-
Notifications
You must be signed in to change notification settings - Fork 125
/
TUNING
48 lines (34 loc) · 1.85 KB
/
TUNING
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
==============================================================
Performance tuning and setting up the input data file hpcg.dat
==============================================================
Current as of release HPCG - 3.1 - March 11, 2019
Check the website http://hpcg-benchmark.org/ for the latest
information.
After building the executable hpcg_build/bin/xhpcg, (where hpcg_build is your
build directory for HPCG) one may want to modify the
input data file hpcg.dat. This file should reside in the same directory
as the executable hpcg_build/bin/xhpcg. An example hpcg.dat file is provided
by default. This file contains information about the problem sizes,
machine configuration, and algorithm features to be used by the executable.
It is 4 lines long. All the selected parameters will be printed in the
output generated by the executable.
================================
Description of the hpcg.dat file
================================
* Line 1: (unused) Typically one would use this line for its own good. For
example, it could be used to summarize the con- tent of the input file. By
default this line reads:
HPCG benchmark input file
* Line 2: (unused) same as Line 1. By default, this line reads:
Sandia National Laboratories; University of Tennessee, Knoxville
* Line 3: This line specifies the local (to an MPI process) dimensions of the
problem. By default, this line reads:
104 104 104
which means that each MPI process will be computing a solution for a cube of
size 104 points.
* Line 4: This line specifies the number of seconds of how long the timed
portion of the benchmark should run. By default, this line reads:
60
which means that the timed portion of the benchmark will run 1 minute.
This length of time is not sufficient for submitting an official run
but does give sufficient data for tuning the benchmark in most cases.