GNU Octave  4.4.1
A high-level interpreted language, primarily intended for numerical computations, mostly compatible with Matlab
sdot3.f
Go to the documentation of this file.
1 c Copyright (C) 2009-2018 VZLU Prague, a.s., Czech Republic
2 c
3 c Author: Jaroslav Hajek <highegg@gmail.com>
4 c
5 c This file is part of Octave.
6 c
7 c Octave is free software: you can redistribute it and/or modify it
8 c under the terms of the GNU General Public License as published by
9 c the Free Software Foundation, either version 3 of the License, or
10 c (at your option) any later version.
11 c
12 c Octave is distributed in the hope that it will be useful, but
13 c WITHOUT ANY WARRANTY; without even the implied warranty of
14 c MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 c GNU General Public License for more details.
16 c
17 c You should have received a copy of the GNU General Public License
18 c along with Octave; see the file COPYING. If not, see
19 c <https://www.gnu.org/licenses/>.
20 c
21  subroutine sdot3(m,n,k,a,b,c)
22 c purpose: a 3-dimensional dot product.
23 c c = sum (a .* b, 2), where a and b are 3d arrays.
24 c arguments:
25 c m,n,k (in) the dimensions of a and b
26 c a,b (in) real input arrays of size (m,k,n)
27 c c (out) real output array, size (m,n)
28  integer m,n,k,i,j,l
29  real a(m,k,n),b(m,k,n)
30  real c(m,n)
31 
32  real sdot
33  external sdot
34 
35 c quick return if possible.
36  if (m <= 0 .or. n <= 0) return
37 
38  if (m == 1) then
39 c the column-major case.
40  do j = 1,n
41  c(1,j) = sdot(k,a(1,1,j),1,b(1,1,j),1)
42  end do
43  else
44 c We prefer performance here, because that's what we generally
45 c do by default in reduction functions. Besides, the accuracy
46 c of xDOT is questionable. Hence, do a cache-aligned nested loop.
47  do j = 1,n
48  do i = 1,m
49  c(i,j) = 0d0
50  end do
51  do l = 1,k
52  do i = 1,m
53  c(i,j) = c(i,j) + a(i,l,j)*b(i,l,j)
54  end do
55  end do
56  end do
57  end if
58 
59  end subroutine
subroutine sdot3(m, n, k, a, b, c)
Definition: sdot3.f:22