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
|
This directory contains a benchmarking and testing program
for fftw3.
The `bench' program has a zillion options, because we use it for
benchmarking other FFT libraries as well. This file only documents
the basic usage of bench.
Usage: bench <commands>
where each command is as follows:
-s <problem>
--speed <problem>
Benchmarks the speed of <problem>.
The syntax for problems is [i|o][r|c][f|b]<size>, where
i/o means in-place or out-of-place. In-place is the default
r/c means real or complex transform. Complex is the default
f/b means forward or backward transform. Forward is the default.
<size> is an arbitrary multidimensional sequence of integers
separated by the character 'x'.
(The syntax for problems is actually richer, but we do not document
it here. See the man page for fftw-wisdom for more information.)
Example:
ib256 : in-place backward complex transform of size 256
32x64 : out-of-place forward complex 2D transform of 32 rows
and 64 columns.
-y <problem>
--verify <problem>
Verify that FFTW is computing the correct answer.
The program does not output anything unless an error occurs or
verbosity is at least one.
-v<n>
Set verbosity to <n>, or 1 if <n> is omitted.
-oestimate
-opatient
-oexhaustive
Plan with FFTW_ESTIMATE, FFTW_PATIENT, or FFTW_EXHAUSTIVE, respectively.
The default is FFTW_MEASURE.
If you benchmark FFTW, please use -opatient.
|
