From 85d1e8de8784c98dc5e9a841c69b3d59f88f1d4c Mon Sep 17 00:00:00 2001
From: CGH0S7 <776459475@qq.com>
Date: Sat, 14 Feb 2026 00:43:39 +0800
Subject: [PATCH] Add Intel SIMD vectorization directives to hot-spot functions

Apply Intel Advisor optimization recommendations:
- Add FORCEINLINE to polint for better inlining
- Add SIMD VECTORLENGTHFOR and UNROLL directives to fderivs,
  fdderivs, symmetry_bd, and kodis functions

This improves vectorization efficiency of finite difference
computations.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 AMSS_NCKU_source/diff_new.f90 | 6 +++++-
 AMSS_NCKU_source/fmisc.f90    | 5 +++++
 AMSS_NCKU_source/kodiss.f90   | 2 ++
 3 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/AMSS_NCKU_source/diff_new.f90 b/AMSS_NCKU_source/diff_new.f90
index 93954f1..ad4c2d8 100644
--- a/AMSS_NCKU_source/diff_new.f90
+++ b/AMSS_NCKU_source/diff_new.f90
@@ -69,10 +69,12 @@
   fy = ZEO
   fz = ZEO
 
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
+!DIR$ UNROLL PARTIAL(4)
   do k=1,ex(3)-1
   do j=1,ex(2)-1
   do i=1,ex(1)-1
-! x direction   
+! x direction
         if(i+1 <= imax .and. i-1 >= imin)then
 !
 !              - f(i-1) + f(i+1)
@@ -371,6 +373,8 @@
   fxz = ZEO
   fyz = ZEO
 
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
+!DIR$ UNROLL PARTIAL(4)
   do k=1,ex(3)-1
   do j=1,ex(2)-1
   do i=1,ex(1)-1
diff --git a/AMSS_NCKU_source/fmisc.f90 b/AMSS_NCKU_source/fmisc.f90
index 1b57677..4baf147 100644
--- a/AMSS_NCKU_source/fmisc.f90
+++ b/AMSS_NCKU_source/fmisc.f90
@@ -883,13 +883,17 @@ subroutine symmetry_bd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+1,1:extc(2),1:extc(3))*SoA(1)
    enddo
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
    do i=0,ord-1
       funcc(:,-i,1:extc(3)) = funcc(:,i+1,1:extc(3))*SoA(2)
    enddo
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
    do i=0,ord-1
       funcc(:,:,-i) = funcc(:,:,i+1)*SoA(3)
    enddo
@@ -1112,6 +1116,7 @@ end subroutine d2dump
 ! Lagrangian polynomial interpolation
 !------------------------------------------------------------------------------
 
+!DIR$ ATTRIBUTES FORCEINLINE :: polint
   subroutine polint(xa, ya, x, y, dy, ordn)
   implicit none
 
diff --git a/AMSS_NCKU_source/kodiss.f90 b/AMSS_NCKU_source/kodiss.f90
index a12ada4..a516393 100644
--- a/AMSS_NCKU_source/kodiss.f90
+++ b/AMSS_NCKU_source/kodiss.f90
@@ -65,6 +65,8 @@ real*8,intent(in) :: eps
 !                       dx^4
 
 !  note the sign (-1)^r-1, now r=2
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
+!DIR$ UNROLL PARTIAL(4)
   do k=1,ex(3)
   do j=1,ex(2)
   do i=1,ex(1)