Skip to content

Commit dba1980

Browse files
kaiwu-astrokaiwu-astro
committed
1 parent abd55c4 commit dba1980

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

59 files changed

+14166
-1031
lines changed

‎Makefile

+47
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,47 @@
1+
#
2+
# Makefile for NBODY6++ June 2003 R.Sp., Changed by Long Wang on Aug 20, 2014
3+
#
4+
# Please use:
5+
# make
6+
# make install
7+
#
8+
9+
RESULT = nbody6++.sse.gpu.mpi
10+
INSTALLDIR = /usr/local
11+
EXTRATOOLS = nb6++dumpb2a libinitial.so libnb6out3.so nb6++snapshot
12+
ifeq ("x", "x $(EXTRATOOLS)")
13+
EXTRAOBJS = installtools
14+
endif
15+
16+
VPATH=./build
17+
18+
$(RESULT) :
19+
$(MAKE) -C ./build
20+
21+
clean:
22+
rm -f ./build/*.o ./build/*.so ./build/$(RESULT) $(EXTRATARGET)
23+
24+
install: $(RESULT) $(EXTRAOBJS)
25+
@[ -d $(INSTALLDIR)/bin ] || mkdir $(INSTALLDIR)/bin
26+
@[ -d $(INSTALLDIR)/share ] || mkdir $(INSTALLDIR)/share
27+
@[ -d $(INSTALLDIR)/share/doc ] || mkdir $(INSTALLDIR)/share/doc
28+
cp ./build/$(RESULT) $(INSTALLDIR)/bin
29+
cp ./doc/nbody6++_manual.pdf $(INSTALLDIR)/share/doc/
30+
31+
32+
installtools:
33+
@[ -d $(INSTALLDIR)/bin ] || mkdir $(INSTALLDIR)/bin
34+
@[ -d $(INSTALLDIR)/include ] || mkdir $(INSTALLDIR)/include
35+
@[ -d $(INSTALLDIR)/lib ] || mkdir $(INSTALLDIR)/lib
36+
cp ./build/nb6++dumpb2a $(INSTALLDIR)/bin
37+
cp ./build/nb6++snapshot $(INSTALLDIR)/bin
38+
cp ./include/initial.h $(INSTALLDIR)/include
39+
cp ./build/libinitial.so $(INSTALLDIR)/lib
40+
cp ./include/nb6out3.h $(INSTALLDIR)/include
41+
cp ./build/libnb6out3.so $(INSTALLDIR)/lib
42+
43+
uninstall:
44+
rm -f $(INSTALLDIR)/bin/$(RESULT)
45+
rm -f $(INSTALLDIR)/lib/libinitial.so $(INSTALLDIR)/lib/libnb6out3.so
46+
rm -f $(INSTALLDIR)/share/doc/nbody6++_manual.pdf
47+
rm -f $(INSTALLDIR)/bin/nb6++dumpb2a $(INSTALLDIR)/bin/nb6++snapshot

‎Makefile.in

+47
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,47 @@
1+
#
2+
# Makefile for NBODY6++ June 2003 R.Sp., Changed by Long Wang on Aug 20, 2014
3+
#
4+
# Please use:
5+
# make
6+
# make install
7+
#
8+
9+
RESULT = nbody6++@RESULT@
10+
INSTALLDIR = @prefix@
11+
EXTRATOOLS = nb6++dumpb2a libinitial.so libnb6out3.so nb6++snapshot
12+
ifeq ("x@EXTRARESULT@", "x $(EXTRATOOLS)")
13+
EXTRAOBJS = installtools
14+
endif
15+
16+
VPATH=./build
17+
18+
$(RESULT) @EXTRARESULT@:
19+
$(MAKE) -C ./build
20+
21+
clean:
22+
rm -f ./build/*.o ./build/*.so ./build/$(RESULT) $(EXTRATARGET)
23+
24+
install: $(RESULT) $(EXTRAOBJS)
25+
@[ -d $(INSTALLDIR)/bin ] || mkdir $(INSTALLDIR)/bin
26+
@[ -d $(INSTALLDIR)/share ] || mkdir $(INSTALLDIR)/share
27+
@[ -d $(INSTALLDIR)/share/doc ] || mkdir $(INSTALLDIR)/share/doc
28+
cp ./build/$(RESULT) $(INSTALLDIR)/bin
29+
cp ./doc/nbody6++_manual.pdf $(INSTALLDIR)/share/doc/
30+
31+
32+
installtools: @EXTRARESULT@
33+
@[ -d $(INSTALLDIR)/bin ] || mkdir $(INSTALLDIR)/bin
34+
@[ -d $(INSTALLDIR)/include ] || mkdir $(INSTALLDIR)/include
35+
@[ -d $(INSTALLDIR)/lib ] || mkdir $(INSTALLDIR)/lib
36+
cp ./build/nb6++dumpb2a $(INSTALLDIR)/bin
37+
cp ./build/nb6++snapshot $(INSTALLDIR)/bin
38+
cp ./include/initial.h $(INSTALLDIR)/include
39+
cp ./build/libinitial.so $(INSTALLDIR)/lib
40+
cp ./include/nb6out3.h $(INSTALLDIR)/include
41+
cp ./build/libnb6out3.so $(INSTALLDIR)/lib
42+
43+
uninstall:
44+
rm -f $(INSTALLDIR)/bin/$(RESULT)
45+
rm -f $(INSTALLDIR)/lib/libinitial.so $(INSTALLDIR)/lib/libnb6out3.so
46+
rm -f $(INSTALLDIR)/share/doc/nbody6++_manual.pdf
47+
rm -f $(INSTALLDIR)/bin/nb6++dumpb2a $(INSTALLDIR)/bin/nb6++snapshot

