Fixes are available
November 2012 Update for XL Fortran for Blue Gene/Q, V14.1
February 2013 Update for XL Fortran for Blue Gene/Q, V14.1
May 2013 Update for XL Fortran for Blue Gene/Q, V14.1
XL Fortran for Blue Gene/Q Fix Pack 5 (August 2013 Update) for 14.1
XL Fortran for Blue Gene/Q Fix Pack 6 (November 2013 Update) for 14.1
XL Fortran for Blue Gene/Q Fix Pack 7 (February 2014 Update) for 14.1
XL Fortran for Blue Gene/Q Fix Pack 8 (May 2014 Update) for 14.1
XL Fortran for Blue Gene/Q Fix Pack 9 (August 2014 Update) for 14.1
Closed as program error.
Performance degradation is seen when using -O3 -qhot related to SIMD copy of data.
Insert a NOSIMD on affected loops.
PROBLEM DESCRIPTION: A performance degradation is observed when the compiler is vectorizing simple copy loops compared with the system memcpy routine. USERS AFFECTED: Users who use -O3 -qhot and above with the XL C/C++ for BG/Q, V12.1 compiler.
On BG/Q, it is observed that memcpy() is significantly faster than generating QPX simd instructions. For that reason, the compiler will now always generate memcpy() transformations instead of vectorizing the loop, as long as there is no memory overlap between the source and target arrays.
Reported component name
XL FORTRAN FOR
Reported component ID
Last modified date
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:
Fixed component name
XL FORTRAN FOR
Fixed component ID
Applicable component levels