← Back to branch summary

~ubuntu-branches/ubuntu/hoary/kdemultimedia/hoary

« back to all changes in this revision

Viewing changes to mpeglib/lib/splay/op.h

  • Committer: Bazaar Package Importer
  • Author(s): Martin Schulze
  • Date: 2003-01-22 15:00:51 UTC
  • Revision ID: james.westby@ubuntu.com-20030122150051-uihwkdoxf15mi1tn
Tags: upstream-2.2.2
Import upstream version 2.2.2

Show diffs side-by-side

added added

removed removed

Lines of Context:
mpeglib/lib/splay/op.h
 
1
/*
 
2
  unrolled operations, for better Pentium FPU scheduling
 
3
  Copyright (C) 2001  Martin Vogt
 
4
 
 
5
  This program is free software; you can redistribute it and/or modify
 
6
  it under the terms of the GNU Library General Public License as published by
 
7
  the Free Software Foundation.
 
8
 
 
9
  For more information look at the file COPYRIGHT in this package
 
10
 
 
11
 */
 
12
 
 
13
 
 
14
 
 
15
#ifndef __OP_H
 
16
#define __OP_H
 
17
 
 
18
/**
 
19
   The Pentium has two pipelined FPUs which makes it possible
 
20
   to do two operations in one cycle.
 
21
   (If you are lucky)
 
22
 
 
23
*/
 
24
 
 
25
#define PTR_DIST (1024)
 
26
 
 
27
#define OS r1=vp1[0] * dp[0]; \
 
28
           r2=vp1[PTR_DIST-0] * dp[0]; \
 
29
           dp++;
 
30
 
 
31
#define XX1 vp1+=15;dp++;
 
32
 
 
33
#define XX2 r1+=vp1[0] * dp[-1]; \
 
34
            r2+=vp1[PTR_DIST-0] * dp[-1];
 
35
 
 
36
#define OP_END(val) vp1-=val;dp+=val;
 
37
 
 
38
#define OP_END_1(vVal,dVal) vp1+=(vVal-dVal),dp+=dVal
 
39
 
 
40
 
 
41
// this is OP_END(x);XX1; together:
 
42
#define OP_END_2(vVal)      vp1+=(15-vVal),dp+=vVal+1
 
43
 
 
44
 
 
45
// check this to test pipelining
 
46
 
 
47
#define SCHEDULE1(op,r1,r2)       r1;op;r2;
 
48
#define SCHEDULE2(op,r1,r2)       op;r1;r2;
 
49
 
 
50
#define SCHEDULE(a,b,c)    SCHEDULE2(a,b,c);
 
51
 
 
52
 
 
53
#define OP r1+=vp1[-1] * dp[0]; \
 
54
           r2+=vp1[PTR_DIST-1] * dp[0];
 
55
 
 
56
 
 
57
 
 
58
#define OP2 SCHEDULE(OP   ,r1+=vp1[-2]  * dp[1] ,r2+=vp1[PTR_DIST-2]  *dp[1]);
 
59
#define OP3 SCHEDULE(OP2  ,r1+=vp1[-3]  * dp[2] ,r2+=vp1[PTR_DIST-3]  *dp[2]);
 
60
#define OP4 SCHEDULE(OP3  ,r1+=vp1[-4]  * dp[3] ,r2+=vp1[PTR_DIST-4]  *dp[3]);
 
61
#define OP5 SCHEDULE(OP4  ,r1+=vp1[-5]  * dp[4] ,r2+=vp1[PTR_DIST-5]  *dp[4]);
 
62
#define OP6 SCHEDULE(OP5  ,r1+=vp1[-6]  * dp[5] ,r2+=vp1[PTR_DIST-6]  *dp[5]);
 
63
#define OP7 SCHEDULE(OP6  ,r1+=vp1[-7]  * dp[6] ,r2+=vp1[PTR_DIST-7]  *dp[6]);
 
64
#define OP8 SCHEDULE(OP7  ,r1+=vp1[-8]  * dp[7] ,r2+=vp1[PTR_DIST-8]  *dp[7]);
 
65
#define OP9 SCHEDULE(OP8  ,r1+=vp1[-9]  * dp[8] ,r2+=vp1[PTR_DIST-9]  *dp[8]);
 
66
#define OP10 SCHEDULE(OP9 ,r1+=vp1[-10] * dp[9] ,r2+=vp1[PTR_DIST-10] *dp[9]);
 
67
#define OP11 SCHEDULE(OP10,r1+=vp1[-11] * dp[10],r2+=vp1[PTR_DIST-11] *dp[10]);
 
68
#define OP12 SCHEDULE(OP11,r1+=vp1[-12] * dp[11],r2+=vp1[PTR_DIST-12] *dp[11]);
 
69
#define OP13 SCHEDULE(OP12,r1+=vp1[-13] * dp[12],r2+=vp1[PTR_DIST-13] *dp[12]);
 
70
#define OP14 SCHEDULE(OP13,r1+=vp1[-14] * dp[13],r2+=vp1[PTR_DIST-14] *dp[13]);
 
71
#define OP15 SCHEDULE(OP14,r1+=vp1[-15] * dp[14],r2+=vp1[PTR_DIST-15] *dp[14]);
 
72
 
 
73
 
 
74
/*
 
75
#define OP r1+=vp1[-1] * dp[0]; \
 
76
           r2+=vp2[-1] * dp[0];
 
77
 
 
78
 
 
79
 
 
80
#define OP2 SCHEDULE(OP   ,r1+=vp1[-2]  * dp[1] ,r2+=vp2[-2]  * dp[1]);
 
81
#define OP3 SCHEDULE(OP2  ,r1+=vp1[-3]  * dp[2] ,r2+=vp2[-3]  * dp[2]);
 
82
#define OP4 SCHEDULE(OP3  ,r1+=vp1[-4]  * dp[3] ,r2+=vp2[-4]  * dp[3]);
 
83
#define OP5 SCHEDULE(OP4  ,r1+=vp1[-5]  * dp[4] ,r2+=vp2[-5]  * dp[4]);
 
84
#define OP6 SCHEDULE(OP5  ,r1+=vp1[-6]  * dp[5] ,r2+=vp2[-6]  * dp[5]);
 
85
#define OP7 SCHEDULE(OP6  ,r1+=vp1[-7]  * dp[6] ,r2+=vp2[-7]  * dp[6]);
 
86
#define OP8 SCHEDULE(OP7  ,r1+=vp1[-8]  * dp[7] ,r2+=vp2[-8]  * dp[7]);
 
87
#define OP9 SCHEDULE(OP8  ,r1+=vp1[-9]  * dp[8] ,r2+=vp2[-9]  * dp[8]);
 
88
#define OP10 SCHEDULE(OP9 ,r1+=vp1[-10] * dp[9] ,r2+=vp2[-10] * dp[9]);
 
89
#define OP11 SCHEDULE(OP10,r1+=vp1[-11] * dp[10],r2+=vp2[-11] * dp[10]);
 
90
#define OP12 SCHEDULE(OP11,r1+=vp1[-12] * dp[11],r2+=vp2[-12] * dp[11]);
 
91
#define OP13 SCHEDULE(OP12,r1+=vp1[-13] * dp[12],r2+=vp2[-13] * dp[12]);
 
92
#define OP14 SCHEDULE(OP13,r1+=vp1[-14] * dp[13],r2+=vp2[-14] * dp[13]);
 
93
#define OP15 SCHEDULE(OP14,r1+=vp1[-15] * dp[14],r2+=vp2[-15] * dp[14]);
 
94
*/
 
95
 
 
96
#endif