‎README

+70
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,70 @@
1+
2+
June 2021
3+
4+
NBODY6++GPU-Dec2020 - Beijing version Dec2020 - maintained by R. Spurzem and team [email protected]
5+
=====================================================================================================
6+
7+
Note that this version has a common origin with Long Wang's github version https://github.com/nbodyx/ ,
8+
but it differs in quite a number of technical and a few astrophysical issues.
9+
10+
This tarball distribution is preliminary and should be replaced by a proper svn service at the same
11+
location in the future: http://silkroad.bao.ac.cn/repos/Nbody6++GPU-Dec2020/
12+
13+
Recommended usage:
14+
15+
./configure --with-par=b1m --enable-simd=sse --enable-mcmodel=large
16+
make clean ; make -j
17+
18+
It is for up to one million bodies with many initial binaries. The configure script written by
19+
Long Wang has a multitude of further options, check in it or ask.
20+
21+
Sources are in src/Main/ . Due to urgent bugfixes few routines are later then Dec2020.
22+
23+
After make you find the executable and object files in build/ .
24+
25+
This tarball provides two run directories: 16k_sse/ and 100k_sse/ - just test runs for 16k and 100k
26+
single particles, 100 N-body time units. You have to copy the executable file to the run directories.
27+
There is a little script @nb6++.run, which shows an example how to run the code on one (16k) or two
28+
(100k) MPI processes. The environment variable OMP_NUM_THREADS has to be set to the desired value of
29+
OpenMP threads per MPI process. (Maybe your system has it predefined). I also recommend to set
30+
OMP_STACKSIZE=4096M the shell where you run the code.
31+
32+
Documentation: Manual in doc/nbody6++_manual.pdf
33+
34+
This code and the documentation is given without warranty, hopefully it is helpful. All may contain errors.
35+
The code is an offspring of Sverre Aarseth's direct N-body codes see www.sverre.com .
36+
This is the code suitable for parallel and GPU accelerated runs on supercomputers and workstations.
37+
It is inefficient (and even more error prone) for particle numbers below about 50k-100k particles
38+
(depending on hardware). For smaller N you are advised to use Nbody6 and Nbody6GPU for single node/process.
39+
40+
The test runs provided are just creating Plummer models and have initially only single particles. For any
41+
other initial model (special density profiles, initial binaries) it is recommended to provide a dat.10
42+
file in N-body input format (see manual). Such file can be produced by other prograns, like Mccluster.
43+
44+
Seleted References:
45+
https://ui.adsabs.harvard.edu/abs/1999PASP..111.1333A/abstract (Aarseth: NBODY1 to NBODY6)
46+
https://ui.adsabs.harvard.edu/abs/1999JCoAM.109..407S/abstract (Spurzem on NBODY6++)
47+
https://ui.adsabs.harvard.edu/abs/2005MNRAS.363..293H/abstract (Hurley+ on SSE/BSE, earlier references therein)
48+
https://ui.adsabs.harvard.edu/abs/2012MNRAS.424..545N/abstract (Nitadori+: NBODY6GPU)
49+
https://ui.adsabs.harvard.edu/abs/2015MNRAS.450.4070W/abstract (Wang+: NBODY6++GPU)
50+
https://ui.adsabs.harvard.edu/abs/2021arXiv210508067K/abstract (Kamlah+: More on current stellar evol.)
51+
52+
53+
Known Problems:
54+
55+
1. For systems with more than one GPU on one node the association of MPI rank id and GPU bus id is not
56+
well defined, will be improved in next version.
57+
2. Runs with a million or more bodies and huge numbers of binaries (5% or more) use extreme amounts of
58+
computing time for the KS binaries (much much more than should be expected). We work on this.
59+
3. On some systems heap and stack management when using OpenMP and MPI together seem to produce very
60+
strange errors and segmentation faults. The exact reason is not known; we work on this.
61+
62+
=======================================================================================================
63+
64+
65+
66+
67+
68+
69+
70+

