https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97063
Bug ID: 97063
Summary: [ MATMUL intrinsic] The value of result is wrong when
vector (step size is negative) * matrix
Product: gcc
Version: 10.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: xin....@compiler-dev.com
Target Milestone: ---
Consider the following Fortran program:
program p
implicit none
real(kind=16), dimension(6) :: arr1
real(kind=16), dimension(6, 10) :: arr2
real(kind=16), dimension(10) :: arr3, arr6
real(kind=16), dimension(3) :: arr4
real(kind=16), dimension(3, 10) :: arr5
integer :: i, j
data arr1 /0.0_16, 1.0_16, 2.0_16, 3.0_16, 4.0_16, 5.0_16/
do i = 1, 6
do j = 1,10
arr2(i, j) = (i + j) * 1.0_16
end do
end do
write(*,*) "--arr3--:"
arr3 = matmul(arr1(5:1:-2),arr2(5:1:-2, :))
write(*,'(5F5.1)') arr3
write(*,*) "--arr4--:"
arr4 = arr1(5:1:-2)
write(*,'(5F5.1)') arr4
write(*,*) "--arr6--:"
arr6 = matmul(arr4, arr2(5:1:-2, :))
write(*,'(5F5.1)') arr6
end program
Compiled with gfortran 10.1.0, the program prints:
--arr3--:
44.0 53.0 62.0 71.0 80.0
89.0 98.0107.0116.0125.0
--arr4--:
4.0 2.0 0.0
--arr6--:
32.0 38.0 44.0 50.0 56.0
62.0 68.0 74.0 80.0 86.0
As you can see,the value of arr3 is differnet with arr6.
