Hi,
For the upcoming release, we have only a small number of machines on which the tests have been run (http://einsteintoolkit.org/release-info/parse_testsuite_results.php). Usually we have many more. It would be very helpful to find out what machines people are actually wanting to use with the toolkit, so that we can focus testing on machines that people actually care about.
Please could you let us know which machines are important for you?
Hi,
I have tried it on Stampede. I think other people also might find it useful.
Also, for quick tests and runs on laptop, I do on my lenovo thinkpad machine.
Best, -Rahul Kashyap
On Tue, Nov 22, 2016 at 5:41 AM, Ian Hinder ian.hinder@aei.mpg.de wrote:
Hi,
For the upcoming release, we have only a small number of machines on which the tests have been run (http://einsteintoolkit.org/ release-info/parse_testsuite_results.php). Usually we have many more. It would be very helpful to find out what machines people are actually wanting to use with the toolkit, so that we can focus testing on machines that people actually care about.
Please could you let us know which machines are important for you?
-- Ian Hinder http://members.aei.mpg.de/ianhin
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On Tue, Nov 22, 2016 at 11:41:38AM +0100, Ian Hinder wrote:
For the upcoming release, we have only a small number of machines on which the tests have been run (http://einsteintoolkit.org/release-info/parse_testsuite_results.php). Usually we have many more. It would be very helpful to find out what machines people are actually wanting to use with the toolkit, so that we can focus testing on machines that people actually care about.
Please could you let us know which machines are important for you?
Here is a list of machines mentioned on the wiki that are in need, grouped but otherwise unsorted:
carver edison gordon shelob stampede supermuc zwicky
We don't have a volunteer for any of them. Not every maintainer has access to those machines, so if you do, please consider helping. If you are actively using some of these machines, please consider helping. If you have used or are thinking about using one of these machines, please consider helping.
Volunteering would mean to run the testsuites (the entire suite) on 1 and 2 mpi processes, using as many openmp threads as makes sense on that machine for one node. Use the standard development checkout of the toolkit (do a fresh one). Then you either sent the resulting two 'summary.log' files to any of the maintainers, or put them on the ET web pages if you do have commit permissions yourself (all devs should have). At this point you can stop, if all testsuites pass. If not - well - there is more work to do. Also take into account that this process might involve updating the simfactory entries, especially the option list.
If we will not find someone who is willing to do the tests on these machines, they will likely not be included in the testing. It is up to you. Also, feel free to test other machines as well, and let us know about how things went. Ideally, send us the resulting log files, and we'll include them. It is always nice for a computing allocation report of some kind to have a sentence ready like: "Our group also ensured that our computational framework (the Einstein Toolkit) included cluster XYZ in it's regular release testing".
Below, I just copy&paste some commands that might come in handy; maybe not always verbatim, but as guide what to do.
Frank
Get the ET thornlist:
GetComponents --parallel \ https://bitbucket.org/einsteintoolkit/manifest/raw/master/einsteintoolkit.th
(I assume you have simfactory configured and know some basics)
Build:
sim build --thornlist=../einsteintoolkit.th
Run like this (or by hand, setting OMP_NUM_THREADS as well):
sim create-submit tests1 --testsuite --procs 20 --num-threads=20 sim create-submit tests2 --testsuite --procs 20 --num-threads=10
(This is for nodes with 20 cores.)
This will create:
simulations/tests[12]/output-0000/TEST/sim/summary.log
The naming convention for the website script is:
NAME__#mpi_#threads.log
Example:
http://einsteintoolkit.org/release-info/qb__1_20.log
These will be parsed (once uploaded) for this web site:
http://einsteintoolkit.org/release-info/parse_testsuite_results.php
The repository for the website is:
https://svn.einsteintoolkit.org/www/
And you can find the files here:
On Tue, Nov 22, 2016 at 11:46 AM, Frank Loeffler knarf@cct.lsu.edu wrote:
On Tue, Nov 22, 2016 at 11:41:38AM +0100, Ian Hinder wrote:
For the upcoming release, we have only a small number of machines on which the tests have been run (http://einsteintoolkit.org/re lease-info/parse_testsuite_results.php). Usually we have many more. It would be very helpful to find out what machines people are actually wanting to use with the toolkit, so that we can focus testing on machines that people actually care about.
Please could you let us know which machines are important for you?
Here is a list of machines mentioned on the wiki that are in need, grouped but otherwise unsorted:
carver edison gordon shelob stampede supermuc zwicky
NERSC's largest machine is Cori. I would add it to the list above, as it is in some way a successor of Edison.
You mention that "Stampede is about to be decommissioned". Do you have a link and a date for this? Without mentioning Stampede 2, that statement alone is confusing; we should be careful since our release announcement will highlight this note.
-erik
On Tue, Nov 22, 2016 at 12:19:50PM -0700, Erik Schnetter wrote:
NERSC's largest machine is Cori. I would add it to the list above, as it is in some way a successor of Edison.
Right. Do you have access to it, and would maybe even volunteer running there?
You mention that "Stampede is about to be decommissioned". Do you have a link and a date for this? Without mentioning Stampede 2, that statement alone is confusing; we should be careful since our release announcement will highlight this note.
When you log in to Stampede, you get this:
--- Stampede 2 is coming --- --- Stampede 1 will start phased decommissioning in January 2017 --- --- The KNLs are now available <> Start migrating today --- https://portal.tacc.utexas.edu/user-news/-/news/101909 https://portal.tacc.utexas.edu/user-guides/stampede#knl
Frank
Hello all,
I am not sure if I can run anything on cori (not sure what system the allocation is good for) but I can log in and have already signed up for edison so can also run on cori.