‎build/Makefile

+166
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,166 @@
1+
#
2+
# Makefile for NBODY6++ June 2003 R.Sp., Changed by Long Wang on Aug 20, 2014
3+
#
4+
# Please use:
5+
# make
6+
# make install
7+
#
8+
# Extra Tools
9+
# dump_btoa (A small routine for transfering between unformatted
10+
# fort.1/2 and formatted dump.dat. dump.dat is very
11+
# useful when you want to restart the simulation in a
12+
# different system)
13+
14+
FC = mpif77
15+
CXX = g++
16+
CC = gcc
17+
NVCC = nvcc
18+
19+
RESULT = nbody6++.sse.gpu.mpi
20+
#FFLAGS = -O3 -fbounds-check -fbacktrace -fno-automatic -fmax-stack-var-size=0 -fPIC -mcmodel=large -Wall
21+
OMP_FLAGS= -D OMP
22+
MPI_FLAGS = -D PARALLEL -D PUREMPI
23+
GPU_FLAGS = -D GPU
24+
SIMD_FLAGS = -D SIMD
25+
HDF5_FLAGS = -D H5OUTPUT
26+
DEBUG_FLAGS= -D DEBUG
27+
TT_FLAGS= -D TT
28+
INTEL_FLAGS= -D __USE_INTEL -no-wrap-margin
29+
30+
FFLAGS = -g -fbounds-check -fbacktrace -O3 -fPIC -mcmodel=large -fopenmp -I../include $(GPU_FLAGS) $(MPI_FLAGS) ${SIMD_FLAGS} ${OMP_FLAGS}
31+
CXXFLAGS = -O3 -fopenmp -I../include -march=native -fPIC -mcmodel=large -msse ${OMP_FLAGS}
32+
CUFLAGS = -O3 -I ../extra_inc/cuda
33+
LDFLAGS = -o $(RESULT)
34+
35+
MPI_FSOURCES= energy_mpi.f fpoly1_mpi.f fpoly2_mpi.f
36+
GPU_FSOURCES= phicor.f
37+
HDF5_FSOURCES=
38+
TT_FSOURCES= ttforce.f ttgalaxy.f ttinit.f ttcal.f jacobi_transform.f
39+
40+
IRRAVX_OBJECTS=irr.avx.o
41+
IRRSSE_OBJECTS=irr.sse.o
42+
AVX_OBJECTS=reg.avx.o pot.avx.o irr.avx.o
43+
SSE_OBJECTS=reg.sse.o pot.sse.o irr.sse.o
44+
CUDA_OBJECTS = gpunb.velocity.o gpupot.gpu.o
45+
CUDA_OBJECTS_SINGLE = gpunb.gpu.o gpupot.gpu.o
46+
#CUDAOBJECTS = gpunb.gpu.o gpupot.gpu.o gpupot.mdot.o
47+
#CUDAOBJECTS = gpunb.velocity.o gpupot.gpu.o gpupot.mdot.o
48+
#CUDAOBJECTS = gpunb.gpu.o gpupot.gpu.o
49+
EXTRATOOLS=nb6++dumpb2a libinitial.so libnb6out3.so nb6++snapshot
50+
51+
EXTRASRC= $(GPU_FSOURCES) $(MPI_FSOURCES)
52+
EXTRAOBJ= $(CUDA_OBJECTS) ${IRRSSE_OBJECTS}
53+
54+
#VISITFLAGS = -DVISIT
55+
#VISITLD = -L ./lvisit -llvisit_nbody -L/work/Tuc/spurzem/visit/visit/lvisit/lib -llvisit -L/work/Tuc/spurzem/visit/visit/visit20/lib -lvisit
56+
57+
#VPATH= ../src/Chain:../src/Nchain:../src/Init:../src/Main:../src/Cloud:../src/GPU:../src/SE:../src/SIMD:../src/Tidal:../src/Tlist:../src/Tools:../src/Ellan:../src/KS:../src/Output:../src/HIR:../include
58+
VPATH= ../src/Main:../src/Tools:../include
59+
60+
INC = params.h common6.h timing.h commonc.h common2.h kspars.h
