36
H. Murai et al.
^
&&
||
max
10 min
firstmax
firstmin
lastmax
lastmin
Fortran
+
*
-
.and.
5 .or.
.eqv.
.neqv.
max
min
10 iand
ior
ieor
firstmax
firstmin
15 lastmax
lastmin
Note The total result is calculated by combining the partial results on all nodes. The
ordering of the combination is unspecified. Hence, if the target variable is a type of
floating point (e.g. float in XMP/C or real in XMP/Fortran), the difference of
the order can make a little bit difference in the result value from that in the original
serial execution.
3.2.2 Parallelizing Nested Loop
Parallelization of nested loops can be specified similarly to a single one, as follows.
XcalableMP C
#pragma xmp nodes p[2][2]
#pragma xmp template t[10][10]
#pragma xmp distribute t[block][block] onto p
H. Murai et al.
^
&&
||
max
10 min
firstmax
firstmin
lastmax
lastmin
Fortran
+
*
-
.and.
5 .or.
.eqv.
.neqv.
max
min
10 iand
ior
ieor
firstmax
firstmin
15 lastmax
lastmin
Note The total result is calculated by combining the partial results on all nodes. The
ordering of the combination is unspecified. Hence, if the target variable is a type of
floating point (e.g. float in XMP/C or real in XMP/Fortran), the difference of
the order can make a little bit difference in the result value from that in the original
serial execution.
3.2.2 Parallelizing Nested Loop
Parallelization of nested loops can be specified similarly to a single one, as follows.
XcalableMP C
#pragma xmp nodes p[2][2]
#pragma xmp template t[10][10]
#pragma xmp distribute t[block][block] onto p