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
|
# Template Input File for DBCSR Performance Driver
# -----------------------------------------------------------------------
# npcols MPI grid
# - 0 leaves MPI to find the best grid.
# - Note that the total number of processors must be divisible per npcols
0
# use MPI-RMA
F
# operation
dbcsr_multiply
# matrix sizes (M, N, K)
1000
1000
1000
# sparsity (matrix A, matrix B, matrix C)
0.0d0
0.0d0
0.0d0
# transposes
N
N
# symmetries
N
N
N
# data type
# - 3: double
# - other types, see "Type definitions" in dbcsr/src/data/dbcsr_data_types.F
3
# alpha (real, imag)
1.0d0
0.0d0
# beta (real, imag)
1.0d0
0.0d0
# limits
# - 0 means full size
# - row
# -- limRowL (First full row of limiting submatrix)
0
# -- limRowU
0
# - col
# -- limColL (First full col of limiting submatrix)
0
# -- limColU
0
# - k
# -- limKL first full col of imiting inner product)
0
# -- limKU
0
# retain sparsity (T/F)
F
# number of repetitions
1
# number of different blocks to read (m, n, k)
1
1
1
# the m blocks (multiplicity, block size, ...)
# - this configuration, eg, will generate blocks of
# - size 5 in the m-dimension
1
5
# - this configuration would generate a block of
# - size 5 followed by 2 blocks of size 3 in the m-dimension,
# - followed by a block of size 5, etc. until size M is reached
# - 1
# - 5
# - 2
# - 3
# the n blocks (multiplicity, block size, ...)
1
5
# the k blocks (multiplicity, block size, ...)
1
5
# checksum (check, threshold, references)
F
0.
0.
0.
|
