64-BitBrainstorm_2026/AMSS-NCKU
8
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[WIP] Add CUDA support for Z4C

Browse Source 
Rewritten done by Codex.
This still has errors, do not pick this one now.
This commit is contained in:
ianchb
2026-04-27 08:27:24 +08:00
parent 60fee8f1c1
commit c689cc8dc9
6 changed files with 8282 additions and 73 deletions
Download Patch File Download Diff File Expand all files Collapse all files

92
AMSS_NCKU_source/makefile
View File

@@ -49,6 +49,10 @@ endif
bssn_rhs_cuda.o: bssn_rhs_cuda.cu bssn_rhs.h macrodef.h
$(Cu) $(CUDA_APP_FLAGS) -c $< -o $@ $(CUDA_LIB_PATH)
# CUDA rewrite of Z4C Cartesian RHS
z4c_rhs_cuda.o: z4c_rhs_cuda.cu z4c_rhs_cuda.h bssn_rhs.h macrodef.h ricci_gamma.h
$(Cu) $(CUDA_APP_FLAGS) -c $< -o $@ $(CUDA_LIB_PATH)
# C rewrite of BSSN RHS kernel and helpers
bssn_rhs_c.o: bssn_rhs_c.C
${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
@@ -65,14 +69,14 @@ kodiss_c.o: kodiss_c.C
lopsided_c.o: lopsided_c.C
${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
lopsided_kodis_c.o: lopsided_kodis_c.C
${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
z4c_rhs_c.o: z4c_rhs_c.C
${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
#interp_lb_profile.o: interp_lb_profile.C interp_lb_profile.h
# ${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
lopsided_kodis_c.o: lopsided_kodis_c.C
${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
z4c_rhs_c.o: z4c_rhs_c.C
${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
#interp_lb_profile.o: interp_lb_profile.C interp_lb_profile.h
# ${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
## TwoPunctureABE uses fixed optimal flags with its own PGO profile, independent of CXXAPPFLAGS
TP_PROFDATA = /home/$(shell whoami)/AMSS-NCKU/pgo_profile/TwoPunctureABE.profdata
@@ -88,13 +92,16 @@ TwoPunctureABE.o: TwoPunctureABE.C
# Input files
## CUDA BSSN RHS switch
## 1 : use the rewritten CUDA bssn_rhs backend
## 0 : keep the normal CPU/Fortran selection below
USE_CUDA_BSSN ?= 0
## CUDA BSSN RHS switch
## 1 : use the rewritten CUDA bssn_rhs backend
## 0 : keep the normal CPU/Fortran selection below
USE_CUDA_BSSN ?= 0
USE_CUDA_Z4C ?= 0
CXXAPPFLAGS += -DUSE_CUDA_BSSN=$(USE_CUDA_BSSN)
CUDA_APP_FLAGS += -DUSE_CUDA_BSSN=$(USE_CUDA_BSSN)
CXXAPPFLAGS += -DUSE_CUDA_Z4C=$(USE_CUDA_Z4C)
CUDA_APP_FLAGS += -DUSE_CUDA_Z4C=$(USE_CUDA_Z4C)
## Kernel implementation switch (set USE_CXX_KERNELS=0 to fall back to Fortran)
ifeq ($(USE_CXX_KERNELS),0)
@@ -107,22 +114,25 @@ endif
CFILES_CUDA_BSSN = bssn_rhs_cuda.o
ifeq ($(USE_CUDA_BSSN),1)
CFILES = $(CFILES_CUDA_BSSN)
else
CFILES = $(CFILES_CPU)
endif
ifeq ($(USE_CXX_Z4C_KERNELS),1)
CFILES += z4c_rhs_c.o
Z4C_F90_OBJ =
else
Z4C_F90_OBJ = Z4c_rhs.o
endif
## RK4 kernel switch (independent from USE_CXX_KERNELS)
ifeq ($(USE_CXX_RK4),1)
RK4_C_OBJ = rungekutta4_rout_c.o
ifeq ($(USE_CUDA_BSSN),1)
CFILES = $(CFILES_CUDA_BSSN)
else
CFILES = $(CFILES_CPU)
