Re: [easybuild] toolchain which uses existing, optimized, MPI/ScaLAPACK and build environment?

I think that is a fair assumption to make. IMO, we shouldn’t expect EB, or any other tool for that matter, to solve all of our problems and just work optimally out of the box in every single case, without operator intervention. The EB toolchains are tested in a limited number of environments.

Re: [easybuild] Lmod integration - Families

I’ll cover it a bit, but not too deeply. In any case, I don’t think Davide will come. Anyway, what are the issues you are seeing? I think that using a toolchain based module naming scheme this would work similarly to the hierarchical scheme (ie: it will prevent you from loading a compiler on

Re: [easybuild] LUA modules: whatis statement split up across multiple lines when printed

nneth Hoste" <easybuild-requ...@lists.ugent.be on behalf of kenneth.ho...@ugent.be> wrote: On 20/01/2017 10:49, Alvarez, Damian wrote: > I like it. I wanted actually to make PR including an extra “contact” field in the easyconfig file (we currently have

Re: [easybuild] LUA modules: whatis statement split up across multiple lines when printed

, "easybuild-requ...@lists.ugent.be on behalf of Alvarez, Damian" <easybuild-requ...@lists.ugent.be on behalf of d.alva...@fz-juelich.de> wrote: Related to this: Maybe it would be useful to add a small --dump-filter option, that simply dumps a list of easyconfigs (or

[easybuild] Re: [Lmod-users] NAMD license validation

Hi Maxime, At JSC we install NAMD in a directory that belongs to the namd group. Access to others is not allowed. To add users to that group they have to send us proof that they acquired a license. I believe there is also an arrangement to communicate with the NAMD developers which users have

Re: [easybuild] Re: [Lmod-users] NAMD license validation

2017 at 18:45 To: "easybuild@lists.ugent.be" <easybuild@lists.ugent.be> Subject: Re: [easybuild] Re: [Lmod-users] NAMD license validation On 20/02/2017 18:40, Alvarez, Damian wrote: Mh, wouldn’t it make sense to use that userInGroup function in EB? I didn’t know it existed, but for

[easybuild] Re: [Lmod-users] NAMD license validation

bruary 2017 at 18:18 To: "Alvarez, Damian" <d.alva...@fz-juelich.de> Cc: "easybuild@lists.ugent.be" <easybuild@lists.ugent.be>, lmod-users <lmod-us...@lists.sourceforge.net> Subject: Re: [Lmod-users] NAMD license validation We do the same thing by making

Re: [easybuild] LUA modules: whatis statement split up across multiple lines when printed

I like it. I wanted actually to make PR including an extra “contact” field in the easyconfig file (we currently have it simply included in the description), but yours is way more complete. However, just as a side note, empty fields should not be created. Having an “usage section” that is empty

Re: [easybuild] Compiling with Intel Xeon Phi (via EasyBuild)

For those that don’t know about it, binfmt_misc can help a LOT in some/many cases when cross-compiling for Xeon Phi. For instance, compiling HDF5 for Xeon Phi without it is a pain (or used to be, I don’t know if HDF5 now explicitly supports it). But with binfmt_misc it is as easy as for

Re: [easybuild] Compiling with Intel Xeon Phi (via EasyBuild)

cess rate with EasyBuild for building for Xeon Phi? What is needed exactly (minimal kernel version, OS packages installed)? Is there a 'standard' script available (or do you have one that you can share)? regards, Kenneth 2016-06-01 9:46 GMT+02:00 Alvarez, Damian <d.alva...@fz-juelich.de<

Re: [easybuild] QuantumEspresso 5.4.0

We have QE 5.4.0 with these buildopts: 'all w90 want xspectra gipaw'. No patches. I am not sure if that helps :-P. From: on behalf of "p.der...@shell.com" Reply-To: "easybuild@lists.ugent.be" Date: Monday 11 July

Re: [easybuild] Any EasyBuild people at SC16?

Hey Ole, I’ll be there. I’ll be presenting the “Scientific Software Management in Real Life: Deployment of EasyBuild on a Large Scale System.” at HUST’16 (one of the SC workshops) on Sunday. I’ll also be around at the Jülich Supercomputing Centre booth at the exhibition and the SC’16

[easybuild] 2nd EasyBuild User Meeting - February 8-10 in Juelich, Germany

Dear EasyBuilders, It is my pleasure to officially announce the 2nd EasyBuild User Meeting! It will take place from the 8th to the 10th of February 2017, in the Juelich Supercomputing Centre, Germany. There is no registration fee, but you still have to register to be allowed to enter the

Re: [easybuild] no module for CUDA-7.5.18

Well, you can simply put CUDA as a dependency in the OpenMPI easyconfig, right? You don’t really need to have a gcccuda toolchain for it, GCC would work just fine. Also, as far as I remember, the CUDA installation doesn’t actively check which version of GCC you are using. That is done at

