Fix to concat0

[EricaCMitchell]

This PR addresses Issue #551

Original code from commit c7e2673 allowed nullptr for buf0 and buf1.

Fixes construction of unique_ptr and ensures buffer size bufsz is a multiple of alignment size_of(void*).

Was able to run mpirun -n 6 moldft input successfully on my machine with a total cpu time of 321.7s.

[evaleev]

@fbischoff please try

[fbischoff]

@evaleev <https://github.com/evaleev> @EricaCMitchell <https://github.com/EricaCMitchell> Unfortunately still not working (see stack trace below). The testsuite doesn’t test on multiple MPI processes, that I will change asap.. timer: guess density 0.13s 0.13s [xeonmax:1655495:0:1655495] Caught signal 11 (Segmentation fault: address not mapped to object at address (nil)) ==== backtrace (tid:1655495) ==== 0 0x000000000003e6f0 __GI___sigaction() :0 1 0x0000000000959ab3 madness::archive::BufferOutputArchive::store<unsigned long>() SCF.cc:0 2 0x0000000000993258 madness::WorldGopInterface::concat0<std::pair<madness::Key<3ul>, double> >() ???:0 3 0x00000000009a66d7 madness::LoadBalanceDeux<3ul>::load_balance() ???:0 4 0x0000000000973571 madness::SCF::initial_guess() ???:0 5 0x00000000008c188b madness::Nemo::value() ???:0 6 0x00000000008c7ca2 madness::Nemo::value() ???:0 7 0x00000000004939e3 main() ???:0 8 0x0000000000029590 __libc_start_call_main() ???:0 9 0x0000000000029640 __libc_start_main_alias_2() :0 10 0x00000000004a2645 _start() ???:0 ================================= -------------------------------------------------------------------------- prterun noticed that process rank 1 with PID 1655495 on node xeonmax exited on signal 11 (Segmentation fault).

…

------------------------------------------------------ Florian Bischoff Humboldt-Universitaet zu Berlin, Institut fuer Chemie Unter den Linden 6, 10099 Berlin phone: +49-30-2093-82708 office: Brook-Taylor-Str 2, room 3'322 ***@***.*** https://tinyurl.com/bischofflab

On Oct 15, 2024, at 9:43 PM, Eduard Valeyev ***@***.***> wrote: @fbischoff <https://github.com/fbischoff> please try — Reply to this email directly, view it on GitHub <#553 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZXPPHG43JWQIDDF5CGM73Z3VV6FAVCNFSM6AAAAABP72MP4KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJUHA3DIMZRGU>. You are receiving this because you were mentioned.

[evaleev]

@fbischoff can an you please share input/metadata?

[fbischoff]

The error occurs both on mac and on the unix cluster, so I guess it’s universal. Here is a minimal example, no input file needed cmake /path/to/patch make -j moldft cd src/apps/moldft mpirun -np 2 moldft --geometry=he

…

------------------------------------------------------ Florian Bischoff Humboldt-Universitaet zu Berlin, Institut fuer Chemie Unter den Linden 6, 10099 Berlin phone: +49-30-2093-82708 office: Brook-Taylor-Str 2, room 3'322 ***@***.*** https://tinyurl.com/bischofflab

On Oct 15, 2024, at 10:32 PM, Eduard Valeyev ***@***.***> wrote: @fbischoff <https://github.com/fbischoff> can an you please share input/metadata? — Reply to this email directly, view it on GitHub <#553 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZXPPBKVTXO4M4S2ST2RZTZ3V3WVAVCNFSM6AAAAABP72MP4KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJUHE3DKMJSGU>. You are receiving this because you were mentioned.

[evaleev]

@fbischoff runs fine for both of us. perhaps build issues on your side? can you please try e740317 again