endif
ifeq ($(USE_CUDA_Z4C),1)
CFILES += z4c_rhs_cuda.o
Z4C_F90_OBJ =
else ifeq ($(USE_CXX_Z4C_KERNELS),1)
CFILES += z4c_rhs_c.o
Z4C_F90_OBJ =
else
Z4C_F90_OBJ = Z4c_rhs.o
endif
## RK4 kernel switch (independent from USE_CXX_KERNELS)
ifeq ($(USE_CXX_RK4),1)
RK4_C_OBJ = rungekutta4_rout_c.o
RK4_F90_OBJ =
else
RK4_C_OBJ =
@@ -130,12 +140,15 @@ RK4_F90_OBJ = rungekutta4_rout.o
endif
CFILES += $(RK4_C_OBJ)
ABE_CUDA_CFILES = $(CFILES_CUDA_BSSN) $(RK4_C_OBJ)
ABE_CUDA_CFILES = $(CFILES_CUDA_BSSN) z4c_rhs_cuda.o $(RK4_C_OBJ)
ABE_LDLIBS = $(LDLIBS)
ifeq ($(USE_CUDA_BSSN),1)
ABE_LDLIBS += -lcudart $(CUDA_LIB_PATH)
endif
ifeq ($(USE_CUDA_Z4C),1)
ABE_LDLIBS += -lcudart $(CUDA_LIB_PATH)
endif
C++FILES = ABE.o Ansorg.o Block.o misc.o monitor.o Parallel.o MPatch.o var.o\
cgh.o bssn_class.o surface_integral.o ShellPatch.o\
@@ -155,13 +168,13 @@ C++FILES = ABE.o Ansorg.o Block.o misc.o monitor.o Parallel.o MPatch.o var.o\
F90FILES_BASE = enforce_algebra.o fmisc.o initial_puncture.o prolongrestrict.o\
prolongrestrict_cell.o prolongrestrict_vertex.o\
$(RK4_F90_OBJ) diff_new.o kodiss.o kodiss_sh.o\
lopsidediff.o sommerfeld_rout.o getnp4.o diff_new_sh.o\
shellfunctions.o bssn_rhs_ss.o Set_Rho_ADM.o\
getnp4EScalar.o bssnEScalar_rhs.o bssn_constraint.o ricci_gamma.o\
fadmquantites_bssn.o $(Z4C_F90_OBJ) Z4c_rhs_ss.o point_diff_new_sh.o\
cpbc.o getnp4old.o NullEvol.o initial_null.o initial_maxwell.o\
getnpem2.o empart.o NullNews.o fourdcurvature.o\
bssn2adm.o adm_constraint.o adm_ricci_gamma.o\
lopsidediff.o sommerfeld_rout.o getnp4.o diff_new_sh.o\
shellfunctions.o bssn_rhs_ss.o Set_Rho_ADM.o\
getnp4EScalar.o bssnEScalar_rhs.o bssn_constraint.o ricci_gamma.o\
fadmquantites_bssn.o $(Z4C_F90_OBJ) Z4c_rhs_ss.o point_diff_new_sh.o\
cpbc.o getnp4old.o NullEvol.o initial_null.o initial_maxwell.o\
getnpem2.o empart.o NullNews.o fourdcurvature.o\
bssn2adm.o adm_constraint.o adm_ricci_gamma.o\
scalar_rhs.o initial_scalar.o NullEvol2.o initial_null2.o\
NullNews2.o tool_f.o
@@ -220,9 +233,10 @@ misc.o : zbesh.o
ABE: $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS)
$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(ABE_LDLIBS)
ABE_CUDA: USE_CUDA_BSSN=1
ABE_CUDA: $(C++FILES) $(ABE_CUDA_CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS)
$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(ABE_CUDA_CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(LDLIBS) -lcudart $(CUDA_LIB_PATH)
ABE_CUDA: USE_CUDA_BSSN=1
ABE_CUDA: USE_CUDA_Z4C=1
ABE_CUDA: $(C++FILES) $(ABE_CUDA_CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS)
$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(ABE_CUDA_CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(LDLIBS) -lcudart $(CUDA_LIB_PATH)
#ABEGPU: $(C++FILES_GPU) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(CUDAFILES)
# $(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES_GPU) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(CUDAFILES) $(LDLIBS)