Re: [easybuild] Find hidden modules by default

Well, alternatively you can add the dependencies to EASYBUILD_HIDE_DEPS and have something like EASYBUILD_HIDE_DEPS="ANTLR,ATK,Autoconf,Automake,Bison,etc”, and those dependencies will be always hidden (so don’t install them directly, or do it with --hidden). If the version number and the

Re: [easybuild] easybuild on a rocks cluster

Hi Yossi, EasyBuild can use the module command that ships with rocks, that’s not an issue. You can tell EasyBuild which modules environment are you using, the “classical” or Lmod. I would recommend you to take a look at Lmod though. EasyBuild changed recently to use Lmod by default (for good

Re: [easybuild] eb file format toolchain version mismatch successful build?

For easyconfig files specified on the command line, we don't check anything, to give people the freedom to name their easyconfig files "test.eb" or whatever (even the .eb extension isn't required). Yay!, freedom to shoot ourselves on the foot ;-P. IMO, if the robot can’t pick it up,

Re: [easybuild] Common config: to hide or not to hide

:57, "easybuild-requ...@lists.ugent.be on behalf of Riccardo Murri" <easybuild-requ...@lists.ugent.be on behalf of riccardo.mu...@uzh.ch> wrote: (Alvarez, Damian, Mon, Apr 03, 2017 at 10:49:46AM +:) > If that helps somebody, this is the list of our hidden dependen

Re: [easybuild] Common config: to hide or not to hide

If that helps somebody, this is the list of our hidden dependencies:

Re: [easybuild] Strategy for hiding modules

You can EasyBuild to regenerate the modules with --module-only. Be careful, as in some corner cases the resulting module might be difficult when simply regenerating the module and when installing software for real. You can also hide the modules using lmod’s 7 hidden modules feature. In your

Re: [easybuild] Bootstrap specific EasyBuild version

Alternatively, for people that regularly contribute back, you could clone the repo[s] somewhere in your home, write a module for it, and use it to install your system wide EB. We use that every time we deploy a new stack. It is close to trivial. This is the TCL module we have for it (and yes,

Re: [easybuild] Does this make sense? (if so, would it make sense to your users?)

I’d argue that binutils are not compilers, and that they are irrelevant for most users. Besides that, looks good to me. We have a similar table on a per-stage basis, but it is currently only exposed to our internal group. Damian On 06/04/17 06:13, "easybuild-requ...@lists.ugent.be on behalf of

Re: [easybuild] Workload managers support

Not really. It depends on the site policy. We launch all the MPI jobs (with ParaStationMPI, IntelMPI and MVAPICH2) with srun. Best, Damian From: on behalf of Kenneth Hoste Reply-To: "easybuild@lists.ugent.be"

Re: [easybuild] [ANN] EasyBuild v3.1.1

Taking about that, my PR needs some attention ;-P https://github.com/hpcugent/easybuild-framework/pull/2134 On 08/03/17 15:12, "easybuild-requ...@lists.ugent.be on behalf of Kenneth Hoste" wrote: So, back up the

Re: [easybuild] Introducing minimal toolchains & optarch changes in existing tree

Just 3 small comments: -If you are going to use just GCC, you can get rid of the leading “GCC:”. -You are specifying -O3, which might conflict with other options (lowopt, noopt or even the default (which is -O2)). Probably you should take that out and tune --default-opt-level instead -You are

[easybuild] Next EasyBuild conf call: Wed July 19th 2017, 5pm CEST

Dear EasyBuilders, The next EasyBuild conf call is planned for tomorrow, Wed July 19th, 5pm CEST; see also https://plus.google.com/u/0/events/cpof57ekded6ulvkt9vbh6i4tb4 Current agenda: * 2017b toolchains * Back up of modules and userInGroup (lmod function) integration * JSC’s new

Re: [easybuild] Next EasyBuild conf call: Wed July 19th 2017, 5pm CEST

Not Kenneth On 18/07/17 19:46, "easybuild-requ...@lists.ugent.be on behalf of Alvarez, Damian" <easybuild-requ...@lists.ugent.be on behalf of d.alva...@fz-juelich.de> wrote: Dear EasyBuilders, The next EasyBuild conf call is planned for tomorrow, Wed July 19th, 5pm

Re: [easybuild] Spider output - whatis vs help

if any) > will be suppressed. > > @Kenneth: Maybe we need to consider the whatis lines as key/value > pairs, allowing to suppress (empty value) or overwrite (non-empty > value) auto-generated whatis entries? > > Markus >

Re: [easybuild] Spider output - whatis vs help

l.com> wrote: On 04-07-17 08:43, Alvarez, Damian wrote: > I think there is no reason to break backwards compatibility. But we still can make it better adding templates like %(pyver)s and allowing whatis to be defined using them. For instance: > > whatis = [

