1/***************************************************************************** 2 * McPAT/CACTI 3 * SOFTWARE LICENSE AGREEMENT 4 * Copyright 2012 Hewlett-Packard Development Company, L.P. 5 * Copyright (c) 2010-2013 Advanced Micro Devices, Inc. 6 * All Rights Reserved 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions are 10 * met: redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer; 12 * redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution; 15 * neither the name of the copyright holders nor the names of its 16 * contributors may be used to endorse or promote products derived from 17 * this software without specific prior written permission. 18 19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 20 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 21 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 22 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 23 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 24 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 25 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 26 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 27 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 28 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 29 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 30 * 31 ***************************************************************************/ 32 33#ifndef __CONST_H__ 34#define __CONST_H__ 35 36#include <math.h> 37#include <stdint.h> 38#include <stdio.h> 39#include <stdlib.h> 40#include <string.h> 41 42/* The following are things you might want to change 43 * when compiling 44 */ 45 46/* 47 * Address bits in a word, and number of output bits from the cache 48 */ 49 50/* 51was: #define ADDRESS_BITS 32 52now: I'm using 42 bits as in the Power4, 53since that's bigger then the 36 bits on the Pentium 4 54and 40 bits on the Opteron 55*/ 56const int ADDRESS_BITS = 42; 57 58/*dt: In addition to the tag bits, the tags also include 1 valid bit, 1 dirty bit, 2 bits for a 4-state 59 cache coherency protocoll (MESI), 1 bit for MRU (change this to log(ways) for full LRU). 60 So in total we have 1 + 1 + 2 + 1 = 5 */ 61const int EXTRA_TAG_BITS = 5; 62 63/* limits on the various N parameters */ 64 65const unsigned int MAXDATAN = 512; // maximum for Ndwl and Ndbl 66const unsigned int MAXSUBARRAYS = 1048576; // maximum subarrays for data and tag arrays 67const unsigned int MAXDATASPD = 256; // maximum for Nspd 68const unsigned int MAX_COL_MUX = 256; 69 70 71 72#define ROUTER_TYPES 3 73#define WIRE_TYPES 6 74 75const double Cpolywire = 0; 76 77 78/* Threshold voltages (as a proportion of Vdd) 79 If you don't know them, set all values to 0.5 */ 80#define VTHFA1 0.452 81#define VTHFA2 0.304 82#define VTHFA3 0.420 83#define VTHFA4 0.413 84#define VTHFA5 0.405 85#define VTHFA6 0.452 86#define VSINV 0.452 87#define VTHCOMPINV 0.437 88#define VTHMUXNAND 0.548 // TODO : this constant must be revisited 89#define VTHEVALINV 0.452 90#define VTHSENSEEXTDRV 0.438 91 92 93//WmuxdrvNANDn and WmuxdrvNANDp are no longer being used but it's part of the old 94//delay_comparator function which we are using exactly as it used to be, so just setting these to 0 95const double WmuxdrvNANDn = 0; 96const double WmuxdrvNANDp = 0; 97 98 99/*===================================================================*/ 100/* 101 * The following are things you probably wouldn't want to change. 102 */ 103 104#define BIGNUM 1e30 105#define INF 9999999 106#define MAX(a,b) (((a)>(b))?(a):(b)) 107#define MIN(a,b) (((a)<(b))?(a):(b)) 108 109/* Used to communicate with the horowitz model */ 110#define RISE 1 111#define FALL 0 112#define NCH 1 113#define PCH 0 114 115 116#define EPSILON 0.5 //v4.1: This constant is being used in order to fix floating point -> integer 117//conversion problems that were occuring within CACTI. Typical problem that was occuring was 118//that with different compilers a floating point number like 3.0 would get represented as either 119//2.9999....or 3.00000001 and then the integer part of the floating point number (3.0) would 120//be computed differently depending on the compiler. What we are doing now is to replace 121//int (x) with (int) (x+EPSILON) where EPSILON is 0.5. This would fix such problems. Note that 122//this works only when x is an integer >= 0. 123/* 124 * Sheng thinks this is more a solution to solve the simple truncate problem 125 * (http://www.cs.tut.fi/~jkorpela/round.html) rather than the problem mentioned above. 126 * Unfortunately, this solution causes nasty bugs (different results when using O0 and O3). 127 * Moreover, round is not correct in CACTI since when an extra fraction of bit/line is needed, 128 * we need to provide a complete bit/line even the fraction is just 0.01. 