Date: Fri, 28 Jun 2002 11:27:51 -0700
Reply-To: "William W. Viergever" <wwvierg@ATTGLOBAL.NET>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "William W. Viergever" <wwvierg@ATTGLOBAL.NET>
Subject: OT: For the Sort-Heads: SyncSort for Windows -- some preliminary
#'s
Content-Type: multipart/alternative;
Hello Sort-Heads:
Have been persued by a SyncSort sales rep over the years <g>, and as they
finally have Windows-SAS interface into SyncSort decided to give it a whirl.
Here's my first test, w/ & w/o Syncsort on a W2K Pro (+SP2) PC, w/ a
*single* PIII 800 Mhz, 1GB of ram, and a few U160 10K 36GB hard drives
(tell Puddin' I'm waiting on my new *dual* 2.2 Ghz <g> box till next month
when the new U320 15K drives and 64-bit/66Mhz controllers hit the street!):
W/O SyncSort for Windows:
1924
*******************************************************************************;
1925 *-- 1st: the 00/01 peel-off ;
1926 data &dsn2..&file1 ;
1927 * set &dsn1..&infile1(OBS = 100000 in = CY00 keep = yr dhs_key
b_clmid from adate ddate days_seg code_sel age)
1928 &dsn1..&infile2(OBS = 100000 in = CY00 keep = yr dhs_key
b_clmid from adate ddate days_seg code_sel age)
1929 &dsn1..&infile3(OBS = 100000 in = CY01 keep = yr dhs_key
b_clmid from adate ddate days_seg code_sel age) ;
1930 set &dsn1..&infile1(in = CY00 keep = yr dhs_key b_clmid from
adate ddate days_seg code_sel age)
1931 &dsn1..&infile2(in = CY00 keep = yr dhs_key b_clmid from
adate ddate days_seg code_sel age)
1932 &dsn1..&infile3(in = CY01 keep = yr dhs_key b_clmid from
adate ddate days_seg code_sel age) ;
1933 if CY01 then yr = 2001 ; *-- a temp fix from the mis-coded
CY01 HIP file ;
1934 run ;
NOTE: There were 6,166,920 observations read from the data set SAS1.HIP00GAC.
NOTE: There were 185,309 observations read from the data set SAS1.HIP00PSY.
NOTE: There were 6,689,825 observations read from the data set SAS1.HIP01GAC.
NOTE: The data set SAS2.TRANS_DET01 has 13,042,054 observations and 9
variables.
NOTE: DATA statement used:
real time 1:58.13
cpu time 1:10.31
FWIW: SAS2.TRANS_DET01 is a little over 900MB large at this point .....
1935
*******************************************************************************;
1936 *-- preliminary sort needed for pseudo-interim billing per benny ;
1937 proc sort data = &dsn2..&file1 out = tmp01 ;
1938 by yr dhs_key b_clmid from adate ddate ;
1939 run ;
NOTE: There were 13,042,054 observations read from the data set
SAS2.TRANS_DET01.
NOTE: The data set WORK.TMP01 has 13,042,054 observations and 9 variables.
NOTE: PROCEDURE SORT used:
real time 33:10.70
cpu time 3:33.75
W/ SyncSort for Windows:
1938
*******************************************************************************;
1939 *-- preliminary sort needed for pseudo-interim billing per benny ;
1940 proc sort data = &dsn2..&file1 out = tmp01 ;
1941 by yr dhs_key b_clmid from adate ddate ;
1942 run ;
NOTE: There were 13,042,054 observations read from the data set
SAS2.TRANS_DET01.
NOTE: The data set WORK.TMP01 has 13,042,054 observations and 9 variables.
NOTE: PROCEDURE SORT used:
real time 22:37.24
cpu time 3:48.62
There are some options to have SyncSort provide more detail on what
commands were sent/processed, etc., but for now, I find the CPU
near-equality to be confusing -- any ideas group?
BTW, once you install SyncSort, and DL/install a SAS v8.2 patch file, you
simply code (why does sound familiar? <vbg>):
OPTIONS SORTPGM=HOST
Note: this is a full-blow SyncSort w/ sorts, merges, copies, joins, as well
as filtering & groupings, reformatting of record layouts, and aggregations
(w/ counts, sub-tots & tots, etc.). Haven't played w/ it's GUI interface
yet (wanted to see raw SAS diffs); will provide feedback as I get into it more.
HTH
Out
------------------------------------------------------------
William W. Viergever Voice : (916) 483-8398
Viergever & Associates Fax : (916) 486-1488
Sacramento, CA 95825 E-mail : wwvierg@attglobal.net
------------------------------------------------------------
[text/html]