[easybuild] Re: [Lmod-users] Spider output - whatis vs help

page, extensions if any) will be suppressed. @Kenneth: Maybe we need to consider the whatis lines as key/value pairs, allowing to suppress (empty value) or overwrite (non-empty value) auto-generated whatis entries? Markus On 07/03/2017 12:25 PM, Alvarez, Damian wrote:

[easybuild] Re: [Lmod-users] Spider output - whatis vs help

As far as I am concerned, you’ve got it right. Thanks Rob! Damian From: Robert McLay <mc...@tacc.utexas.edu> Date: Monday 3 July 2017 at 16:05 To: "Alvarez, Damian" <d.alva...@fz-juelich.de> Cc: "Geimer, Markus" <m.gei...@fz-juelich.de>, Kenneth Hoste <

Re: [easybuild] mpi4py on OpenMPI 2.x (foss/2017a)

We have installed mpi4py in a toolchain with an MPI that doesn’t support MPI_THREAD_MULTIPLE. So far, nobody reported any error. What is the issue exactly there? On 11/04/17 17:24, "easybuild-requ...@lists.ugent.be on behalf of Henkel, Andreas"

[easybuild] Bug in Intel MPI and potentially other MPI runtimes

Dear EasyBuilders, We have seen a bug in Intel MPI when used in combination with Intel compilers 2017. GCC 5.4.0 can also potentially expose that issue, even though triggering the issue with that compiler is way more complicated (we needed some statistics to assess its existence in a pretty

Re: [easybuild] Software/easyconfigs update avail check

Just to add on that, you could create drop-in replacements for selected packages. You could create MariaDB-10.X-intel-2016b.eb (for instance), update the easyconfig file every time there is a new release in the 10.X series, and reinstall it on top of the old one, so bug fixes and security

Re: [easybuild] upgrade eb failed

Did you check if you have an updated lmod cache? If you try to load the module after making sure that LMOD_CACHED_LOADS is not set, does it work? From: on behalf of Yann Sagon Reply-To: "easybuild@lists.ugent.be"

Re: [easybuild] MariaDB-10.1.13-intel-2016a.eb

It looks to me like a bad idea to set LD_PRELOAD in this case. As I see it, jemalloc can be used in two cases: 1) To transparently enhance runtime performance. In this case it is purely a runtime dependency (not build dependency) and LD_PRELOAD is needed when running the code that depends on

Re: [easybuild] Troubles with running EB-built Qt5

Similar issue: https://software.intel.com/en-us/articles/intel-compiler-version-16-not-compatible-with-recent-libcso6 Damian On 22/11/17 12:43, "easybuild-requ...@lists.ugent.be on behalf of Paul Melis" wrote: Nope,

Re: [easybuild] Building GCCcore that will run on AMD and Intel CPU

I do not know why you are experiencing this, but I don’t think --optarch plays any role here. --optarch=GENERIC will look for the appropriate value for the compiler used, but GCCcore is built on top of dummy, which doesn’t have any optarch defined. So EB shouldn’t add any architecture specific

Re: [easybuild] Redhead/CentOS 7.4 and Intel compiler issues

Hey Ake, Could you elaborate on that? I mean, I do expect the compiler to have bugs, but we haven’t seen anything particularly bad so far, just these 2 things: -The issue with glibc in CentOS 7.4 -An internal compiler error when optimizing certain loops in PETSc with complex numbers. Is there

Re: [easybuild] Redhead/CentOS 7.4 and Intel compiler issues

segfaults) I would classify it as a brown-paperbag-release. I.e. don't ever use it. I never bothered testing double precision. On 11/20/2017 07:23 AM, Alvarez, Damian wrote: > Hey Ake, > > Could you elaborate on that? I mean, I do expect the compi

Re: Should we skip foss/2018a (Re: [easybuild] 2018a toolchains)

+1 for going for 18.1. But the glibc issues (not kernel issues) have been solved already (https://access.redhat.com/errata/RHBA-2017:3296). This package should find its way into CentOS 7.4 real quick (maybe it is already there, I am not sure). Damian On 08/12/17 10:35,

Re: [easybuild] Redhead/CentOS 7.4 and Intel compiler issues

There is a bug fix for it: https://access.redhat.com/errata/RHBA-2017:3296 Damian PS: I didn’t see the title of the subject before.. what was the original poster thinking about….? ;-) From: on behalf of "Strube, Alexandre"

Re: [easybuild] DOLFIN and Fenics

We also have DOLFIN 2017.1 available in our repo: https://github.com/easybuilders/JSC/tree/master/Golden_Repo/2017b/d/DOLFIN Sorry for joining late to the party __ Damian On 02.02.18, 14:37, "easybuild-requ...@lists.ugent.be on behalf of John Donners"

[easybuild] Relaxed hierarchies by relying on MPI ABI compatibility rather than MPI vendor/version.

