Sas step run time

[Sirisha1520]

Hi all,

I have a sas code and it reads input file with 4crores records, takes around 66.07 cpu seconds. Is there any way to reduce run time for reading large input files?

Thanks,
Sirisha

[maguiremq]

https://www.youtube.com/watch?v=he9ecikOPCE

[Sajid01]

Well the time depends upon many things. Is it on a local or remote drive, the hardware and software resources available, your code etc, the state of the file (compressed or uncompressed etc). And you are reading 40 million records and there could be many variables in each record.
It is difficult to give any suggestion specific to your case.

[Sirisha1520]

Its a mainframe job using Ps file having few char and date columns around 15 variables with record length 80.

[Tom]

---

@Sirisha1520 wrote:
Its a mainframe job using Ps file having few char and date columns around 15 variables with record length 80.

---

What slice of the machine did your job get?  Maybe it was busy doing something else that had higher priority than you.

[Kurt_Bremser]

Please use units that the rest of the world understands. Wikipedia told me that a "crore" is 10 million (10 mega).

So you are reading 40 million lines from a mainframe partitioned dataset, which means that it took one CPU second to process ~ 150000 rows. Not bad at first glance.

Looking deeper, 80 bytes per row means a raw data throughput of 11.5 MB/s, which is not really fast.

Please run the import step again, with

options fullstimer;

set, and then post the complete log, so we can see any transformations you make, and the real time/CPU time rate.