Error with powertrace.py script in Sniper 5.3

851 views
Skip to first unread message

kalenji Catali

unread,
Jan 22, 2014, 11:02:29 AM1/22/14
to snip...@googlegroups.com
Hi,


When I configure the fft test example you include in Sniper 5.3 with powertrace.py script I got the following error:


../../run-sniper -n 2 -c gainestown -s powertrace  --roi --viz --viz-aso -- ./fft -p 2
[SNIPER] Start
Executing Python script /home/jabellan/SNIPER/sniper-5.3/test/fft/sim.scripts.py
[SNIPER] --------------------------------------------------------------------------------
[SNIPER] Sniper using Pin frontend
[SNIPER] Running pre-ROI region in  CACHE_ONLY mode
[SNIPER] Running application ROI in DETAILED mode
[SNIPER] Running post-ROI region in FAST_FORWARD mode
[SNIPER] --------------------------------------------------------------------------------

FFT with Blocking Transpose
   1024 Complex Doubles
   2 Processors
   65536 Cache lines
   16 Byte line size
   4096 Bytes per page

[SNIPER] Enabling performance models
[SNIPER] Setting instrumentation mode to DETAILED
[SNIPER] Disabling performance models
[SNIPER] Leaving ROI after 2.89 seconds
[SNIPER] Simulated 1.0M instructions, 0.5M cycles, 2.10 IPC
[SNIPER] Simulation speed 351.0 KIPS (175.5 KIPS / target core - 5698.3ns/instr)
[SNIPER] Setting instrumentation mode to FAST_FORWARD

                 PROCESS STATISTICS
            Computation      Transpose     Transpose
 Proc          Time            Time        Fraction
    0               157             22       0.14013

                 TIMING INFORMATION
Start time                        :      -1844408320
Initialization finish time        :      -1844408297
Overall finish time               :      -1844408140
Total time with initialization    :              180
Total time without initialization :              157
Overall transpose time            :               22
Overall transpose fraction        :          0.14013

/home/jabellan//SNIPER/sniper-5.3/tools/../mcpat/mcpat-1.0.cache: error while loading shared libraries: libdb-4.3.so: cannot open shared object file: No such file or directory
Traceback (most recent call last):
  File "/home/jabellan//SNIPER/sniper-5.3/tools/mcpat.py", line 1281, in <module>
    main(jobid = jobid, resultsdir = resultsdir, powertype = powertype, config = config, outputfile = outputfile, no_graph = no_graph, print_stack = not no_text, partial = partial)
  File "/home/jabellan//SNIPER/sniper-5.3/tools/mcpat.py", line 175, in main
    raise ValueError('No valid McPAT output found')
ValueError: No valid McPAT output found
[SNIPER] End
[SNIPER] Elapsed time: 4.71 seconds
[SNIPER] Generating visualization in viz/


Optional: Run '../../tools/cpistack.py' in this directory to generate cpi-stack output for this run
Optional: Run '../../tools/mcpat.py' in this directory to generate power output for this run
Optional: Run '../../tools/dumpstats.py' in this directory to view detailed statistics for this run
Optional: Run '../../tools/gen_topology.py' in this directory to view the system topology for this run

Any idea?


Thanks a lot!

kalenji Catali

unread,
Jan 22, 2014, 11:19:06 AM1/22/14
to snip...@googlegroups.com
Now, it works in another machine I have. It was a problem of libraries...

However, I would like to know if there is another script similar to ipctrace.py where power of cores are shown every N nanoseconds of execution.
Do you have this kind of script? If not, how could I implement this periodic sampling of cores' power?

Thanks a lot!

Wim Heirman

unread,
Jan 23, 2014, 5:22:17 AM1/23/14
to snip...@googlegroups.com
Kalenji,

libdb-4.3.so should have been downloaded together with the mcpat executable in sniper/mcpat. Maybe there is an LD_LIBRARY_PATH problem when running it through powertrace.py

But in any case, powertrace.py is pretty old, we haven't used it for a while. Instead, take a look at energystats.py which also periodically runs mcpat and makes the results available as regular statistics -- so they can be logged using periodic-stats, or read by other Python code using e.g. sim.stats.get('core', 0, 'energy-dynamic'). It should also support per-core DVFS and pass the right voltage/frequency to mcpat. If you want a quick trace you can combine this with the stattrace script: the command below prints out each core's dynamic energy consumption in intervals of 10 us:

./run-sniper -n4 --roi -senergystats -sstattrace:core.energy-dynamic -- ./test/fft/fft -p4

I'm not sure about the units but I think it will be femtoJoules per nanoseconds (femtoJoules is the unit used by energystats.py which also applies to what you'll find in dumpstats, stattrace.py normalizes everything to be the statistic's delta per nanosecond).

