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

feat: Implement hybrid MPI+OpenMP parallelization

Browse Source 
- Enable -qopenmp in makefile.inc
- Add OpenMP directives to 4th order derivatives in diff_new.f90
- Update makefile_and_run.py to dynamic calculate OMP_NUM_THREADS based on 96 cores and remove hardcoded CPU binding
This commit is contained in:
CGH0S7
2026-02-06 13:25:07 +08:00
parent 26c81d8e81
commit 082f9c3423
3 changed files with 67 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
AMSS_NCKU_source/diff_new.f90
View File

@@ -69,6 +69,7 @@
fy = ZEO
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -151,6 +152,7 @@
fx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -218,6 +220,7 @@
fy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -282,6 +285,7 @@
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -371,6 +375,7 @@
fxz = ZEO
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -469,6 +474,7 @@
fxx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -531,6 +537,7 @@
fyy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -594,6 +601,7 @@
fzz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -657,6 +665,7 @@
fxy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -719,6 +728,7 @@
fxz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -780,6 +790,7 @@
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -866,6 +877,7 @@
fxz = ZEO
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -997,6 +1009,7 @@
fy = ZEO
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1151,6 +1164,7 @@
fx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1227,6 +1241,7 @@
fy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1297,6 +1312,7 @@
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1401,6 +1417,7 @@
fxz = ZEO
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1576,6 +1593,7 @@
fxx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1643,6 +1661,7 @@
fyy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1712,6 +1731,7 @@
fzz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1781,6 +1801,7 @@
fxy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1851,6 +1872,7 @@
fxz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -1919,6 +1941,7 @@
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2011,6 +2034,7 @@
fy = ZEO
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2127,6 +2151,7 @@
fx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2212,6 +2237,7 @@
fy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2288,6 +2314,7 @@
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2406,6 +2433,7 @@
fxz = ZEO
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2593,6 +2621,7 @@
fxx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2665,6 +2694,7 @@
fyy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2740,6 +2770,7 @@
fzz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2815,6 +2846,7 @@
fxy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2895,6 +2927,7 @@
fxz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -2973,6 +3006,7 @@
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3080,6 +3114,7 @@
fy = ZEO
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3216,6 +3251,7 @@
fx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3311,6 +3347,7 @@
fy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3395,6 +3432,7 @@
fz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3530,6 +3568,7 @@
fxz = ZEO
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3802,6 +3841,7 @@
fxx = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3883,6 +3923,7 @@
fyy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -3967,6 +4008,7 @@
fzz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -4051,6 +4093,7 @@
fxy = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -4153,6 +4196,7 @@
fxz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1
@@ -4253,6 +4297,7 @@
fyz = ZEO
!$omp parallel do collapse(3) schedule(static)
do k=1,ex(3)-1
do j=1,ex(2)-1
do i=1,ex(1)-1

8
AMSS_NCKU_source/makefile.inc
View File

@@ -15,16 +15,16 @@ LDLIBS = -L${MKLROOT}/lib -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -lifcore
## -xHost: Optimize for the host CPU architecture (Intel/AMD compatible)
## -fp-model fast=2: Aggressive floating-point optimizations
## -fma: Enable fused multiply-add instructions
## Note: OpenMP has been disabled (-qopenmp removed) due to performance issues
CXXAPPFLAGS = -O3 -xHost -fp-model fast=2 -fma \
## Note: OpenMP enabled for hybrid MPI+OpenMP
CXXAPPFLAGS = -O3 -xHost -fp-model fast=2 -fma -qopenmp \
-Dfortran3 -Dnewc -I${MKLROOT}/include
f90appflags = -O3 -xHost -fp-model fast=2 -fma \
f90appflags = -O3 -xHost -fp-model fast=2 -fma -qopenmp \
-fpp -I${MKLROOT}/include
f90 = ifx
f77 = ifx
CXX = icpx
CC = icx
CLINKER = mpiicpx
CLINKER = mpiicpx -qopenmp
Cu = nvcc
CUDA_LIB_PATH = -L/usr/lib/cuda/lib64 -I/usr/include -I/usr/lib/cuda/include