<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Jun 18, 2015 at 5:12 PM, Mirco Wahab <span dir="ltr"><<a href="mailto:mirco.wahab@chemie.tu-freiberg.de" target="_blank">mirco.wahab@chemie.tu-freiberg.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Mark,<br>
<span class=""><br>
On 17.06.2015 23:31, Mark Abraham wrote:<br>
> I've solved three correctness issues with the OpenCL implementation now,<br>
> and NPT on AMD looks quite nice. Can you please try the code at<br>
> <a href="https://gerrit.gromacs.org/#/c/4314/26" rel="noreferrer" target="_blank">https://gerrit.gromacs.org/#/c/4314/26</a> and let us know if the situation<br>
> is improved?<br>
<br>
</span>I tried but could not really compile under VS-2013/MSVC-18. The problem<br>
is, as is already stated in a comment, the macro expansion into an<br>
alignment declaration, which is not supported by MSVC:<br></blockquote><div><br></div><div>This is weird. Because it works for me with VS-2013. Do you have the latest Update (2013 Update 4)?</div><div><br></div><div>Roland </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
[basedefinitions.h]<br>
244:<br>
#if defined(_MSC_VER) && (_MSC_VER >= 1700 || defined(__ICL))<br>
# define GMX_ALIGNMENT 1<br>
# define GMX_ALIGNED(type, alignment)<br>
__declspec(align(alignment*sizeof(type))) type<br>
#elif defined(__GNUC__) || defined(__clang__)<br>
# define GMX_ALIGNMENT 1<br>
...<br>
<br>
<br>
I then changed the "(alignment*sizeof(type))" expression<br>
into "32" which would be probably the value it's been<br>
expanded to:<br>
# define GMX_ALIGNED(type, alignment) __declspec(align(32)) type<br>
but tested other values (16, 64).<br>
<br>
I'm not sure this is the show-stopper, but any trial yielded the<br>
same result. mdrun segfaults immediately after initializing<br>
independent of the input file (but working fine in nb=cpu mode).<br>
<br>
I'm attaching the log file here.<br>
<br>
regards,<br>
<br>
M.<br>
<br>
<br>
<br>
--------------------- 8< [crashed md.log w/gpu] ------------------------<br>
Log file opened on Thu Jun 18 23:09:02 2015<br>
Host: DENEB pid: 4884 rank ID: 0 number of ranks: 1<br>
:-) GROMACS - gmx mdrun, VERSION 5.1-beta1-dev (-:<br>
<br>
GROMACS is written by:<br>
Emile Apol Rossen Apostolov Herman J.C. Berendsen Par<br>
Bjelkmar<br>
Aldert van Buuren Rudi van Drunen Anton Feenstra Sebastian<br>
Fritsch<br>
Gerrit Groenhof Christoph Junghans Anca Hamuraru Vincent<br>
Hindriksen<br>
Dimitrios Karkoulis Peter Kasson Carsten Kutzner Per<br>
Larsson<br>
Justin A. Lemkul Magnus Lundborg Pieter Meulenhoff Erik<br>
Marklund<br>
Teemu Murtola Szilard Pall Sander Pronk Roland<br>
Schulz<br>
Alexey Shvetsov Michael Shirts Alfons Sijbers Peter<br>
Tieleman<br>
Teemu Virolainen Christian Wennberg Maarten Wolf<br>
and the project leaders:<br>
Mark Abraham, Berk Hess, Erik Lindahl, and David van der Spoel<br>
<br>
Copyright (c) 1991-2000, University of Groningen, The Netherlands.<br>
Copyright (c) 2001-2015, The GROMACS development team at<br>
Uppsala University, Stockholm University and<br>
the Royal Institute of Technology, Sweden.<br>
check out <a href="http://www.gromacs.org" rel="noreferrer" target="_blank">http://www.gromacs.org</a> for more information.<br>
<br>
GROMACS is free software; you can redistribute it and/or modify it<br>
under the terms of the GNU Lesser General Public License<br>
as published by the Free Software Foundation; either version 2.1<br>
of the License, or (at your option) any later version.<br>
<br>
GROMACS: gmx mdrun, VERSION 5.1-beta1-dev<br>
Executable: D:\GromacsCL\bin\gmx.exe<br>
Data prefix: D:\GromacsCL<br>
Command line:<br>
gmx mdrun -v<br>
<br>
GROMACS version: VERSION 5.1-beta1-dev<br>
Precision: single<br>
Memory model: 64 bit<br>
MPI library: thread_mpi<br>
OpenMP support: enabled (GMX_OPENMP_MAX_THREADS = 32)<br>
GPU support: enabled<br>
OpenCL support: enabled<br>
invsqrt routine: gmx_software_invsqrt(x)<br>
SIMD instructions: SSE2<br>
FFT library: fftw3<br>
RDTSCP usage: enabled<br>
