Hi, all:
I have been a IML user for years.
However, I recently encountered a very weird problem when using IML on the linux system.
Basically I wrote a sas program for a new estimation method created with macros and IML, and then did some simulations on linux system.
The SAS IML was failed with the error message as follows:
ERROR: (execution) Unable to allocate sufficient memory. At least 3115064 more bytes required.
operation : CHOOSE at line 578 column 167
operands : _TEM1001, *LIT1083, dwn
_TEM1001 624 rows 624 cols (numeric)
*LIT1083 1 row 1 col (numeric)
0
dwn 624 rows 624 cols (numeric)
For the above message, the corresponding codes in my program is "dwn=choose(wn=.,0,wn);" where wn is a 624*624 matrix.
Well, the point is that I did the same simulations (same codes on same server, I promise) last Nov, but everything was fine without memory problem.
Here is the information about the sas version and the linux system I used.
Last Nov. :
NOTE: SAS (r) Proprietary Software 9.3 (TS1M0)
NOTE: This session is executing on the Linux 2.6.18-238.9.1.el5 (LIN X64) platform.
Recently:
NOTE: SAS (r) Proprietary Software 9.3 (TS1M0)
NOTE: This session is executing on the Linux 2.6.18-274.7.1.el5 (LIN X64) platform.
Does the 'out of memory' issue really result from the system difference?
Any clue?
I have been struggling with this problem for a while.
I would appreciate anyone who can help out.
Thanks,
Vivian