source: README @ 5df7203

Revision 5df7203, 3.4 KB checked in by Hal Finkel <hfinkel@…>, 7 years ago (diff)

add glob processing

  • Property mode set to 100644
Line 
1memlog - A Memory-Allocation Logging Tool
2
3This tool attempts to help you answer the question:
4  Why is my application using so much memory?
5
6** LINKING **
7
8How to use it depends on how your application is linked:
9
10For dynamically-linked applications, you can:
11
12 1. Use LD_PRELOAD: Set LD_PRELOAD=/path/to/memlog/libmemlog.so when you run
13    your application.
14
15 2. Link directly: Add the following to your linker flags:
16      -L/path/to/memlog -Wl,-rpath,/path/to/memlog -lmemlog
17
18For statically-linked applications, add the following to your linker flags:
19
20  -Wl,--wrap,malloc,--wrap,free,--wrap,realloc,--wrap,calloc,--wrap,memalign \
21    /path/to/memlog/memlog_s.o -lpthread -ldl
22
23** RUNNING **
24
25When your application runs, you should find in your current directory files
26named 'HOST.PID.memlog', one for each process. These contain the raw tracing
27information, and are only somewhat human readable. You can create a ps/pdf
28file detailing the memory allocated when each process reached its peak memory
29use by running:
30
31  /path/to/memlog/memlog_analyze /path/to/HOST.PID.memlog
32
33this will generate files named HOST.PID.memlog.dot, HOST.PID.memlog.ps and
34HOST.PID.memlog.pdf. You'll probably find the pdf file most convenient for
35viewing. HOST.PID.memlog.txt is also generated, providing the same information
36in textual form.
37
38If you pass the --leaks option to memlog_analyze, it will provide data on
39allocations active at the end of the program (leaks) instead of those active
40when the peak memory usage is first reached.
41
42You might have many runs of the same application (or output from many ranks of
43an MPI job), and you'd like to pick the one for analysis with the highest
44memory usage. If you provide a glob pattern to memlog_analyze it will do this
45for you. Make sure you quote the glob pattern so that your shell does not
46expand it.
47
48  /path/to/memlog/memlog_analyze "/path/to/*.memlog"
49
50When running under common batch systems, the files are named
51JOB_ID.HOST.PID.memlog, and when running under the BG/Q CNK, the process's rank
52is used instead of the node-local PID.
53
54Note that te peak memory usage is determined by monitoring the processes's
55maximum resident set size, not just the total allocated heap memory.
56
57memlog_analyze depends on dot (from the graphviz package) and ps2pdf (from the
58ghostscript package), plus various tools from the binutils package.
59
60** RELATED WORK **
61
62Why was memlog created? There are several other tools that can support this use
63case, but none of them would work in our environment properly. They were
64either too slow, not runnable under the BG/Q CNK, not thread safe, did not
65properly support big-endian PPC64, supported only either static or dynamic
66linking, did not collect full backtraces, or just did not produce
67sufficiently-informative peak-usage output.
68
69That having been said, some other tools that might interest you:
70  Valgrind Massif - http://valgrind.org/docs/manual/ms-manual.html
71  Google Performance Tools - http://google-perftools.googlecode.com/svn/trunk/doc/heapprofile.html
72  memtrail - https://github.com/jrfonseca/memtrail
73  LeakTracer - http://www.andreasen.org/LeakTracer/
74  glibc mtrace - http://www.gnu.org/s/hello/manual/libc/Allocation-Debugging.html
75  Heaptrack - http://milianw.de/blog/heaptrack-a-heap-memory-profiler-for-linux
76  MemProf - http://www.secretlabs.de/projects/memprof/
77
78The dot/pdf output produced by memlog was definitely inspired by that produced
79by Google's pprof tool in the aforementioned package.
80
Note: See TracBrowser for help on using the repository browser.