Yours, Roland
On Tue, Nov 22, 2016 at 12:19:50PM -0700, Erik Schnetter wrote:
NERSC's largest machine is Cori. I would add it to the list above, as >it is in some way a successor of Edison.
Right. Do you have access to it, and would maybe even volunteer running there?
You mention that "Stampede is about to be decommissioned". Do you have >a link and a date for this? Without mentioning Stampede 2, that statement alone is confusing; we should be careful since our release announcement will highlight this note.
When you log in to Stampede, you get this:
--- Stampede 2 is coming --- --- Stampede 1 will start phased decommissioning in January 2017 --- --- The KNLs are now available <> Start migrating today --- https://portal.tacc.utexas.edu/user-news/-/news/101909 https://portal.tacc.utexas.edu/user-guides/stampede#knlFrank
Hi,
Thanks for the reminder that Stampede is to be decommissioned. i haven;t ported to Stampede2 yet. Let me port first and then I'll follow your instructions for the tests.
-Rahul
On Tue, Nov 22, 2016 at 2:36 PM, Roland Haas rhaas@illinois.edu wrote:
Hello all,
I am not sure if I can run anything on cori (not sure what system the allocation is good for) but I can log in and have already signed up for edison so can also run on cori.
Yours, Roland
On Tue, Nov 22, 2016 at 12:19:50PM -0700, Erik Schnetter wrote:
NERSC's largest machine is Cori. I would add it to the list above, as
it is
in some way a successor of Edison.
Right. Do you have access to it, and would maybe even volunteer running
there?
You mention that "Stampede is about to be decommissioned". Do you have
a
link and a date for this? Without mentioning Stampede 2, that statement alone is confusing; we should be careful since our release announcement will highlight this note.
When you log in to Stampede, you get this:
--- Stampede 2 is coming ------ Stampede 1 will start phased decommissioning in January 2017--- --- The KNLs are now available <> Start migrating today --- https://portal.tacc.utexas.edu/user-news/-/news/101909 https://portal.tacc.utexas.edu/user-guides/stampede#knl
Frank
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://keys.gnupg.net.
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On Tue, Nov 22, 2016 at 04:37:19PM -0500, rahul kashyap wrote:
Let me port first and then I'll follow your instructions for the tests.
I've put together an experimental simfactory entry to the new nodes. It runs, and tests pass (except the ones from Dissipation). It should be considered experimental because it does not fully implement proper logging in, and I also used a compute node to compile (because I didn't want to mess with cross-compiling on my first try). This works, but makes using the machine a bit more 'hand-work' than others.
I'll upload those soon to simfactory, marked as 'experimental'.
Frank
Hi Frank,
How was the performance of KNL? I had a chance to run qc0 on a KNL test machine here (single node). It was 4.6x slower than Haswell 2.3GHz (AVX). The tests were done with -O1 option to bypass the issue of Intel 17.
More on the machine: KNL 7230 Centos 7.2 with xppsl-1.4.3 Intel compiler 17.0.1 Quadrant mode Cache/Flat modes tested with AVX512/AVX2/AVX
Vectors thorn is going to be updated for AVX512?
Please take a look at the attached option file.
Hee Il
2016-11-23 6:42 GMT+09:00 Frank Loeffler knarf@cct.lsu.edu:
On Tue, Nov 22, 2016 at 04:37:19PM -0500, rahul kashyap wrote:
Let me port first and then I'll follow your instructions for the tests.
I've put together an experimental simfactory entry to the new nodes. It runs, and tests pass (except the ones from Dissipation). It should be considered experimental because it does not fully implement proper logging in, and I also used a compute node to compile (because I didn't want to mess with cross-compiling on my first try). This works, but makes using the machine a bit more 'hand-work' than others.
I'll upload those soon to simfactory, marked as 'experimental'.
Frank
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On Tue, Nov 29, 2016 at 7:22 PM, Hee Il Kim heeilkim@gmail.com wrote:
Hi Frank,
How was the performance of KNL? I had a chance to run qc0 on a KNL test machine here (single node). It was 4.6x slower than Haswell 2.3GHz (AVX). The tests were done with -O1 option to bypass the issue of Intel 17.
That's approximately what I would expect. I've been fighting with an (unrelated, luckily) segfault until a few days ago.
More on the machine:
KNL 7230 Centos 7.2 with xppsl-1.4.3 Intel compiler 17.0.1 Quadrant mode Cache/Flat modes tested with AVX512/AVX2/AVX
Vectors thorn is going to be updated for AVX512?
Yes, the respective changes are implemented, and will be available after the release freeze is lifted.
-erik
Please take a look at the attached option file.
Hee Il
2016-11-23 6:42 GMT+09:00 Frank Loeffler knarf@cct.lsu.edu:
On Tue, Nov 22, 2016 at 04:37:19PM -0500, rahul kashyap wrote:
Let me port first and then I'll follow your instructions for the tests.
I've put together an experimental simfactory entry to the new nodes. It runs, and tests pass (except the ones from Dissipation). It should be considered experimental because it does not fully implement proper logging in, and I also used a compute node to compile (because I didn't want to mess with cross-compiling on my first try). This works, but makes using the machine a bit more 'hand-work' than others.
I'll upload those soon to simfactory, marked as 'experimental'.
Frank
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
users@lists.einsteintoolkit.org
-
Erik Schnetter -
Frank Loeffler -
Hee Il Kim -
Ian Hinder -
rahul kashyap -
Roland Haas