LCPQ/qmcchem
10
1
Fork 1
mirror of https://gitlab.com/scemama/qmcchem.git synced 2025-01-02 17:45:39 +01:00
Code Releases Activity

Merged outer loop in sparse_full_mv

Browse Source
This commit is contained in:
Anthony Scemama 2016-06-03 11:53:34 +02:00
parent 837c5bbbf7
commit 0e243a8afc
1 changed files with 13 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
src/mo.irp.f
View File

@ -710,16 +710,6 @@ subroutine sparse_full_mv(A,LDA, &
d32 = B2(kao+2) d32 = B2(kao+2)
d42 = B2(kao+3) d42 = B2(kao+3)
do k=0,LDA-1,$IRP_ALIGN/4
!DIR$ VECTOR ALIGNED
do j=1,$IRP_ALIGN/4
C1(j+k) = C1(j+k) + A(j+k,k_vec(1))*d11 + A(j+k,k_vec(2))*d21&
+ A(j+k,k_vec(3))*d31 + A(j+k,k_vec(4))*d41
C2(j+k) = C2(j+k) + A(j+k,k_vec(1))*d12 + A(j+k,k_vec(2))*d22&
+ A(j+k,k_vec(3))*d32 + A(j+k,k_vec(4))*d42
enddo
enddo
d13 = B3(kao ) d13 = B3(kao )
d23 = B3(kao+1) d23 = B3(kao+1)
d33 = B3(kao+2) d33 = B3(kao+2)
@ -730,7 +720,20 @@ subroutine sparse_full_mv(A,LDA, &
d34 = B4(kao+2) d34 = B4(kao+2)
d44 = B4(kao+3) d44 = B4(kao+3)
d15 = B5(kao )
d25 = B5(kao+1)
d35 = B5(kao+2)
d45 = B5(kao+3)
do k=0,LDA-1,$IRP_ALIGN/4 do k=0,LDA-1,$IRP_ALIGN/4
!DIR$ VECTOR ALIGNED
do j=1,$IRP_ALIGN/4
C1(j+k) = C1(j+k) + A(j+k,k_vec(1))*d11 + A(j+k,k_vec(2))*d21&
+ A(j+k,k_vec(3))*d31 + A(j+k,k_vec(4))*d41
C2(j+k) = C2(j+k) + A(j+k,k_vec(1))*d12 + A(j+k,k_vec(2))*d22&
+ A(j+k,k_vec(3))*d32 + A(j+k,k_vec(4))*d42
enddo
!DIR$ VECTOR ALIGNED !DIR$ VECTOR ALIGNED
do j=1,$IRP_ALIGN/4 do j=1,$IRP_ALIGN/4
C3(j+k) = C3(j+k) + A(j+k,k_vec(1))*d13 + A(j+k,k_vec(2))*d23& C3(j+k) = C3(j+k) + A(j+k,k_vec(1))*d13 + A(j+k,k_vec(2))*d23&
@ -738,14 +741,7 @@ subroutine sparse_full_mv(A,LDA, &
C4(j+k) = C4(j+k) + A(j+k,k_vec(1))*d14 + A(j+k,k_vec(2))*d24& C4(j+k) = C4(j+k) + A(j+k,k_vec(1))*d14 + A(j+k,k_vec(2))*d24&
+ A(j+k,k_vec(3))*d34 + A(j+k,k_vec(4))*d44 + A(j+k,k_vec(3))*d34 + A(j+k,k_vec(4))*d44
enddo enddo
enddo
d15 = B5(kao )
d25 = B5(kao+1)
d35 = B5(kao+2)
d45 = B5(kao+3)
do k=0,LDA-1,$IRP_ALIGN/4
!DIR$ VECTOR ALIGNED !DIR$ VECTOR ALIGNED
do j=1,$IRP_ALIGN/4 do j=1,$IRP_ALIGN/4
C5(j+k) = C5(j+k) + A(j+k,k_vec(1))*d15 + A(j+k,k_vec(2))*d25& C5(j+k) = C5(j+k) + A(j+k,k_vec(1))*d15 + A(j+k,k_vec(2))*d25&