应《网络安全法》要求，自2017年10月1日起，未进行实名认证将不得使用互联网跟帖服务。为保障您的帐号能够正常使用，请尽快对帐号进行手机号验证，感谢您的理解与支持！

24小时热门版块排行榜    

小木虫论坛-学术科研互动平台 » 计算模拟区 » 第一性原理 » Vasp&MedeA » 【求助】为什么vasp运行时cpu使用率很不均衡？
51/1返回列表
查看: 2806  |  回复: 9
只看楼主 @他人 存档 新回复提醒 (忽略) 收藏    在APP中查看
当前只显示满足指定条件的回帖，点击这里查看本话题的所有回帖

tubaxu

金虫 (小有名气)

[交流] 【求助】为什么vasp运行时cpu使用率很不均衡？ 已有4人参与

我自己编译通过了vasp，用单节点8个cpu实验时，发现第一个cpu使用率总是超过100%，甚至500%，或者800%，而第六、第七、第八个后面的cpu，利用率只有30%或50%，弄了好多次也不行。但是同样机子上别人装的vasp，却能保持8个cpu都是100%的运算。
   鄙人刚刚入门，水平很菜，请各位达人给予指点，不胜感谢。
回复此楼

» 猜你喜欢

» 本主题相关价值贴推荐，对您同样有帮助:
凡人作一事，便须全副精神注在此一事，首尾不懈。人而无恒，终生一无所成。
1楼 2010-04-21 20:11:01
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

tubaxu

金虫 (小有名气)
引用回帖:
Originally posted by shrek826 at 2010-04-22 21:01:54:


在.bashrc文件中加入一行
export OMP_NUM_THREADS=1
再看看效率如何
如果还是这样，建议你使用其他的mpi来并行

多谢帮忙。我在自己的个人用户下，输入vi ~/.bashrc，在最后加上那一行。结果仍然是一样的，没解决。

我发现可以用“mpiexec”命令实现并行，是不是说明机子里面装的mpich2？？

把我的makefile文件附在后面，如果有心情的朋友，帮忙看看吧。


.SUFFIXES: .inc .f .f90 .F
#-----------------------------------------------------------------------
# Makefile for Intel Fortran compiler for P4 systems
#
# The makefile was tested only under Linux on Intel platforms
# (Suse 5.3, libc 6 <-> glibc 2.X)
#
# it might be required to change some of library pathes, since
# LINUX installation vary a lot
# Hence check ***ALL**** options in this makefile very carefully
#-----------------------------------------------------------------------
#
# Mind that some Linux distributions (Suse 6.1) have a bug in
# libm causing small errors in the error-function (total energy
# is therefore wrong by about 1meV/atom). The recommended
# solution is to update libc.
#
# BLAS must be installed on the machine
# there are several options:
# 1) very slow but works:
#   retrieve the lapackage from ftp.netlib.org
#   and compile the blas routines (BLAS/SRC directory)
#   please use g77 or f77 for the compilation. When I tried to
#   use pgf77 or pgf90 for BLAS, VASP hang up when calling
#   ZHEEV  (however this was with lapack 1.1 now I use lapack 2.0)
# 2) most desirable: get an optimized BLAS
#   for a list of optimized BLAS try
#     http://www.kachinatech.com/~hjjou/scilib/opt_blas.html
#
# the two most reliable packages around are presently:
# 3a) Intels own optimised BLAS (PIII, P4, Itanium)
#     http://developer.intel.com/software/products/mkl/
#   this is really excellent when you use Intel CPU's
#
# 3b) or obtain the atlas based BLAS routines
#     http://math-atlas.sourceforge.net/
#   you certainly need atlas on the Athlon, since the  mkl
#   routines are not optimal on the Athlon.
#   If you want to use atlas based BLAS, check the lines around LIB=
#
# 3c) brand new and mindblowing fast SSE  (4 GFlops on P4, 2.53 GHz)
#   Kazushige Goto's BLAS
#   http://www.cs.utexas.edu/users/kgoto/signup_first.html
#
#
#-----------------------------------------------------------------------