61+
62+
SOURCE = ${EXTRASRC} nbody6.f file_init.f ellan.f eigenvalue.f indexx.f \
63+
adjust.f assess.f bindat.f binev.f binout.f binpop.f block.f bodies.f \
64+
brake.f brake2.f brake3.f bsetid.f chaos0.f chaos.f \
65+
check.f checkl.f chrect.f clint.f cloud.f cloud0.f \
66+
cmbody.f cmcorr.f cmfirr.f cmfreg.f coal.f comenv.f core.f corerd.f \
67+
cputim.f data.f decide.f deform.f degen.f delay.f \
68+
dgcore.f dtchck.f eccmod.f ecirc.f edot.f efac2.f efac3.f \
69+
expel.f escape.f events.f expand.f fclose.f \
70+
fcloud.f fcorr.f fdisk.f fhalo.f ficorr.f findj.f findm.f \
71+
flyby.f fnuc.f fpcorr.f fpert.f fpoly1.f fpoly2.f \
72+
gcinit.f gcint.f giant.f giant3.f gntage.f grrad.f hcorr.f \
73+
hiarch.f hicirc.f hidat.f higrow.f himax.f himod.f \
74+
hipop.f hirect.f histab.f hivel.f hmdot.f hmdot2.f hotsys.f \
75+
hrdiag.f hrplot.f hut.f hut2.f iblock.f imf.f imfbd.f imf2.f \
76+
impact.f induce.f input.f insert.f instar.f intgrt.f \
77+
jacobi.f kick.f kick2.f ksapo.f kscorr.f \
78+
ksin2.f ksinit.f ksint.f kslist.f ksmod.f ksperi.f kspert.f \
79+
kspoly.f kspred.f ksrect.f ksreg.f ksres.f ksres2.f ksterm.f \
80+
kstide.f lagr.f lagr2.f levels.f magbrk.f matrix.f mdot.f merge.f \
81+
merge2.f mix.f mloss.f mlwind.f modify.f mrenv.f mydump.f \
82+
nbint.f nblist.f nbpot.f nbrem.f nbrest.f \
83+
newtev.f nstab.f offset.f orbit.f output.f peri.f permit.f \
84+
pfac.f poti.f proto_star.f qtides.f ran2.f regint.f \
85+
remove.f rename_ks.f reset.f reset2.f resolv.f rkint.f rl.f roche.f \
86+
rpmax.f rpmax2.f rpmin.f scale.f search.f setup.f short.f shrink.f \
87+
sort1.f spiral.f stability.f star.f start.f stepk.f steps.f stumpf.f \
88+
subint.f synch.f tcirc.f tides.f tides2.f \
89+
tides3.f touch.f tpert.f trdot.f trdot2.f trflow.f tstab.f tstep.f \
90+
units.f unpert.f update.f verify.f xtrnl0.f xtrnld.f xtrnlf.f xtrnlp.f \
91+
xtrnlv.f xvpred.f zare.f zcnsts.f zero.f zfuncs.f \
92+
triple.f derqp3.f difsy3.f erel3.f extend.f qpmod3.f stabl3.f \
93+
stablz.f start3.f subsys.f tperi.f trans3.f \
94+
quad.f derqp4.f difsy4.f endreg.f erel4.f ichain.f newreg.f newsys.f \
95+
qpmod4.f rchain.f rsort.f stabl4.f start4.f status.f trans4.f \
96+
cfuncs.f chain.f chstab.f const.f cstab2.f cstab3.f cstab4.f cstab5.f \
97+
derqp.f difsy1.f erel.f hpsort.f inclin.f invert.f ksphys.f physks.f \
98+
qforce.f qpmod.f r2sort.f recoil.f redraw.f select.f slow.f stablc.f \
99+
swcond.f switch.f transk.f transq.f transx.f vector.f xtf.f \
