_pdriveW;
      nr_dectreesegments++;
      if(v_or_h) {
        v_inv_predecode += vertical_step;
        Cdectreesegments[nr_dectreesegments-1] = GlobalCbitmetal*vertical_step;
        Rdectreesegments[nr_dectreesegments-1] = 0.5*GlobalRbitmetal*vertical_step;
        Cline = gatecap(previous_ndriveW+previous_pdriveW,0)+GlobalCbitmetal*vertical_step;
        v_or_h = 0;
        vertical_step *= 2;
        nr_subarrays_left /= 2;
      }
      else {
        h_inv_predecode += horizontal_step;
        Cdectreesegments[nr_dectreesegments-1] = GlobalCwordmetal*horizontal_step;
        Rdectreesegments[nr_dectreesegments-1] = 0.5*GlobalRwordmetal*horizontal_step;
        Cline = gatecap(previous_ndriveW+previous_pdriveW,0)+GlobalCwordmetal*horizontal_step;
        v_or_h = 1;
        horizontal_step *= 2;
        nr_subarrays_left /= 2;
      }
      Cload = Cline / gatecap(1.0,0.0);

      current_ndriveW = Cload*SizingRatio/3;
      current_pdriveW = 2*Cload*SizingRatio/3;

      WdecdrivetreeN[nr_dectreesegments-1] = current_ndriveW;
    }

    if(nr_dectreesegments >= 10) {
      printf("Too many segments in the data decoder H-tree. Overflowing the preallocated array!");
      exit(1);
    }
    wire_cap = GlobalCbitmetal*v_inv_predecode + GlobalCwordmetal*h_inv_predecode;
    wire_res = 0.5*(GlobalRbitmetal*v_inv_predecode + GlobalRwordmetal*h_inv_predecode);


  }

  Wdecdriven_second = current_ndriveW;
  Wdecdrivep_second = current_pdriveW;

  // Size of second driver

  Wdecdriven_first = (Wdecdriven_second + Wdecdrivep_second)*SizingRatio/3;
  Wdecdrivep_first = 2*(Wdecdriven_second + Wdecdrivep_second)*SizingRatio/3;

  // these are the widths of the devices of dataoutput devices
  // will be used in the data_senseamplifier_data and dataoutput_data functions

  l_outdrv_v = 0;
  l_outdrv_h = 0;

  if(!fullyassoc) {

    //v4.1: Fixing double->int type conversion problems. EPSILON is added below to make sure
    //the final int value is the correct one 
    //rows =  (C/(B*A*Ndbl*Nspd));
    //cols = (CHUNKSIZE*B*A*Nspd/Ndwl);
    rows =  (int) (C/(B*A*Ndbl*Nspd) + EPSILON);
    cols = (int) (CHUNKSIZE*B*A*Nspd/Ndwl + EPSILON);
  }
  else {