I am working on MIPS reduction on Mainframe.
I have 750 million records to be sorted. The existing sort card in production was like this
EXEC SORT,CORE=MAX,REGION= 0K ---> it was taking 11 to 12 CPU seconds to complete the sort.
I have tested the below sort cards for 4 times as part of Unit testing.
EXEC SORT,REGION=0M,PARM=ELAP -->took around 6 CPU secs
EXEC SORT,REGION=0M,PARM=CPU --> took 11 CPU secs
But when(PARM=ELAP) implemented in production, it is taking 11 CPU secs.
Please let me know if there is any means to reduce the MIPS for the above amount of data.
DFSORT accepts but does not process the following EXEC/DFSPARM PARM
options: ... ELAP, ...
If you do have DFSORT, contact them through your Support and they will be happy to assist with advice to improve performance.
If you don't have DFSORT, how do you think improving the ELAPsed time would reduce the CPU time (consistently)? Again, contact (presumably) SyncSORT for advice which they will be willing to supply to you.