Add safe BSSN-EScalar kernel and transfer toggles

This commit is contained in:
2026-04-25 01:41:55 +08:00
parent 0f1d0de1e7
commit 0cf58176d9
6 changed files with 379 additions and 413 deletions

View File

@@ -2,11 +2,19 @@
include makefile.inc
ifeq ($(USE_CXX_ESCALAR_KERNEL),1)
ifeq ($(USE_CXX_KERNELS),0)
$(error USE_CXX_ESCALAR_KERNEL=1 requires USE_CXX_KERNELS=1 because bssn_escalar_rhs_c.C reuses the C BSSN kernel)
endif
endif
## polint(ordn=6) kernel selector:
## 1 (default): barycentric fast path
## 0 : fallback to Neville path
POLINT6_USE_BARY ?= 1
POLINT6_FLAG = -DPOLINT6_USE_BARYCENTRIC=$(POLINT6_USE_BARY)
TRANSFER_CACHE_FLAG = -DBSSN_USE_TRANSFER_CACHE=$(USE_TRANSFER_CACHE)
ESCALAR_KERNEL_FLAG = -DBSSN_USE_ESCALAR_C_KERNEL=$(USE_CXX_ESCALAR_KERNEL)
## ABE build flags selected by PGO_MODE (set in makefile.inc, default: opt)
## make -> opt (PGO-guided, maximum performance)
@@ -16,7 +24,8 @@ PROFDATA = /home/$(shell whoami)/AMSS-NCKU/pgo_profile/default.profdata
ifeq ($(PGO_MODE),instrument)
## Phase 1: instrumentation — omit -ipo/-fp-model fast=2 for faster build and numerical stability
CXXAPPFLAGS = -O3 -xHost -fma -fprofile-instr-generate -ipo \
-Dfortran3 -Dnewc -I${MKLROOT}/include $(INTERP_LB_FLAGS)
-Dfortran3 -Dnewc -I${MKLROOT}/include $(INTERP_LB_FLAGS) \
$(TRANSFER_CACHE_FLAG) $(ESCALAR_KERNEL_FLAG)
f90appflags = -O3 -xHost -fma -fprofile-instr-generate -ipo \
-align array64byte -fpp -I${MKLROOT}/include $(POLINT6_FLAG)
else
@@ -26,7 +35,8 @@ else
CXXAPPFLAGS = -O3 -xHost -fp-model fast=2 -fma -ipo \
-Dfortran3 -Dnewc -I${MKLROOT}/include $(INTERP_LB_FLAGS)
-Dfortran3 -Dnewc -I${MKLROOT}/include $(INTERP_LB_FLAGS) \
$(TRANSFER_CACHE_FLAG) $(ESCALAR_KERNEL_FLAG)
f90appflags = -O3 -xHost -fp-model fast=2 -fma -ipo \
-align array64byte -fpp -I${MKLROOT}/include $(POLINT6_FLAG)
endif
@@ -87,7 +97,10 @@ ifeq ($(USE_CXX_KERNELS),0)
CFILES =
else
# C++ mode (default): C rewrite of bssn/bssn-escalar rhs and helper kernels
CFILES = bssn_rhs_c.o bssn_escalar_rhs_c.o fderivs_c.o fdderivs_c.o kodiss_c.o lopsided_c.o lopsided_kodis_c.o
CFILES = bssn_rhs_c.o fderivs_c.o fdderivs_c.o kodiss_c.o lopsided_c.o lopsided_kodis_c.o
ifeq ($(USE_CXX_ESCALAR_KERNEL),1)
CFILES += bssn_escalar_rhs_c.o
endif
endif
## RK4 kernel switch (independent from USE_CXX_KERNELS)