# all CPP processed fortran files have the extension .f
SUFFIX=.f90

#-----------------------------------------------------------------------
# fortran compiler and linker
#-----------------------------------------------------------------------
#FC=ifc
# fortran linker
FCL=$(FC)


#-----------------------------------------------------------------------
# whereis CPP ?? (I need CPP, can't use gcc with proper options)
# that's the location of gcc for SUSE 5.3
#
#  CPP_   =  /usr/lib/gcc-lib/i486-linux/2.7.2/cpp -P -C
#
# that's probably the right line for some Red Hat distribution:
#
#  CPP_   =  /usr/lib/gcc-lib/i386-redhat-linux/2.7.2.3/cpp -P -C
#
#  SUSE X.X, maybe some Red Hat distributions:

CPP_ =  ./preprocess <$*.F | /usr/bin/cpp -P -C -traditional >$*$(SUFFIX)

#-----------------------------------------------------------------------
# possible options for CPP:
# NGXhalf             charge density   reduced in X direction
# wNGXhalf            gamma point only reduced in X direction
# avoidalloc          avoid ALLOCATE if possible
# IFC                 work around some IFC bugs
# CACHE_SIZE          1000 for PII,PIII, 5000 for Athlon, 8000-12000 P4
# RPROMU_DGEMV        use DGEMV instead of DGEMM in RPRO (usually  faster)
# RACCMU_DGEMV        use DGEMV instead of DGEMM in RACC (faster on P4)
#-----------------------------------------------------------------------

#CPP     = $(CPP_)  -DHOST=\"LinuxIFC\" \
#          -Dkind8 -DNGXhalf -DCACHE_SIZE=12000 -DPGF90 -Davoidalloc \
#          -DRPROMU_DGEMV  -DRACCMU_DGEMV

#-----------------------------------------------------------------------
# general fortran flags  (there must a trailing blank on this line)
#-----------------------------------------------------------------------

FFLAGS =  -FR -lowercase

#-----------------------------------------------------------------------
# optimization
# we have tested whether higher optimisation improves performance
# -xW   SSE2 optimization
# -axW  SSE2 optimization,  but also generate code executable on all mach.
# -tpp7 P4 optimization
# -prefetch
#-----------------------------------------------------------------------

OFLAG=-O3 -xW

OFLAG_HIGH = $(OFLAG)
OBJ_HIGH =
OBJ_NOOPT =
DEBUG  = -FR -O0
INLINE = $(OFLAG)


#-----------------------------------------------------------------------
# the following lines specify the position of BLAS  and LAPACK
# on P4, VASP works fastest with Intels mkl performance library
# so that's what I recommend
#-----------------------------------------------------------------------

# Atlas based libraries
#ATLASHOME= $(HOME)/archives/BLAS_OPT/ATLAS/lib/Linux_P4SSE2/
#BLAS=   -L$(ATLASHOME)  -lf77blas -latlas

# use specific libraries (default library path points to other libraries)
#BLAS= $(ATLASHOME)/libf77blas.a $(ATLASHOME)/libatlas.a

# use the mkl Intel libraries for p4 (www.intel.com)
# mkl.5.1
# set -DRPROMU_DGEMV  -DRACCMU_DGEMV in the CPP lines
#BLAS=-L/opt/intel/mkl/lib/32 -lmkl_p4  -lpthread

# mkl.5.2 requires also to -lguide library
# set -DRPROMU_DGEMV  -DRACCMU_DGEMV in the CPP lines
#BLAS=-L/opt/intel/mkl/lib/32 -lmkl_p4 -lguide -lpthread

# even faster Kazushige Goto's BLAS
# http://www.cs.utexas.edu/users/kgoto/signup_first.html
BLAS=  /home/xuyf/GotoBLAS2/libgoto2_nehalemp-r1.07p1.so

