File: engine_sse.cpp

package info (click to toggle)
openems 0.0.35+dfsg.1-3
  • links: PTS, VCS
  • area: main
  • in suites: buster
  • size: 6,288 kB
  • sloc: cpp: 40,259; yacc: 580; lex: 350; makefile: 258; sh: 169; ruby: 19
file content (179 lines) | stat: -rw-r--r-- 7,662 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
/*
*	Copyright (C) 2010 Thorsten Liebig (Thorsten.Liebig@gmx.de)
*
*	This program is free software: you can redistribute it and/or modify
*	it under the terms of the GNU General Public License as published by
*	the Free Software Foundation, either version 3 of the License, or
*	(at your option) any later version.
*
*	This program is distributed in the hope that it will be useful,
*	but WITHOUT ANY WARRANTY; without even the implied warranty of
*	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
*	GNU General Public License for more details.
*
*	You should have received a copy of the GNU General Public License
*	along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/

#ifndef SSE_CORRECT_DENORMALS
#include <xmmintrin.h>
#endif

#include "engine_sse.h"

//! \brief construct an Engine_sse instance
//! it's the responsibility of the caller to free the returned pointer
Engine_sse* Engine_sse::New(const Operator_sse* op)
{
	cout << "Create FDTD engine (SSE)" << endl;
	Engine_sse* e = new Engine_sse(op);
	e->Init();
	return e;
}

Engine_sse::Engine_sse(const Operator_sse* op) : Engine(op)
{
	m_type = SSE;
	Op = op;
	f4_volt = 0;
	f4_curr = 0;
	numVectors =  ceil((double)numLines[2]/4.0);

	// speed up the calculation of denormal floating point values (flush-to-zero)
#ifndef SSE_CORRECT_DENORMALS
	unsigned int oldMXCSR = _mm_getcsr(); //read the old MXCSR setting
	unsigned int newMXCSR = oldMXCSR | 0x8040; // set DAZ and FZ bits
	_mm_setcsr( newMXCSR ); //write the new MXCSR setting to the MXCSR
#endif
}

Engine_sse::~Engine_sse()
{
	//_mm_setcsr( oldMXCSR ); // restore old setting
	Reset();
}

void Engine_sse::Init()
{
	Engine::Init();

	Delete_N_3DArray(volt,numLines);
	volt=NULL; // not used
	Delete_N_3DArray(curr,numLines);
	curr=NULL; // not used

	f4_volt = Create_N_3DArray_v4sf(numLines);
	f4_curr = Create_N_3DArray_v4sf(numLines);
}

void Engine_sse::Reset()
{
	Engine::Reset();
	Delete_N_3DArray_v4sf(f4_volt,numLines);
	f4_volt = 0;
	Delete_N_3DArray_v4sf(f4_curr,numLines);
	f4_curr = 0;
}

void Engine_sse::UpdateVoltages(unsigned int startX, unsigned int numX)
{
	unsigned int pos[3];
	bool shift[2];
	f4vector temp;

	pos[0] = startX;
	for (unsigned int posX=0; posX<numX; ++posX)
	{
		shift[0]=pos[0];
		for (pos[1]=0; pos[1]<numLines[1]; ++pos[1])
		{
			shift[1]=pos[1];
			for (pos[2]=1; pos[2]<numVectors; ++pos[2])
			{
				// x-polarization
				f4_volt[0][pos[0]][pos[1]][pos[2]].v *= Op->f4_vv[0][pos[0]][pos[1]][pos[2]].v;
				f4_volt[0][pos[0]][pos[1]][pos[2]].v += Op->f4_vi[0][pos[0]][pos[1]][pos[2]].v * ( f4_curr[2][pos[0]][pos[1]][pos[2]].v - f4_curr[2][pos[0]][pos[1]-shift[1]][pos[2]].v - f4_curr[1][pos[0]][pos[1]][pos[2]].v + f4_curr[1][pos[0]][pos[1]][pos[2]-1].v );

