<snip> I fear to consume to much resources if I convert the data to decimal each time. </snip>
<snip> I wouldn't be too worried about using a CVD and AP for your total fields. </snip> I would!. I once worked on a COBOL program with only display fields in the code. This program would "peg" a 158. When I changed the Working storage display fields to the appropriate internal representation, the program no longer registered. To make a long story short, *EVERY* subscript and data field that represented a number went through PACK, CVB, (operation), CVD, UNPACK, at least 4 times during the course of execution. BEWARE of CVD/CVB! To answer the original question, Perhaps using Floating point instructions (either Hex or IEEE) will help. However, I have no experience with the CXFR, (operation), CFXR sequence of instructions as far as CPU utilization is concerned. It seems to me a higher level language (SAS, ...) would be of more assistance.... HTH, <snip> > I'm reading some SMF30 records and want to accumulate several 4 byte > binary values (like CPU SU, IO SU, ...) for jobs which achieve the same > criteria. But the summation grows over the 4 byte capacity (I use AR in > Amode31). Is there a way to use 2 register for the summation. Or can I use > Amode64-Code only for the addition. I fear to consume to much resources if I > convert the data to decimal each time. > </snip> ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