Regards,
Wim




--
--
--
You received this message because you are subscribed to the Google
Groups "Sniper simulator" group.
To post to this group, send email to snip...@googlegroups.com
To unsubscribe from this group, send email to
snipersim+...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/snipersim?hl=en
 
---
You received this message because you are subscribed to the Google Groups "Sniper simulator" group.
To unsubscribe from this group and stop receiving emails from it, send an email to snipersim+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

kalenji Catali

unread,
Jan 23, 2014, 10:30:42 AM1/23/14
to snip...@googlegroups.com
Thank you very much!!
Regards,
Kalenji

kalenji Catali

unread,
Jan 24, 2014, 1:12:15 PM1/24/14
to snip...@googlegroups.com
Hi Wim,

I would like to know if the mcpat version installed in Sniper5.3 was optimized to reduce simulation time. 
If not, I don't know how to apply the patch sniper-mcpat-1.0.patch for CACTI optimization.

Thank you!

Regards,
Jose

Trevor E. Carlson

unread,
Jan 24, 2014, 2:35:04 PM1/24/14
to snip...@googlegroups.com, kalenji Catali
Kalenji,

Sniper 5.3 includes McPAT that is optimized to cache the cacti results. If you add the --power option to the sniper/test/fft Makefile, and run make twice, you will see that the second run is much faster.

Definitely get back to us if you run into any issues,
Trevor

Ghizlane Mouslih

unread,
May 15, 2015, 2:35:28 AM5/15/15
to snip...@googlegroups.com
Hi,

Please, I'm using the latest version of Sniper and i want to print out each core's dynamic energy consumption. I used this command 

./run-sniper -n4 --roi -senergystats -sstattrace:core.energy-dynamic -- ./test/fft/fft -p4

And i got the following error :

Run benchmark under the Sniper simulator
Usage:
  ./run-sniper  -p <program>  -i <inputsize (test)> -n <ncores (1)>  -m <machines (1)>  -d <outputdir (.)>  -c <config-file>  -r <sniper-root-dir>  -g <options>
Benchmarks:
  cpu2006:
    perlbench bzip2 gcc bwaves gamess mcf milc zeusmp gromacs cactusADM leslie3d namd gobmk dealII soplex povray calculix hmmer sjeng GemsFDTD libquantum h264ref tonto lbm omnetpp astar wrf sphinx3 xalancbmk
  jikes:
    lusearch jython antlr pmd avrora fop luindex xalan bloat eclipse lusearch.fix hsqldb sunflow pjbb2005
  local:
    pi

Traceback (most recent call last):
  File "./run-sniper", line 235, in <module>
    usage()
  File "./run-sniper", line 26, in usage
    module = __import__(module)
  File "/home/yuanqing/research/sniper-6.1/benchmarks/npb/__init__.py", line 8, in <module>
    OPTIONS = eval(open(os.path.join(HOME, '../tools/hooks', 'buildconf.py')).read())
IOError: [Errno 2] No such file or directory: '/home/yuanqing/research/sniper-6.1/benchmarks/npb/../tools/hooks/buildconf.py'


Can you help please?

Thanks in advance.

Ghizlane

Trevor Carlson

unread,
May 15, 2015, 3:11:15 AM5/15/15
to snip...@googlegroups.com, Ghizlane Mouslih
Ghzilane,

It appears that you are mixing the Sniper commands with the Benchmarks directory. The run-sniper command in $BENCHMARKS_ROOT allow you to run the benchmarks included (or added) in the benchmarks distribution. Instead, you will want to use the run-sniper command in $SNIPER_ROOT, and this should work.

$ cd ${SNIPER_ROT}
$ ./run-sniper -n4 --roi -senergystats -- ./test/fft/fft -p4

Trevor

-- 
-- 
--
You received this message because you are subscribed to the Google
Groups "Sniper simulator" group.
To post to this group, send email to snip...@googlegroups.com
To unsubscribe from this group, send email to
snipersim+...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/snipersim?hl=en

--- 
You received this message because you are subscribed to the Google Groups "Sniper simulator" group.
To unsubscribe from this group and stop receiving emails from it, send an email to snipersim+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Ghizlane Mouslih

unread,
May 15, 2015, 4:02:39 AM5/15/15
to snip...@googlegroups.com, ghizlan...@gmail.com
Hi Trevor,

You are right, i was mixing the sniper command with the benchmark directory as i want to see the energy consumption while running the benchmark. I have an other question please, can i vizualise the output by using --viz ?
Reply all
Reply to author
Forward
0 new messages