				// y-polarization
				f4_volt[1][pos[0]][pos[1]][pos[2]].v *= Op->f4_vv[1][pos[0]][pos[1]][pos[2]].v;
				f4_volt[1][pos[0]][pos[1]][pos[2]].v += Op->f4_vi[1][pos[0]][pos[1]][pos[2]].v * ( f4_curr[0][pos[0]][pos[1]][pos[2]].v - f4_curr[0][pos[0]][pos[1]][pos[2]-1].v - f4_curr[2][pos[0]][pos[1]][pos[2]].v + f4_curr[2][pos[0]-shift[0]][pos[1]][pos[2]].v);

				// z-polarization
				f4_volt[2][pos[0]][pos[1]][pos[2]].v *= Op->f4_vv[2][pos[0]][pos[1]][pos[2]].v;
				f4_volt[2][pos[0]][pos[1]][pos[2]].v += Op->f4_vi[2][pos[0]][pos[1]][pos[2]].v * ( f4_curr[1][pos[0]][pos[1]][pos[2]].v - f4_curr[1][pos[0]-shift[0]][pos[1]][pos[2]].v - f4_curr[0][pos[0]][pos[1]][pos[2]].v + f4_curr[0][pos[0]][pos[1]-shift[1]][pos[2]].v);
			}

			// for pos[2] = 0
			// x-polarization
			temp.f[0] = 0;
			temp.f[1] = f4_curr[1][pos[0]][pos[1]][numVectors-1].f[0];
			temp.f[2] = f4_curr[1][pos[0]][pos[1]][numVectors-1].f[1];
			temp.f[3] = f4_curr[1][pos[0]][pos[1]][numVectors-1].f[2];
			f4_volt[0][pos[0]][pos[1]][0].v *= Op->f4_vv[0][pos[0]][pos[1]][0].v;
			f4_volt[0][pos[0]][pos[1]][0].v += Op->f4_vi[0][pos[0]][pos[1]][0].v * ( f4_curr[2][pos[0]][pos[1]][0].v - f4_curr[2][pos[0]][pos[1]-shift[1]][0].v - f4_curr[1][pos[0]][pos[1]][0].v + temp.v );

			// y-polarization
			temp.f[0] = 0;
			temp.f[1] = f4_curr[0][pos[0]][pos[1]][numVectors-1].f[0];
			temp.f[2] = f4_curr[0][pos[0]][pos[1]][numVectors-1].f[1];
			temp.f[3] = f4_curr[0][pos[0]][pos[1]][numVectors-1].f[2];
			f4_volt[1][pos[0]][pos[1]][0].v *= Op->f4_vv[1][pos[0]][pos[1]][0].v;
			f4_volt[1][pos[0]][pos[1]][0].v += Op->f4_vi[1][pos[0]][pos[1]][0].v * ( f4_curr[0][pos[0]][pos[1]][0].v - temp.v - f4_curr[2][pos[0]][pos[1]][0].v + f4_curr[2][pos[0]-shift[0]][pos[1]][0].v);

			// z-polarization
			f4_volt[2][pos[0]][pos[1]][0].v *= Op->f4_vv[2][pos[0]][pos[1]][0].v;
			f4_volt[2][pos[0]][pos[1]][0].v += Op->f4_vi[2][pos[0]][pos[1]][0].v * ( f4_curr[1][pos[0]][pos[1]][0].v - f4_curr[1][pos[0]-shift[0]][pos[1]][0].v - f4_curr[0][pos[0]][pos[1]][0].v + f4_curr[0][pos[0]][pos[1]-shift[1]][0].v);
		}
		++pos[0];
	}
}

void Engine_sse::UpdateCurrents(unsigned int startX, unsigned int numX)
{
	unsigned int pos[5];
	f4vector temp;

	pos[0] = startX;
	for (unsigned int posX=0; posX<numX; ++posX)
	{
		for (pos[1]=0; pos[1]<numLines[1]-1; ++pos[1])
		{
			for (pos[2]=0; pos[2]<numVectors-1; ++pos[2])
			{
				// x-pol
				f4_curr[0][pos[0]][pos[1]][pos[2]].v *= Op->f4_ii[0][pos[0]][pos[1]][pos[2]].v;
				f4_curr[0][pos[0]][pos[1]][pos[2]].v += Op->f4_iv[0][pos[0]][pos[1]][pos[2]].v * ( f4_volt[2][pos[0]][pos[1]][pos[2]].v - f4_volt[2][pos[0]][pos[1]+1][pos[2]].v - f4_volt[1][pos[0]][pos[1]][pos[2]].v + f4_volt[1][pos[0]][pos[1]][pos[2]+1].v);