# LAPACK, simplest use vasp.4.lib/lapack_double
LAPACK= ../vasp.4.lib/lapack_double.o

# use atlas optimized part of lapack
#LAPACK= ../vasp.4.lib/lapack_atlas.o  -llapack -lcblas

# use the mkl Intel lapack
#LAPACK= -lmkl_lapack

#-----------------------------------------------------------------------

LIB  = -L../vasp.4.lib -ldmy \
     ../vasp.4.lib/linpack_double.o $(LAPACK) \
     $(BLAS)

# options for linking (for compiler version 6.X) nothing is required
#LINK    =
# compiler version 7.0 generates some vector statments which are located
# in the svml library, add the LIBPATH and the library (just in case)
LINK    =  -L/opt/intel/Compiler/11.1/069/lib/intel64/ -lsvml -i-dynamic

#-----------------------------------------------------------------------
# fft libraries:
# VASP.4.6 can use fftw.30 (http://www.fftw.org)
# since this version is faster on P4 machines, we recommend to use it
#-----------------------------------------------------------------------

#FFT3D   = fft3dfurth.o fft3dlib.o
#FFT3D   = fftw3d.o fft3dlib.o   /opt/libs/fftw-3.0/lib/libfftw3.a


#=======================================================================
# MPI section, uncomment the following lines
#
# one comment for users of mpich or lam:
# You must *not* compile mpi with g77/f77, because f77/g77            
# appends *two* underscores to symbols that contain already an        
# underscore (i.e. MPI_SEND becomes mpi_send__).  The pgf90
# compiler however appends only one underscore.
# Precompiled mpi version will also not work !!!
#
# We found that mpich.1.2.1 and lam-6.5.X are stable
# mpich.1.2.1 was configured with
#  ./configure -prefix=/usr/local/mpich_nodvdbg -fc="pgf77 -Mx,119,0x200000"  \
# -f90="pgf90 -Mx,119,0x200000" \
# --without-romio --without-mpe -opt=-O \
#
# lam was configured with the line
#  ./configure  -prefix /usr/local/lam-6.5.X --with-cflags=-O -with-fc=pgf90 \
# --with-f77flags=-O --without-romio
#
# lam was generally faster and we found an average communication
# band with of roughly 160 MBit/s (full duplex)
#
# please note that you might be able to use a lam or mpich version
# compiled with f77/g77, but then you need to add the following
# options: -Msecond_underscore (compilation) and -g77libs (linking)
#
# !!! Please do not send me any queries on how to install MPI, I will
# certainly not answer them !!!!
#=======================================================================
#-----------------------------------------------------------------------
# fortran linker for mpi: if you use LAM and compiled it with the options
# suggested above,  you can use the following line
#-----------------------------------------------------------------------

FC=mpif77
#FCL=$(FC)

#-----------------------------------------------------------------------
# additional options for CPP in parallel version (see also above):
# NGZhalf               charge density   reduced in Z direction
# wNGZhalf              gamma point only reduced in Z direction
# scaLAPACK             use scaLAPACK (usually slower on 100 Mbit Net)
#-----------------------------------------------------------------------

CPP    = $(CPP_) -DMPI  -DHOST=\"LinuxIFC\" -DIFC \
     -Dkind8 -DNGZhalf -DCACHE_SIZE=4000 -DPGF90 -Davoidalloc \
     -DMPI_BLOCK=500 -DPROC_GROUP=8 \
    -DRPROMU_DGEMV  -DRACCMU_DGEMV

#-----------------------------------------------------------------------
# location of SCALAPACK
# if you do not use SCALAPACK simply uncomment the line SCA
#-----------------------------------------------------------------------

BLACS=$(HOME)/archives/SCALAPACK/BLACS/
#SCA_=$(HOME)/archives/SCALAPACK/SCALAPACK

#SCA= $(SCA_)/libscalapack.a  \
$(BLACS)/LIB/blacsF77init_MPI-LINUX-0.a $(BLACS)/LIB/blacs_MPI-LINUX-0.a $(BLACS)/LIB/blacsF77init_MPI-LINUX-0.a

SCA=

#-----------------------------------------------------------------------
# libraries for mpi
#-----------------------------------------------------------------------