129 * So, in later version than 6.5 we use (int)ceil() to get double to int conversion. 130 */ 131 132#define EPSILON2 0.1 133#define EPSILON3 0.6 134 135 136#define MINSUBARRAYROWS 16 //For simplicity in modeling, for the row decoding structure, we assume 137//that each row predecode block is composed of at least one 2-4 decoder. When the outputs from the 138//row predecode blocks are combined this means that there are at least 4*4=16 row decode outputs 139#define MAXSUBARRAYROWS 262144 //Each row predecode block produces a max of 2^9 outputs. So 140//the maximum number of row decode outputs will be 2^9*2^9 141#define MINSUBARRAYCOLS 2 142#define MAXSUBARRAYCOLS 262144 143 144 145#define INV 0 146#define NOR 1 147#define NAND 2 148 149 150#define NUMBER_TECH_FLAVORS 4 151 152#define NUMBER_INTERCONNECT_PROJECTION_TYPES 2 //aggressive and conservative 153//0 = Aggressive projections, 1 = Conservative projections 154#define NUMBER_WIRE_TYPES 4 //local, semi-global and global 155//1 = 'Semi-global' wire type, 2 = 'Global' wire type 156 157 158const int dram_cell_tech_flavor = 3; 159 160 161#define VBITSENSEMIN 0.08 //minimum bitline sense voltage is fixed to be 80 mV. 162 163#define fopt 4.0 164 165#define INPUT_WIRE_TO_INPUT_GATE_CAP_RATIO 0 166#define BUFFER_SEPARATION_LENGTH_MULTIPLIER 1 167#define NUMBER_MATS_PER_REDUNDANT_MAT 8 168 169#define NUMBER_STACKED_DIE_LAYERS 1 170 171// this variable can be set to carry out solution optimization for 172// a maximum area allocation. 173#define STACKED_DIE_LAYER_ALLOTED_AREA_mm2 0 //6.24 //6.21//71.5 174 175// this variable can also be employed when solution optimization 176// with maximum area allocation is carried out. 177#define MAX_PERCENT_AWAY_FROM_ALLOTED_AREA 50 178 179// this variable can also be employed when solution optimization 180// with maximum area allocation is carried out. 181#define MIN_AREA_EFFICIENCY 20 182 183// this variable can be employed when solution with a desired 184// aspect ratio is required. 185#define STACKED_DIE_LAYER_ASPECT_RATIO 1 186 187// this variable can be employed when solution with a desired 188// aspect ratio is required. 189#define MAX_PERCENT_AWAY_FROM_ASPECT_RATIO 101 190 191// this variable can be employed to carry out solution optimization 192// for a certain target random cycle time. 193#define TARGET_CYCLE_TIME_ns 1000000000 194 195#define NUMBER_PIPELINE_STAGES 4 196 197// this can be used to model the length of interconnect 198// between a bank and a crossbar 199#define LENGTH_INTERCONNECT_FROM_BANK_TO_CROSSBAR 0 //3791 // 2880//micron 200 201#define IS_CROSSBAR 0 202#define NUMBER_INPUT_PORTS_CROSSBAR 8 203#define NUMBER_OUTPUT_PORTS_CROSSBAR 8 204#define NUMBER_SIGNALS_PER_PORT_CROSSBAR 256 205 206 207#define MAT_LEAKAGE_REDUCTION_DUE_TO_SLEEP_TRANSISTORS_FACTOR 1 208#define LEAKAGE_REDUCTION_DUE_TO_LONG_CHANNEL_HP_TRANSISTORS_FACTOR 1 209 210#define PAGE_MODE 0 211 212#define MAIN_MEM_PER_CHIP_STANDBY_CURRENT_mA 60 213// We are actually not using this variable in the CACTI code. We just want to acknowledge that 214// this current should be multiplied by the DDR(n) system VDD value to compute the standby power 215// consumed during precharge. 216 217 218const double VDD_STORAGE_LOSS_FRACTION_WORST = 0.125; 219const double CU_RESISTIVITY = 0.022; //ohm-micron 220const double BULK_CU_RESISTIVITY = 0.018; //ohm-micron 221const double PERMITTIVITY_FREE_SPACE = 8.854e-18; //F/micron 222 223const static uint32_t sram_num_cells_wl_stitching_ = 16; 224const static uint32_t dram_num_cells_wl_stitching_ = 64; 225const static uint32_t comm_dram_num_cells_wl_stitching_ = 256; 226const static double num_bits_per_ecc_b_ = 8.0; 227 228const double bit_to_byte = 8.0; 229 230#define MAX_NUMBER_GATES_STAGE 20 231#define MAX_NUMBER_HTREE_NODES 20 232#define NAND2_LEAK_STACK_FACTOR 0.2 233#define NAND3_LEAK_STACK_FACTOR 0.2 234#define NOR2_LEAK_STACK_FACTOR 0.2 235#define INV_LEAK_STACK_FACTOR 0.5 236#define MAX_NUMBER_ARRAY_PARTITIONS 1000000 237 238// abbreviations used in this project 239// ---------------------------------- 240// 241// num : number 242// rw : read/write 243// rd : read 244// wr : write 245// se : single-ended 246// sz : size 247// F : feature 248// w : width 249// h : height or horizontal 250// v : vertical or velocity 251 252 253enum ram_cell_tech_type_num { 254 itrs_hp = 0, 255 itrs_lstp = 1, 256 itrs_lop = 2, 257 lp_dram = 3, 258 comm_dram = 4 259}; 260 261const double pppm[4] = {1, 1, 1, 1}; 262const double pppm_lkg[4] = {0, 1, 1, 0}; 263const double pppm_dyn[4] = {1, 0, 0, 0}; 264const double pppm_Isub[4] = {0, 1, 0, 0}; 265const double pppm_Ig[4] = {0, 0, 1, 0}; 266const double pppm_sc[4] = {0, 0, 0, 1}; 267 268 269 270#endif 271