Fix CUDA AMR symmetry drift

This commit is contained in:
2026-05-17 23:46:15 +08:00
parent a0b43bae04
commit f2264989d8
5 changed files with 37 additions and 18 deletions

View File

@@ -18,9 +18,9 @@ OMP_FLAG = -qopenmp
ifeq ($(PGO_MODE),instrument)
## Intel Phase 1: instrumentation — omit -ipo/-fp-model fast=2 for faster build and numerical stability
CXXAPPFLAGS = -O3 -xHost -fma -fprofile-instr-generate -ipo \
CXXAPPFLAGS = -O3 -march=znver5 -fma -fprofile-instr-generate -ipo \
-Dfortran3 -Dnewc $(MKL_INC) $(INTERP_LB_FLAGS)
f90appflags = -O3 -xHost -fma -fprofile-instr-generate -ipo \
f90appflags = -O3 -march=znver5 -fma -fprofile-instr-generate -ipo \
-align array64byte -fpp $(MKL_INC) $(POLINT6_FLAG)
else
## opt (default): maximum performance with PGO profile data -fprofile-instr-use=$(PROFDATA) \
@@ -28,23 +28,23 @@ else
## INTERP_LB_FLAGS has been turned off too, now tested and found to be negative optimization
CXXAPPFLAGS = -O3 -xHost -fp-model fast=2 -fma -ipo \
CXXAPPFLAGS = -O3 -march=znver5 -fp-model fast=2 -fma -ipo \
-Dfortran3 -Dnewc $(MKL_INC) $(INTERP_LB_FLAGS)
f90appflags = -O3 -xHost -fp-model fast=2 -fma -ipo \
f90appflags = -O3 -march=znver5 -fp-model fast=2 -fma -ipo \
-align array64byte -fpp $(MKL_INC) $(POLINT6_FLAG)
endif
TP_OPTFLAGS = -O3 -xHost -fp-model fast=2 -fma -ipo \
TP_OPTFLAGS = -O3 -march=znver5 -fp-model fast=2 -fma -ipo \
-Dfortran3 -Dnewc $(MKL_INC)
else
## NVHPC defaults: mpicc/mpicxx/mpifort wrappers
## PGO_MODE is ignored in this branch.
OMP_FLAG = -mp
CXXAPPFLAGS = -O3 -tp=host -Mcache_align -Mfma \
CXXAPPFLAGS = -O3 -march=znver5 -tp=host -Mcache_align -Mfma \
-Dfortran3 -Dnewc $(MKL_INC) $(INTERP_LB_FLAGS)
f90appflags = -O3 -tp=host -Mcache_align -Mfma -Mpreprocess \
f90appflags = -O3 -march=znver5 -tp=host -Mcache_align -Mfma -Mpreprocess \
$(MKL_INC) $(POLINT6_FLAG)
TP_OPTFLAGS = -O3 -tp=host -Mcache_align -Mfma \
TP_OPTFLAGS = -O3 -march=znver5 -tp=host -Mcache_align -Mfma \
-Dfortran3 -Dnewc $(MKL_INC)
endif