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
|
--- cinterface.tex.orig Thu Mar 10 17:02:44 2005
+++ cinterface.tex Thu Mar 10 17:05:51 2005
@@ -1,3 +1,10 @@
+\documentclass{article}
+\oddsidemargin 0.25in
+\textwidth 6.0in
+\usepackage{fancyvrb}
+
+\begin{document}
+
%
% cinterface.tex
%
@@ -100,13 +107,13 @@
two dimensional arrays, {\tt ORDER}. The standard dictates the following
enumerated types:
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
enum CBLAS_ORDER {CblasRowMajor=101, CblasColMajor=102};
enum CBLAS_TRANSPOSE {CblasNoTrans=111, CblasTrans=112, CblasConjTrans=113};
enum CBLAS_UPLO {CblasUpper=121, CblasLower=122};
enum CBLAS_DIAG {CblasNonUnit=131, CblasUnit=132};
enum CBLAS_SIDE {CblasLeft=141, CblasRight=142};
-\end{verbatim}
+\end{Verbatim}
\subsubsection{Considered methods}
{\it
@@ -227,22 +234,22 @@
{\it
\begin{quotation}
This is the area where use of a structure is most desired. Again, the most
-common suggestion is a structure such as
+common suggestion is a structure such as\\
\verb+struct NON_PORTABLE_COMPLEX {float r; float i;};+.
If one is willing to use this structure throughout one's code, then this
provides a natural and convenient mechanism. If, however, the programmer has
utilized a different structure for complex, this ease of use breaks down. Then,
something like the following code fragment is required:
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
NON_PORTABLE_COMPLEX ctmp;
float cdot[2];
ctmp = cblas_cdotc(n, x, 1, y, 1);
cdot[0] = ctmp.r;
cdot[1] = ctmp.i;
-\end{verbatim}
-which is certainly much less convenient than:
+\end{Verbatim}
+which is certainly much less convenient than:\\
\verb+cblas_cdotc_sub(n, x, 1, y, 1, cdot)+.
It should also be noted that the primary reason for having a function instead
@@ -348,13 +355,13 @@
and column $j$,
we must allocate $m$ pointers, and assign them in a section of code such as:
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
float **A, **subA;
subA = malloc(m*sizeof(float*));
for (k=0; k != m; k++) subA[k] = A[i+k] + j;
cblas_rout(... subA ...);
-\end{verbatim}
+\end{Verbatim}
The same operation must be done if we wish to use a row or column as a vector.
This is not only an inconvenience, but can add up to a non-negligible
@@ -381,35 +388,35 @@
\noindent
{\bf Example 1: making a library call with a C 2D array:}
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
double A[50][25]; /* standard C 2D array */
cblas_rout(CblasRowMajor, ... &A[i][j], 25, ...);
-\end{verbatim}
-
+\end{Verbatim}
+
\noindent
{\bf Example 2: Legal use of pointer to pointer style programming and the CBLAS}
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
double **A, *p;
- A = malloc(M);
+ A = malloc(M*sizeof(double *));
p = malloc(M*N*sizeof(double));
for (i=0; i < M; i++) A[i] = &p[i*N];
cblas_rout(CblasRowMajor, ... &A[i][j], N, ...);
-\end{verbatim}
+\end{Verbatim}
\noindent
{\bf Example 3: Illegal use of pointer to pointer style programming and the CBLAS}
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
double **A, *p;
- A = malloc(M);
+ A = malloc(M*sizeof(double *));
p = malloc(M*N*sizeof(double));
for (i=0; i < M; i++) A[i] = malloc(N*sizeof(double));
cblas_rout(CblasRowMajor, ... &A[i][j], N, ...);
-\end{verbatim}
+\end{Verbatim}
Note that Example 3 is illegal because the rows of A have no guaranteed stride.
@@ -506,7 +513,7 @@
%{\footnotesize {\bf Current Status:}\\
%First vote taken on section~\ref{sec-cblash}, passed 9/0/1 with 16 eligible voters, 8/98.}\\
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
#ifndef CBLAS_H
#define CBLAS_H
#include <stddef.h>
@@ -1072,7 +1079,7 @@
void *C, const int ldc);
#endif
-\end{verbatim}
+\end{Verbatim}
\subsection{Using Fortran 77 BLAS to support row-major BLAS operations}
\label{app-ArrayStore}
@@ -1541,7 +1548,7 @@
rows contain the subdiagonals.
{\samepage
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
------ Super diagonal KU
----------- Super diagonal 2
------------ Super diagonal 1
@@ -1549,7 +1556,7 @@
------------ Sub diagonal 1
----------- Sub diagonal 2
------ Sub diagonal KL
-\end{verbatim}
+\end{Verbatim}
}
If we have a row-major storage, and thus a row-oriented algorithm, we will
@@ -1559,7 +1566,7 @@
to line up with the main diagonal along rows.
{\samepage
-\begin{verbatim}
+\begin{Verbatim}[fontsize=\small,fontfamily=tt,fontshape=rm]
KL D KU
| | | |
| | | | |
@@ -1567,7 +1574,7 @@
| | | | | |
| | | | |
| | | |
-\end{verbatim}
+\end{Verbatim}
}
Now, let us contrast these two storage schemes. Both store
@@ -1675,3 +1682,4 @@
With these ideas in mind, the analysis for the dense routines may be used
unchanged for packed.
\clearpage
+\end{document}
|