#LIB     = -L../vasp.4.lib -ldmy  \
#      ../vasp.4.lib/linpack_double.o $(LAPACK) \
#      $(SCA) $(BLAS)

# FFT: fftmpi.o with fft3dlib of Juergen Furthmueller
FFT3D   = fftmpi.o fftmpi_map.o fft3dlib.o

# fftw.3.0 is slighly faster and should be used if available
#FFT3D   = fftmpiw.o fftmpi_map.o fft3dlib.o   /opt/libs/fftw-3.0/lib/libfftw3.a

#-----------------------------------------------------------------------
# general rules and compile lines
#-----------------------------------------------------------------------
BASIC=   symmetry.o symlib.o   lattlib.o  random.o   

SOURCE=  base.o     mpi.o      smart_allocate.o      xml.o  \
         constant.o jacobi.o   main_mpi.o  scala.o   \
         asa.o      lattice.o  poscar.o   ini.o      setex.o     radial.o  \
         pseudo.o   mgrid.o    mkpoints.o wave.o      wave_mpi.o  $(BASIC) \
         nonl.o     nonlr.o    dfast.o    choleski2.o    \
         mix.o      charge.o   xcgrad.o   xcspin.o    potex1.o   potex2.o  \
         metagga.o  constrmag.o pot.o      cl_shift.o force.o    dos.o      elf.o      \
         tet.o      hamil.o    steep.o    \
         chain.o    dyna.o     relativistic.o LDApU.o sphpro.o  paw.o   us.o \
         ebs.o      wavpre.o   wavpre_noio.o broyden.o \
         dynbr.o    rmm-diis.o reader.o   writer.o   tutor.o xml_writer.o \
         brent.o    stufak.o   fileio.o   opergrid.o stepver.o  \
         dipol.o    xclib.o    chgloc.o   subrot.o   optreal.o   davidson.o \
         edtest.o   electron.o shm.o      pardens.o  paircorrection.o \
         optics.o   constr_cell_relax.o   stm.o    finite_diff.o \
         elpol.o    setlocalpp.o

INC=

vasp: $(SOURCE) $(FFT3D) $(INC) main.o
        rm -f vasp
        $(FCL) -o vasp $(LINK) main.o  $(SOURCE)   $(FFT3D) $(LIB)
makeparam: $(SOURCE) $(FFT3D) makeparam.o main.F $(INC)
        $(FCL) -o makeparam  $(LINK) makeparam.o $(SOURCE) $(FFT3D) $(LIB)
zgemmtest: zgemmtest.o base.o random.o $(INC)
        $(FCL) -o zgemmtest $(LINK) zgemmtest.o random.o base.o $(LIB)
dgemmtest: dgemmtest.o base.o random.o $(INC)
        $(FCL) -o dgemmtest $(LINK) dgemmtest.o random.o base.o $(LIB)
ffttest: base.o smart_allocate.o mpi.o mgrid.o random.o ffttest.o $(FFT3D) $(INC)
        $(FCL) -o ffttest $(LINK) ffttest.o mpi.o mgrid.o random.o smart_allocate.o base.o $(FFT3D) $(LIB)
kpoints: $(SOURCE) $(FFT3D) makekpoints.o main.F $(INC)
        $(FCL) -o kpoints $(LINK) makekpoints.o $(SOURCE) $(FFT3D) $(LIB)

clean:       
        -rm -f *.f *.o *.L ; touch *.F

main.o: main$(SUFFIX)
        $(FC) $(FFLAGS)$(DEBUG)  $(INCS) -c main$(SUFFIX)
xcgrad.o: xcgrad$(SUFFIX)
        $(FC) $(FFLAGS) $(INLINE)  $(INCS) -c xcgrad$(SUFFIX)
xcspin.o: xcspin$(SUFFIX)
        $(FC) $(FFLAGS) $(INLINE)  $(INCS) -c xcspin$(SUFFIX)

makeparam.o: makeparam$(SUFFIX)
        $(FC) $(FFLAGS)$(DEBUG)  $(INCS) -c makeparam$(SUFFIX)

