LI75827: IMPROVING P7 MASS VECTOR LIBRARIES PERFORMANCE

XL C/C++ for Linux Fix Pack 9 (August 2013 Update) for 11.1
April 2011 Update for XL C/C++ for Linux, V11.1
April 2011 Update for XL Fortran for Linux, V13.1
July 2011 Update for XL Fortran for Linux, V13.1
July 2011 Update for XL C/C++ for Linux, V11.1
October 2011 Update for XL C/C++ for Linux, V11.1
October 2011 Update for XL Fortran for Linux, V13.1
June 2012 Update for XL C/C++ for Linux, V11.1
June 2012 Update for XL Fortran for Linux, V13.1
October 2012 Update for XL C/C++ for Linux, V11.1
October 2012 Update for XL Fortran for Linux, V13.1
XL C/C++ for Linux Fix Pack 8 (June 2013 Update) for 11.1
XL Fortran for Linux Fix Pack 8 (June 2013 Update) for 13.1

APAR status

• Closed as program error.

Error description

• P7 MASS vector library single precision functions show reduced
  performance when input/output vectors are not quad-word aligned.
  

Local fix

• In some cases using -lmassvp6 or -lmassvp5 may yield better
  performance. Otherwise users must explicitly quad-word align
  arguments to MASS calls.
  

Problem summary

• The vsdiv function failed to generate the correct result
  whenever the 32bit floating point vectors where not aligned.
  

Problem conclusion

• Fixed vsdiv function problem.
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  XL MASS LIB COM

• Fixed component ID

  5724X1602

Applicable component levels

• R610 PSY

     UP