Project

General

Profile

Bug #2554

write lowest energy coordinates fail with CG

Added by Carlo Camilloni 4 months ago. Updated 2 months ago.

Status:
Closed
Priority:
Normal
Assignee:
Category:
core library
Target version:
Affected version - extra info:
Affected version:
Difficulty:
uncategorized
Close

Description

Step 405, Epot=-8.872469e+04, Fnorm=7.410e+00, Fmax=7.881e+01 (atom 3770)

writing lowest energy coordinates.

-------------------------------------------------------
Program: gmx mdrun, version 2018.2
Source file: src/gromacs/mdlib/minimize.cpp (line 582)
Function: void write_em_traj(FILE *, t_commrec *, gmx_mdoutf_t, gmx_bool, gmx_bool, const char *, gmx_mtop_t *, t_inputrec *, gmx_int64_t, em_state_t *, t_state *, ObservablesHistory *)

Assertion failed:
Condition: bX
The code below assumes that (with domain decomposition), x is collected to
state_global in the call above.

For more information and tips for troubleshooting, please check the GROMACS
website at http://www.gromacs.org/Documentation/Errors
-------------------------------------------------------
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
with errorcode 1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.

topol.tpr (219 KB) topol.tpr Carlo Camilloni, 06/15/2018 08:43 AM

Related issues

Related to GROMACS - Bug #2578: with PME on GPU, EM setup does not propagate state->x pinning settingsClosed

Associated revisions

Revision 9232fa9a (diff)
Added by Mark Abraham 3 months ago

Fix CG with multiple domains

The global state can only be accessed on master rank, so we should set
the flags there and then permit the DD code to propagate the setup of
the cg_p vector.

Fixes #2554

Change-Id: If294bf000eb0266125ecc6c6bf937aac0390e2b8

Revision 0fe41117 (diff)
Added by Berk Hess 2 months ago

Really fix CG with multiple domains

Commit 9232fa9a incorrectly claimed to fix CG with multiple domains.
The change fixes all issues due to the change of the global state
only being available on the master rank.

Note that there is still an issue with writing the final trajectory
frame.

Refs #2554

Change-Id: I1ed3363f722daad9e6aef00d9ded6647acc820f2

Revision 1fadc372 (diff)
Added by Berk Hess 2 months ago

Fix assertion failure in CG

When the final step coincided with a coordinate output step,
conjugate gradient minimization would exit with an assertion failure
instead of writing confout.gro.

Fixes #2554

Change-Id: I53fffccb5310259132724b83ee452ef01f8ac9c8

History

#1 Updated by Carlo Camilloni 4 months ago

More info: this happened on my Mac with native clang compilers and open-mpi 2.1.3
running it with or without -nb cpu and in scalar

gmx_mpi mdrun -v

GROMACS version: 2018.1
Precision: single
Memory model: 64 bit
MPI library: MPI
OpenMP support: disabled
GPU support: OpenCL
SIMD instructions: AVX2_256
FFT library: fftw-3.3.7-sse2-avx
RDTSCP usage: enabled
TNG support: enabled
Hwloc support: hwloc-1.11.0
Tracing support: disabled
Built on: 2018-03-22 13:43:21
Built by: _biodip.unimi.it [CMAKE]
Build OS/arch: Darwin 17.4.0 x86_64
Build CPU vendor: Intel
Build CPU brand: Intel(R) Core(TM) i7-6700HQ CPU @ 2.60GHz
Build CPU family: 6 Model: 94 Stepping: 3
Build CPU features: aes apic avx avx2 clfsh cmov cx8 cx16 f16c fma hle htt intel lahf mmx msr nonstop_tsc pcid pclmuldq pdcm pdpe1gb popcnt pse rdrnd rdtscp rtm sse2 sse3 sse4.1 sse4.2 ssse3 tdt x2apic
C compiler: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/cc AppleClang 9.0.0.9000039
C compiler flags: -march=core-avx2 -Wno-unknown-pragmas -O3 -DNDEBUG
C++ compiler: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/c++ AppleClang 9.0.0.9000039
C++ compiler flags: -march=core-avx2 -std=c++11 -Wno-unknown-pragmas -O3 -DNDEBUG
OpenCL include dir: /System/Library/Frameworks/OpenCL.framework
OpenCL library: /System/Library/Frameworks/OPENCL.framework
OpenCL version: 1.2

#2 Updated by Carlo Camilloni 3 months ago

Is there any comment about this? Is rather limiting to not be able to use CG minimisation...

#3 Updated by Gerrit Code Review Bot 3 months ago

Gerrit received a related patchset '3' for Issue #2554.
Uploader: Mark Abraham ()
Change-Id: gromacs~release-2018~If294bf000eb0266125ecc6c6bf937aac0390e2b8
Gerrit URL: https://gerrit.gromacs.org/8102

#4 Updated by Mark Abraham 3 months ago

  • Related to Bug #2578: with PME on GPU, EM setup does not propagate state->x pinning settings added

#5 Updated by Mark Abraham 3 months ago

  • Status changed from New to In Progress

Carlo Camilloni wrote:

Is there any comment about this? Is rather limiting to not be able to use CG minimisation...

Sure, it will be straightforward to fix. I suspect that further fixes will be needed for at least PME-on-GPU support.

Sorry about the delays, the relevant people have been travelling, sick, and/or on summer holidays.

#6 Updated by Carlo Camilloni 3 months ago

The current patch does not solve the problem, furthermore I compiled also a scalar version of the code and I still get the same error message
writing lowest energy coordinates.

-------------------------------------------------------
Program: mdrun_d, version 2018.2
Source file: src/gromacs/mdlib/minimize.cpp (line 582)
Function: void write_em_traj(FILE *, t_commrec *, gmx_mdoutf_t, gmx_bool, gmx_bool, const char *, gmx_mtop_t *, t_inputrec *, gmx_int64_t, em_state_t *, t_state *, ObservablesHistory *)

Assertion failed:
Condition: bX
The code below assumes that (with domain decomposition), x is collected to
state_global in the call above.

#7 Updated by Mark Abraham 3 months ago

Yes thanks, we know that further fixes are needed.

#8 Updated by Mark Abraham 3 months ago

  • Status changed from In Progress to Resolved

#9 Updated by Berk Hess 2 months ago

  • Category set to core library
  • Assignee set to Berk Hess
  • Target version set to 2018.3

This issue is not critical for the 2018.3 release, but would be nice to have fixed.
I'll look into it.

#10 Updated by Gerrit Code Review Bot 2 months ago

Gerrit received a related patchset '1' for Issue #2554.
Uploader: Berk Hess ()
Change-Id: gromacs~release-2018~I1ed3363f722daad9e6aef00d9ded6647acc820f2
Gerrit URL: https://gerrit.gromacs.org/8181

#11 Updated by Gerrit Code Review Bot 2 months ago

Gerrit received a related patchset '1' for Issue #2554.
Uploader: Berk Hess ()
Change-Id: gromacs~release-2018~I53fffccb5310259132724b83ee452ef01f8ac9c8
Gerrit URL: https://gerrit.gromacs.org/8182

#12 Updated by Berk Hess 2 months ago

  • Status changed from Resolved to Fix uploaded

#13 Updated by Berk Hess 2 months ago

  • Status changed from Fix uploaded to Resolved

#14 Updated by Paul Bauer 2 months ago

  • Status changed from Resolved to Closed

Also available in: Atom PDF