100+
ycopy.f ysave.f \
101+
absorb.f chaos2.f chdata.f chfind.f chfirr.f chinit.f chlist.f chmod_chain.f \
102+
chpot.f chterm.f expel2.f fchain.f ghost.f giant2.f kcpert.f \
103+
reduce.f reinit.f renew.f setsys.f tchain.f xcpred.f xtpert.f \
104+
xbpredall.f jpred.f jpred_int.f fpoly1_ks.f fpoly2_ks.f cmfirr_cor.f cmfirr_ucor.f \
105+
kcpert_cor.f ksparmpi.f remove_ks.f nbint_cor.f string_left.f \
106+
sort_tlist.f next_tlist.f shrink_tlist.f repair_tlist.f remove_tlist.f \
107+
add_tlist.f exchange_tlist.f replace_tlist.f delay_store_tlist.f \
108+
delay_add_tlist.f delay_remove_tlist.f rmesc_tlist.f shift_tlist.f k_step.f \
109+
imbhinit.f bhplot.f energy.f kspinit.f kspreg.f counter_reset.f util_gpu.f \
110+
regcor_gpu.f fpoly0.f tail0.f ntint.f xtrnlt.f fbulge.f steps2.f tstep2.f \
111+
custom_output.f custom_output_facility.f ksres_op.f kstran.f \
112+
global_output.f global_params_gether.f
113+
114+
.SUFFIXES : .o .F
115+
116+
OBJECTS = $(SOURCE:.f=.o)
117+
118+
TARGET: $(RESULT)
119+
120+
$(RESULT): $(OBJECTS) $(EXTRAOBJ)
121+
$(FC) $(FFLAGS) $(LDFLAGS) $(OBJECTS) $(EXTRAOBJ) -lstdc++ -lcudart
122+
123+
libinitial.so: initial.h initial.cpp
124+
$(CXX) -shared $(CXXFLAGS) ../src/Tools/initial.cpp -o libinitial.so
125+
126+
libnb6out3.so: nb6out3.h nb6out3.cxx
127+
$(CXX) -shared $(CXXFLAGS) ../src/Tools/nb6out3.cxx -o libnb6out3.so
128+
129+
nb6++dumpb2a: dump_btoa.F
130+
$(FC) $(FFLAGS) $^ -o nb6++dumpb2a
131+
132+
nb6++snapshot: libinitial.so libnb6out3.so snapshot.cpp
133+
$(CXX) $(CXXFLAGS) ../src/Tools/snapshot.cpp -L./ -lnb6out3 -linitial -o nb6++snapshot
134+
135+
$(OBJECTS): $(INC)
136+
137+
irr.avx.o: irr.avx.cpp
138+
$(CXX) $(CXXFLAGS) $^ -c -o $@
139+
140+
reg.avx.o: reg.avx.cpp
141+
$(CXX) $(CXXFLAGS) $^ -c -o $@
142+
143+
pot.avx.o: pot.avx.cpp
144+
$(CXX) $(CXXFLAGS) $^ -c -o $@
145+
146+
irr.sse.o: irr.sse.cpp
147+
$(CXX) $(CXXFLAGS) $^ -c -o $@
148+
149+
reg.sse.o: reg.sse.cpp
150+
$(CXX) $(CXXFLAGS) $^ -c -o $@
151+
152+
pot.sse.o: pot.sse.cpp
153+
$(CXX) $(CXXFLAGS) $^ -c -o $@
154+
155+
gpunb.velocity.o: gpunb.velocity.cu
156+
$(NVCC) -c $(CUFLAGS) -Xcompiler "$(CXXFLAGS)" $^
157+
158+
gpunb.gpu.o: gpunb.gpu.cu
159+
$(NVCC) -c $(CUFLAGS) -Xcompiler "$(CXXFLAGS)" $^
160+
161+
gpupot.gpu.o: gpupot.gpu.cu
162+
$(NVCC) -c $(CUFLAGS) -Xcompiler "$(CXXFLAGS)" $^
163+
164+
gpupot.mdot.o: gpupot.mdot.cu
165+
$(NVCC) -c $(CUFLAGS) -Xcompiler "$(CXXFLAGS)" $^
166+

0 commit comments

Comments
 (0)