C++11 compilation: disabled<br>
TNG support: enabled<br>
Tracing support: disabled<br>
Built on: Unknown date<br>
Built by: Anonymous@unknown [CMAKE]<br>
Build OS/arch: Windows-6.2 AMD64<br>
Build CPU vendor: AuthenticAMD<br>
Build CPU brand: AMD Phenom(tm) II X6 1090T Processor<br>
Build CPU family: 16 Model: 10 Stepping: 0<br>
Build CPU features: apic clfsh cmov cx8 cx16 htt lahf_lm misalignsse mmx<br>
msr nonstop_tsc pdpe1gb popcnt pse rdtscp sse2 sse3 sse4a<br>
C compiler: C:/Program Files (x86)/Microsoft Visual Studio<br>
12.0/VC/bin/x86_amd64/cl.exe MSVC 18.0.31101.0<br>
C compiler flags: /DWIN32 /D_WINDOWS /W3 /MD /O2 /Ob2 /D NDEBUG<br>
C++ compiler: C:/Program Files (x86)/Microsoft Visual Studio<br>
12.0/VC/bin/x86_amd64/cl.exe MSVC 18.0.31101.0<br>
C++ compiler flags: /DWIN32 /D_WINDOWS /W3 /GR /EHsc /wd4800<br>
/wd4355 /wd4996 /wd4305 /wd4244 /wd4101 /wd4267 /wd4090 /MD /O2 /Ob2 /D<br>
NDEBUG<br>
Boost version: 1.57.0 (external)<br>
OpenCL include dir: C:/Program Files (x86)/AMD APP SDK/3.0-0-Beta/include<br>
OpenCL library: C:/Program Files (x86)/AMD APP<br>
SDK/3.0-0-Beta/lib/x86_64/OpenCL.lib<br>
OpenCL version: 2.0<br>
<br>
<br>
Running on 1 node with total 6 cores, 6 hardware threads, 1 compatible GPU<br>
Hardware detected:<br>
CPU info:<br>
Vendor: AuthenticAMD<br>
Brand: AMD Phenom(tm) II X6 1090T Processor<br>
Family: 16 model: 10 stepping: 0<br>
CPU features: apic clfsh cmov cx8 cx16 htt lahf_lm misalignsse mmx<br>
msr nonstop_tsc pdpe1gb popcnt pse rdtscp sse2 sse3 sse4a<br>
SIMD instructions most likely to fit this hardware: SSE2<br>
SIMD instructions selected at GROMACS compile time: SSE2<br>
GPU info:<br>
Number of GPUs detected: 1<br>
#0: name: Pitcairn, vendor: Advanced Micro Devices, Inc., device<br>
version: OpenCL 1.2 AMD-APP (1642.5), stat: compatible<br>
<br>
<br>
++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++<br>
B. Hess and C. Kutzner and D. van der Spoel and E. Lindahl<br>
GROMACS 4: Algorithms for highly efficient, load-balanced, and scalable<br>
molecular simulation<br>
J. Chem. Theory Comput. 4 (2008) pp. 435-447<br>
-------- -------- --- Thank You --- -------- --------<br>
<br>
<br>
++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++<br>
D. van der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E. Mark and H. J. C.<br>
Berendsen<br>
GROMACS: Fast, Flexible and Free<br>
J. Comp. Chem. 26 (2005) pp. 1701-1719<br>
-------- -------- --- Thank You --- -------- --------<br>
<br>
<br>
++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++<br>
E. Lindahl and B. Hess and D. van der Spoel<br>
GROMACS 3.0: A package for molecular simulation and trajectory analysis<br>
J. Mol. Mod. 7 (2001) pp. 306-317<br>
-------- -------- --- Thank You --- -------- --------<br>
<br>
<br>
++++ PLEASE READ AND CITE THE FOLLOWING REFERENCE ++++<br>
H. J. C. Berendsen, D. van der Spoel and R. van Drunen<br>
GROMACS: A message-passing parallel molecular dynamics implementation<br>
Comp. Phys. Comm. 91 (1995) pp. 43-56<br>
-------- -------- --- Thank You --- -------- --------<br>
<br>
Input Parameters:<br>
integrator = md<br>
tinit = 0<br>
dt = 0.001<br>
nsteps = 100000<br>
init-step = 0<br>
simulation-part = 1<br>
comm-mode = Linear<br>
nstcomm = 100<br>
bd-fric = 0<br>
ld-seed = 665864<br>
emtol = 10<br>
emstep = 0.01<br>
niter = 20<br>
fcstep = 0<br>
nstcgsteep = 1000<br>
nbfgscorr = 10<br>
rtpi = 0.05<br>
nstxout = 0<br>
nstvout = 0<br>
nstfout = 0<br>
nstlog = 100<br>
nstcalcenergy = 100<br>
nstenergy = 1000<br>
nstxout-compressed = 100<br>
compressed-x-precision = 1000<br>
cutoff-scheme = Verlet<br>
nstlist = 25<br>
ns-type = Grid<br>
pbc = xyz<br>
periodic-molecules = FALSE<br>
verlet-buffer-tolerance = 0.005<br>
rlist = 1.097<br>
rlistlong = 1.097<br>
nstcalclr = 25<br>
coulombtype = Reaction-Field<br>
coulomb-modifier = Potential-shift<br>
rcoulomb-switch = 0<br>
rcoulomb = 1<br>
epsilon-r = 1<br>
epsilon-rf = inf<br>
vdw-type = Cut-off<br>
vdw-modifier = Potential-shift<br>