Dear EasyBuilders and lmod users, I have a question for the community. Currently EasyBuild supports to deploy its software stack in a hierarchical manner, as intended and supported by lmod (ie: load compiler, that expands $MODULEPATH and the MPIs become visible, load an MPI, which expands

Re: [easybuild] UUID_1.0

enneth On 24/09/2018 14:09, Jure Pečar wrote: > On Mon, 24 Sep 2018 11:18:36 + > "Alvarez, Damian" wrote: > >> Alternatively you can just take it from here: https://raw.githubusercontent.com/karelzak/util-linux/master/libuuid/src/l

Re: [easybuild] UUID_1.0

Alternatively you can just take it from here: https://raw.githubusercontent.com/karelzak/util-linux/master/libuuid/src/libuuid.sym I believe that's the file we used. Damian On 24.09.18, 12:50, "easybuild-requ...@lists.ugent.be on behalf of Alvarez, Damian" wrote: Hi Jure,

Re: [easybuild] UUID_1.0

Hi Jure, You need this: preconfigopts = 'LDFLAGS="$LDFLAGS -Wl,--version-script=%(builddir)s/libuuid.sym"' However, libuuid.sym is not part of the software. We have include it as a patch here:

Re: [easybuild] Future of Python at GCCcore level?

Hey Åke, did you check the speed of intel vs gcc in python? I checked it some months ago/last year, and there was no clear winner. Sometimes intel was a bit faster, sometimes gcc was a bit faster. The only occasion where intel was definitely faster was in trigonometrical and exponential

Re: [easybuild] Retrospectively hiding modules / hiding by default

A few notes of caution when rebuilding the whole stack: -Doing --module-only is not guaranteed to work in all cases. Sometimes variables are computed during the build, and might have an effect on the module created. EB will print a diff if the module already exist, pay attention to it.

Re: [easybuild] 2019a update of common toolchains

A word of caution regarding Intel MPI 2019: They changed a lot of things under the hood, and we have seen lots of issues in our systems on relatively large jobs (1.5K+ MPI processes). Basically most collective algorithms don't make it through. We have seen that on 2 different InfiniBand

Re: [easybuild] 2019a update of common toolchains

ered multiple issues, > specifically with libfabric (which is default with Intel MPI 2019). > > -Pramod > > On Mon, Jan 7, 2019 at 4:45 PM Alvarez, Damian <mailto:d.alva...@fz-juelich.de>> wrote: > > A word of caution r

Re: [easybuild] Building foss-2019a fails in binutils-2.31.1.eb (Skylake node)

I have binutils 2.31.1 on Skylake. It has been compiled without avx512 support though. I am not sure if that could play a role. Damian On 20.02.19, 14:46, "easybuild-requ...@lists.ugent.be on behalf of Åke Sandgren" wrote: What is the actual error? Look in

Re: [easybuild] cuda with cuda samples

Dear Yann, I have added this to our CUDA easyconfig: postinstallcmds = [ 'cd %(installdir)s/samples && make SMS="35 37"' ] That should be enough, but remember to tune it for the architectures you need. Damian From: on behalf of Yann Sagon Reply-To: "easybuild@lists.ugent.be" Date:

Re: [easybuild] Openblas(foss) matrix issue

If you rely on the easyconfigs that ship with easybuild, you can get reproducible results by sticking to the same easybuild version. PRs in easyconfigs have been accepted in the past, if they fixed bugs or enhanced functionality, and it should be like that IMO. That said, I don’t use OpenBLAS,

[easybuild] next EasyBuild conf call: Wed May 28th 2019, 5pm CEST

Dear EasyBuilders, Our beloved BDFL won't make it tomorrow. So I am taking over the leading of the next EasyBuild conf call, planned for Wed May 28th 2019, at 5pm CEST. You can join the conf call via https://umu.zoom.us/j/561463243 . Current agenda: * update on next EasyBuild release (3.9.2)

Re: [easybuild] Link to EUM17 slides broken

I fixed it now. Thanks! On 09.05.19, 16:48, "easybuild-requ...@lists.ugent.be on behalf of Loris Bennett" wrote: Hi Kenneth, Kenneth Hoste writes: > Hi Loris, > > On 09/05/2019 15:59, Loris Bennett wrote: >> Hi, >> >> Having found the video, I was looking

[easybuild] SysAdmin position at JSC

Dear everyone, I am sending this email with the approval of our beloved BDFL. We (the HPS division at Juelich Supercomputing Centre) are looking for someone to become part of our team. I thought some of you might know someone that might be interested. If so, please don't hesitate to get in