makeparam$(SUFFIX): makeparam.F main.F
#
# MIND: I do not have a full dependency list for the include
# and MODULES: here are only the minimal basic dependencies
# if one strucuture is changed then touch_dep must be called
# with the corresponding name of the structure
#
base.o: base.inc base.F
mgrid.o: mgrid.inc mgrid.F
constant.o: constant.inc constant.F
lattice.o: lattice.inc lattice.F
setex.o: setexm.inc setex.F
pseudo.o: pseudo.inc pseudo.F
poscar.o: poscar.inc poscar.F
mkpoints.o: mkpoints.inc mkpoints.F
wave.o: wave.inc wave.F
nonl.o: nonl.inc nonl.F
nonlr.o: nonlr.inc nonlr.F

$(OBJ_HIGH):
        $(CPP)
        $(FC) $(FFLAGS) $(OFLAG_HIGH) $(INCS) -c $*$(SUFFIX)
$(OBJ_NOOPT):
        $(CPP)
        $(FC) $(FFLAGS) $(INCS) -c $*$(SUFFIX)

fft3dlib_f77.o: fft3dlib_f77.F
        $(CPP)
        $(F77) $(FFLAGS_F77) -c $*$(SUFFIX)

.F.o:
        $(CPP)
        $(FC) $(FFLAGS) $(OFLAG) $(INCS) -c $*$(SUFFIX)
.F$(SUFFIX):
        $(CPP)
$(SUFFIX).o:
        $(FC) $(FFLAGS) $(OFLAG) $(INCS) -c $*$(SUFFIX)

# special rules
#-----------------------------------------------------------------------

# -tpp5|6|7 P, PII-PIII, PIV
# -xW use SIMD (does not pay of on PII, since fft3d uses double prec)
# all other options do no affect the code performance since -O1 is used

fft3dlib.o : fft3dlib.F
        $(CPP)
        $(F77) -FR -lowercase -O1  -xW  -prev_div -unroll0  -vec_report3 -c $*$(SUFFIX)
fft3dfurth.o : fft3dfurth.F
        $(CPP)
        $(F77) -FR -lowercase -O1 -c $*$(SUFFIX)

radial.o : radial.F
        $(CPP)
        $(F77) -FR -lowercase -O1 -c $*$(SUFFIX)

symlib.o : symlib.F
        $(CPP)
        $(F77) -FR -lowercase -O1 -c $*$(SUFFIX)

symmetry.o : symmetry.F
        $(CPP)
        $(F77) -FR -lowercase -O1 -c $*$(SUFFIX)

dynbr.o : dynbr.F
        $(CPP)
        $(F77) -FR -lowercase -O1 -c $*$(SUFFIX)

us.o : us.F
        $(CPP)
        $(F77) -FR -lowercase -O1 -c $*$(SUFFIX)

wave.o : wave.F
        $(CPP)
        $(F77) -FR -lowercase -O0 -c $*$(SUFFIX)

LDApU.o : LDApU.F
        $(CPP)
        $(F77) -FR -lowercase -O2 -c $*$(SUFFIX)
赞一下 回复此楼
凡人作一事，便须全副精神注在此一事，首尾不懈。人而无恒，终生一无所成。
6楼2010-04-23 18:36:38
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖
查看全部 10 个回答

shrek826

金虫 (小有名气)
tubaxu(金币+3): 2010-04-22 19:17
你用的什么mpi编译？如果是mpich-1.2.7，有可能出现这种情况
赞一下 回复此楼
2楼2010-04-21 23:20:14
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

zzy870720z

荣誉版主 (文坛精英)

优秀版主优秀版主优秀版主优秀版主
★
aylayl08(金币+1):谢谢解答 2010-04-22 10:07
tubaxu(金币+3): 2010-04-22 19:17
这个好像有时候会出这种问题，应该是并行的问题，听人说过单节点6核的效率不错，不知道是不是？
赞一下(1人) 回复此楼
博学、审问、慎思、明辨、笃学
3楼2010-04-21 23:43:17
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

tubaxu

金虫 (小有名气)
引用回帖:
Originally posted by shrek826 at 2010-04-21 23:20:14:
你用的什么mpi编译？如果是mpich-1.2.7，有可能出现这种情况