				// y-pol
				f4_curr[1][pos[0]][pos[1]][pos[2]].v *= Op->f4_ii[1][pos[0]][pos[1]][pos[2]].v;
				f4_curr[1][pos[0]][pos[1]][pos[2]].v += Op->f4_iv[1][pos[0]][pos[1]][pos[2]].v * ( f4_volt[0][pos[0]][pos[1]][pos[2]].v - f4_volt[0][pos[0]][pos[1]][pos[2]+1].v - f4_volt[2][pos[0]][pos[1]][pos[2]].v + f4_volt[2][pos[0]+1][pos[1]][pos[2]].v);

				// z-pol
				f4_curr[2][pos[0]][pos[1]][pos[2]].v *= Op->f4_ii[2][pos[0]][pos[1]][pos[2]].v;
				f4_curr[2][pos[0]][pos[1]][pos[2]].v += Op->f4_iv[2][pos[0]][pos[1]][pos[2]].v * ( f4_volt[1][pos[0]][pos[1]][pos[2]].v - f4_volt[1][pos[0]+1][pos[1]][pos[2]].v - f4_volt[0][pos[0]][pos[1]][pos[2]].v + f4_volt[0][pos[0]][pos[1]+1][pos[2]].v);
			}

			// for pos[2] = numVectors-1
			// x-pol
			temp.f[0] = f4_volt[1][pos[0]][pos[1]][0].f[1];
			temp.f[1] = f4_volt[1][pos[0]][pos[1]][0].f[2];
			temp.f[2] = f4_volt[1][pos[0]][pos[1]][0].f[3];
			temp.f[3] = 0;
			f4_curr[0][pos[0]][pos[1]][numVectors-1].v *= Op->f4_ii[0][pos[0]][pos[1]][numVectors-1].v;
			f4_curr[0][pos[0]][pos[1]][numVectors-1].v += Op->f4_iv[0][pos[0]][pos[1]][numVectors-1].v * ( f4_volt[2][pos[0]][pos[1]][numVectors-1].v - f4_volt[2][pos[0]][pos[1]+1][numVectors-1].v - f4_volt[1][pos[0]][pos[1]][numVectors-1].v + temp.v);

			// y-pol
			temp.f[0] = f4_volt[0][pos[0]][pos[1]][0].f[1];
			temp.f[1] = f4_volt[0][pos[0]][pos[1]][0].f[2];
			temp.f[2] = f4_volt[0][pos[0]][pos[1]][0].f[3];
			temp.f[3] = 0;
			f4_curr[1][pos[0]][pos[1]][numVectors-1].v *= Op->f4_ii[1][pos[0]][pos[1]][numVectors-1].v;
			f4_curr[1][pos[0]][pos[1]][numVectors-1].v += Op->f4_iv[1][pos[0]][pos[1]][numVectors-1].v * ( f4_volt[0][pos[0]][pos[1]][numVectors-1].v - temp.v - f4_volt[2][pos[0]][pos[1]][numVectors-1].v + f4_volt[2][pos[0]+1][pos[1]][numVectors-1].v);

			// z-pol
			f4_curr[2][pos[0]][pos[1]][numVectors-1].v *= Op->f4_ii[2][pos[0]][pos[1]][numVectors-1].v;
			f4_curr[2][pos[0]][pos[1]][numVectors-1].v += Op->f4_iv[2][pos[0]][pos[1]][numVectors-1].v * ( f4_volt[1][pos[0]][pos[1]][numVectors-1].v - f4_volt[1][pos[0]+1][pos[1]][numVectors-1].v - f4_volt[0][pos[0]][pos[1]][numVectors-1].v + f4_volt[0][pos[0]][pos[1]+1][numVectors-1].v);
		}
		++pos[0];
	}
}