On Tue, Jun 25, 2013 at 12:11 PM, Djurre de Jong-Bruinink <djurredej...@yahoo.com> wrote: > Dear Gromacs developers/users, > > After suggestions on this mailing list to use intel over gcc compilers, we > recently obtained the newest intel compilers (2013.4.183). I was kind of > disappointed to find that there is no speed-up at all when comparing a gcc > and intel compiled g4.6.2. Is this expected for our hardware, would the > speed-up only become apparent when using specific functions (OMP, > group-scheme) or am I doing something wrong in the compilation? > > An exert fromboth log files is below, the complete logfiles can be found > there: > > http://md.chem.rug.nl/~djurre/logs/N6intel.log > http://md.chem.rug.nl/~djurre/logs/N6gcc.log
You're using a real-MPI process per core, and you have six cores per processor. The recommended procedure is to map cores to OpenMP threads, and choose the number of MPI processes per processor (and thus the number of OpenMP threads per MPI process) to maximize performance. See http://www.gromacs.org/Documentation/Acceleration_and_parallelization#Multi-level_parallelization.3a_MPI.2fthread-MPI_.2b_OpenMP The correct balance will depend on the hardware, simulation, degree of parallelism and compiler. The performance from default settings will generally not be terrible, but getting the maximum performance will certainly require some effort from the user (and in some cases you won't see differences in compilers until you reach this regime). Cheers, Mark > Thanks in advance, > Djurre de Jong > > > Log file opened on Tue Jun 25 11:33:25 2013 > Host: node041 pid: 13590 nodeid: 0 nnodes: 72 > Gromacs version: VERSION 4.6.2 > Precision: single > Memory model: 64 bit > MPI library: MPI > OpenMP support: enabled > GPU support: disabled > invsqrt routine: gmx_software_invsqrt(x) > CPU acceleration: SSE2 > FFT library: fftw-3.3.2-sse2 > Large file support: enabled > RDTSCP usage: enabled > Built on: Mon Jun 24 20:54:51 CEST 2013 > Built by: p238199@login01 [CMAKE] > Build OS/arch: Linux 2.6.18-308.11.1.el5 x86_64 > Build CPU vendor: AuthenticAMD > Build CPU brand: Six-Core AMD Opteron(tm) Processor 2435 > Build CPU family: 16 Model: 8 Stepping: 0 > Build CPU features: apic clfsh cmov cx8 cx16 htt lahf_lm misalignsse mmx msr > nonstop_tsc pdpe1gb popcnt pse rdtscp sse2 sse3 sse4a > C compiler: /cm/shared/apps/openmpi/intel/64/1.4.5/bin/mpicc Intel > icc (ICC) 13.1.2 20130514 > C compiler flags: -msse2 -std=gnu99 -Wall -ip -funroll-all-loops -O3 > -DNDEBUG > > > Core t (s) Wall t (s) (%) > Time: 10651.270 222.008 4797.7 > (ns/day) (hour/ns) > Performance: 77.836 0.308 > > > ###################################################################################### > > Host: node041 pid: 13728 nodeid: 0 nnodes: 72 > Gromacs version: VERSION 4.6.2 > Precision: single > Memory model: 64 bit > MPI library: MPI > OpenMP support: enabled > GPU support: disabled > invsqrt routine: gmx_software_invsqrt(x) > CPU acceleration: SSE2 > FFT library: fftw-3.3.2-sse2 > Large file support: enabled > RDTSCP usage: enabled > Built on: Tue Jun 25 10:25:13 CEST 2013 > Built by: p238199@login01 [CMAKE] > Build OS/arch: Linux 2.6.18-308.11.1.el5 x86_64 > Build CPU vendor: AuthenticAMD > Build CPU brand: Six-Core AMD Opteron(tm) Processor 2435 > Build CPU family: 16 Model: 8 Stepping: 0 > Build CPU features: apic clfsh cmov cx8 cx16 htt lahf_lm misalignsse mmx msr > nonstop_tsc pdpe1gb popcnt pse rdtscp sse2 sse3 sse4a > C compiler: /cm/shared/apps/openmpi/gcc/64/1.4.5/bin/mpicc GNU gcc > (GCC) 4.7.2 > C compiler flags: -msse2 -Wextra -Wno-missing-field-initializers > -Wno-sign-compare -Wall -Wno-unused -Wunused-value -fomit-frame-pointer > -funroll-all-loops -fexcess-precision=fast -O3 -DNDEBUG > > > Core t (s) Wall t (s) (%) > Time: 10516.210 219.223 4797.0 > (ns/day) (hour/ns) > Performance: 78.825 0.304 > -- > gmx-users mailing list gmx-users@gromacs.org > http://lists.gromacs.org/mailman/listinfo/gmx-users > * Please search the archive at > http://www.gromacs.org/Support/Mailing_Lists/Search before posting! > * Please don't post (un)subscribe requests to the list. Use the > www interface or send it to gmx-users-requ...@gromacs.org. > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists -- gmx-users mailing list gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! * Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