机子里面装的就mpich，没看那个版本（不会看），但不是mpich2。为什么别人编译的可以，我编译的就不行呢？
赞一下 回复此楼
凡人作一事，便须全副精神注在此一事，首尾不懈。人而无恒，终生一无所成。
4楼2010-04-22 00:55:11
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖
查看全部 10 个回答
普通表情 龙 兔 虎 猫 高级回复 (可上传附件)
最具人气热帖推荐 [查看全部] 作者 回/看 最后发表
[硕博家园] 售SCI一区文章，我:8.O.55.1.O.54,科目齐全,可伽急 +3 ASdOkHsho7FD 2026-08-28 5/250 2026-08-29 06:40 by gy1nBQXYQJqL
[基金申请] 投票:  有多少人是今天查系统知道结果的？ +16 爱看书的可乐 2026-08-26 18/900 2026-08-29 06:26 by winsaint
[教师之家] 售SCI-T0P文章，我:8O.5.5.1.O.54,科目齐全,可+急 +3 ASdOkHsho7FD 2026-08-28 5/250 2026-08-29 06:21 by gy1nBQXYQJqL
[基金申请] 为什么资助数各大高校都创新高，自己申请怎么就这么难 +12 Kittylucky 2026-08-27 13/650 2026-08-29 00:04 by superceng
[基金申请] 系统查不到 +11 董八千 2026-08-26 11/550 2026-08-28 18:06 by Leogzhya
[基金申请] 有没有仍没收到信息的 +3 德尚中行 2026-08-27 3/150 2026-08-28 15:24 by rongshuxia
[教师之家] 导师吐槽：我怎么摊上了这么个极品研究生！ +8 苏东坡二世 2026-08-23 8/400 2026-08-28 14:41 by lkforward
[基金申请] 基金不中，共勉 +11 eulota 2026-08-26 11/550 2026-08-28 14:22 by 火星超人xi
[基金申请] 2026年叶企孙基金 +3 bud_bud 2026-08-27 6/300 2026-08-28 11:53 by bud_bud
[基金申请] 哪位高人中了，把查询到的截图贴出来让我看看，让我长长见识 +5 yuleib84 2026-08-26 6/300 2026-08-28 00:02 by yudaoqian88
[基金申请] 面上合作单位盖章 +5 ssyjh 2026-08-27 5/250 2026-08-27 20:50 by gdfollow
[文学芳草园] 梦想 +3 myrtle 2026-08-26 3/150 2026-08-27 10:01 by angelyueyi
[基金申请] 为什么 国际(地区)合作与交流项目 没有放榜？ 10+3 majunge000 2026-08-26 11/550 2026-08-27 08:42 by 北京莱茵编辑
[基金申请] 系统进不去 +4 yanglien 2026-08-26 5/250 2026-08-26 11:10 by wenfengw83
[基金申请] 项目信息和经费信息在系统里都可以看到了 +6 wittyboy 2026-08-26 14/700 2026-08-26 10:55 by wittyboy
[基金申请] 国合可查了 +3 paperzjh 2026-08-26 3/150 2026-08-26 10:41 by LemmonTr
[基金申请] 今天务委会开完了，明天出结果吗 +19 angus9576 2026-08-25 23/1150 2026-08-26 10:03 by zp519
[职场人生] 学生撞见辅导员送外卖，第二天全班都沉默了 +3 鲸鱼融金_浙江_� 2026-08-22 3/150 2026-08-26 09:01 by zzuzxg
[基金申请] 如果此刻你正在为国基感到焦虑，不妨来听听这首《基金之外》 +8 scalable 2026-08-24 8/400 2026-08-25 12:52 by jnhyjjm
[基金申请] 没有任何消息-是不是就凉了 +9 图啦图啦 2026-08-24 10/500 2026-08-25 11:59 by 南海小哥
信息提示
关闭
请填处理意见
关闭