rvdw-switch = 0<br>
rvdw = 1<br>
DispCorr = No<br>
table-extension = 1<br>
fourierspacing = 0.16<br>
fourier-nx = 0<br>
fourier-ny = 0<br>
fourier-nz = 0<br>
pme-order = 4<br>
ewald-rtol = 1e-005<br>
ewald-rtol-lj = 0.001<br>
lj-pme-comb-rule = Geometric<br>
ewald-geometry = 0<br>
epsilon-surface = 0<br>
implicit-solvent = No<br>
gb-algorithm = Still<br>
nstgbradii = 1<br>
rgbradii = 1<br>
gb-epsilon-solvent = 80<br>
gb-saltconc = 0<br>
gb-obc-alpha = 1<br>
gb-obc-beta = 0.8<br>
gb-obc-gamma = 4.85<br>
gb-dielectric-offset = 0.009<br>
sa-algorithm = Ace-approximation<br>
sa-surface-tension = 2.05016<br>
tcoupl = V-rescale<br>
nsttcouple = 25<br>
nh-chain-length = 0<br>
print-nose-hoover-chain-variables = FALSE<br>
pcoupl = No<br>
pcoupltype = Isotropic<br>
nstpcouple = -1<br>
tau-p = 2<br>
compressibility (3x3):<br>
compressibility[ 0]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
compressibility[ 1]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
compressibility[ 2]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
ref-p (3x3):<br>
ref-p[ 0]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
ref-p[ 1]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
ref-p[ 2]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
refcoord-scaling = No<br>
posres-com (3):<br>
posres-com[0]=0.00000e+000<br>
posres-com[1]=0.00000e+000<br>
posres-com[2]=0.00000e+000<br>
posres-comB (3):<br>
posres-comB[0]=0.00000e+000<br>
posres-comB[1]=0.00000e+000<br>
posres-comB[2]=0.00000e+000<br>
QMMM = FALSE<br>
QMconstraints = 0<br>
QMMMscheme = 0<br>
MMChargeScaleFactor = 1<br>
qm-opts:<br>
ngQM = 0<br>
constraint-algorithm = Lincs<br>
continuation = FALSE<br>
Shake-SOR = FALSE<br>
shake-tol = 0.0001<br>
lincs-order = 4<br>
lincs-iter = 1<br>
lincs-warnangle = 30<br>
nwall = 0<br>
wall-type = 9-3<br>
wall-r-linpot = -1<br>
wall-atomtype[0] = -1<br>
wall-atomtype[1] = -1<br>
wall-density[0] = 0<br>
wall-density[1] = 0<br>
wall-ewald-zfac = 3<br>
pull = FALSE<br>
rotation = FALSE<br>
interactiveMD = FALSE<br>
disre = No<br>
disre-weighting = Conservative<br>
disre-mixed = FALSE<br>
dr-fc = 1000<br>
dr-tau = 0<br>
nstdisreout = 100<br>
orire-fc = 0<br>
orire-tau = 0<br>
nstorireout = 100<br>
free-energy = no<br>
cos-acceleration = 0<br>
deform (3x3):<br>
deform[ 0]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
deform[ 1]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
deform[ 2]={0.00000e+000, 0.00000e+000, 0.00000e+000}<br>
simulated-tempering = FALSE<br>
E-x:<br>
n = 0<br>
E-xt:<br>
n = 0<br>
E-y:<br>
n = 0<br>
E-yt:<br>
n = 0<br>
E-z:<br>
n = 0<br>
E-zt:<br>
n = 0<br>
swapcoords = no<br>
adress = FALSE<br>
userint1 = 0<br>
userint2 = 0<br>
userint3 = 0<br>
userint4 = 0<br>
userreal1 = 0<br>
userreal2 = 0<br>
userreal3 = 0<br>
userreal4 = 0<br>
grpopts:<br>
nrdf: 17493<br>
ref-t: 300<br>
tau-t: 0.2<br>
annealing: No<br>
annealing-npoints: 0<br>
acc: 0 0 0<br>
nfreeze: N N N<br>
energygrp-flags[ 0]: 0<br>
<div class="HOEnZb"><div class="h5"><br>
--<br>
Gromacs Developers mailing list<br>
<br>
* Please search the archive at <a href="http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List" rel="noreferrer" target="_blank">http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List</a> before posting!<br>
<br>
* Can't post? Read <a href="http://www.gromacs.org/Support/Mailing_Lists" rel="noreferrer" target="_blank">http://www.gromacs.org/Support/Mailing_Lists</a><br>
<br>
* For (un)subscribe requests visit<br>
<a href="https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers" rel="noreferrer" target="_blank">https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers</a> or send a mail to <a href="mailto:gmx-developers-request@gromacs.org">gmx-developers-request@gromacs.org</a>.<br>
</div></div></blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature">ORNL/UT Center for Molecular Biophysics <a href="http://cmb.ornl.gov" target="_blank">cmb.ornl.gov</a><br>865-241-1537, ORNL PO BOX 2008 MS6309</div>
</div></div>