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

[WIP]Add CUDA support for BSSN RHS with new kernel and update makefiles

Browse Source
This commit is contained in:
ianchb
2026-02-27 21:46:43 +08:00
parent c001939461
commit 724e9cd415
4 changed files with 2524 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
AMSS_NCKU_source/makefile
View File

@@ -37,6 +37,10 @@ endif
.cu.o:
$(Cu) $(CUDA_APP_FLAGS) -c $< -o $@ $(CUDA_LIB_PATH)
# CUDA rewrite of BSSN RHS (drop-in replacement for bssn_rhs_c + stencil helpers)
bssn_rhs_cuda.o: bssn_rhs_cuda.cu macrodef.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 $@
@@ -79,6 +83,9 @@ else
CFILES = bssn_rhs_c.o fderivs_c.o fdderivs_c.o kodiss_c.o lopsided_c.o
endif
# CUDA rewrite: bssn_rhs_cuda.o replaces all CFILES (stencils are built-in)
CFILES_CUDA = bssn_rhs_cuda.o
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\
bssnEScalar_class.o perf.o Z4c_class.o NullShellPatch.o\
@@ -159,9 +166,12 @@ $(CUDAFILES): bssn_gpu.h gpu_mem.h gpu_rhsSS_mem.h
misc.o : zbesh.o
# projects
ABE: $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS)
$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(LDLIBS)
ABE: $(C++FILES) $(CFILES_CUDA) $(F90FILES) $(F77FILES) $(AHFDOBJS)
$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(CFILES_CUDA) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(LDLIBS) -lcudart $(CUDA_LIB_PATH)
ABE_CUDA: $(C++FILES) $(CFILES_CUDA) $(F90FILES) $(F77FILES) $(AHFDOBJS)
$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(CFILES_CUDA) $(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)
@@ -169,4 +179,4 @@ TwoPunctureABE: $(TwoPunctureFILES)
$(CLINKER) $(TP_OPTFLAGS) -qopenmp -o $@ $(TwoPunctureFILES) $(LDLIBS)
clean:
rm *.o ABE ABEGPU TwoPunctureABE make.log -f
rm *.o ABE ABE_CUDA ABEGPU TwoPunctureABE make.log -f