Programming the statistical procedures from SAS

ERROR: Integer overflow on computing amount of memory required

Accepted Solution Solved
Reply
Contributor
Posts: 69
Accepted Solution

ERROR: Integer overflow on computing amount of memory required

Hello expert,

 

My data is observations of firms data in stock exchange. I got the following error in LOG:

ERROR: Integer overflow on computing amount of memory required. A request to allocate 28439.9M bytes of memory can not be

honored.

 

My sas code is as follows:

proc mixed data=sasfile06 method=ml covtest nobound;

class Old;

model ROA=INDUSTRY YEAR/s intercept;

random OLD RD MARKETING LOANS/type=un s;

repeated / type=un subject=ID;

run;

 

I probably did a mistake but could not find any clue via interent or forum as a solution.

 

Any remarks will acceptable, THANKS

 

 


Accepted Solutions
Solution
‎06-06-2016 03:44 PM
SAS Super FREQ
Posts: 3,311

Re: ERROR: Integer overflow on computing amount of memory required

I would guess that you are estimating many, many random effects and that SAS cannot allocate enough memory to hold the symmetric matrix for the "Z" design matrix.  You might want to run PROC FREQ to find out how many columns are in your Z matrix:

 

proc freq data=sasfile06;

tables OLD RD MARKETING LOANS;

run;

 

See the doc for PROC MIXED that discusses memory and computing time for hints about how to handle large data and random effects with many levels.

View solution in original post


All Replies
Solution
‎06-06-2016 03:44 PM
SAS Super FREQ
Posts: 3,311

Re: ERROR: Integer overflow on computing amount of memory required

I would guess that you are estimating many, many random effects and that SAS cannot allocate enough memory to hold the symmetric matrix for the "Z" design matrix.  You might want to run PROC FREQ to find out how many columns are in your Z matrix:

 

proc freq data=sasfile06;

tables OLD RD MARKETING LOANS;

run;

 

See the doc for PROC MIXED that discusses memory and computing time for hints about how to handle large data and random effects with many levels.

☑ This topic is SOLVED.

Need further help from the community? Please ask a new question.

Discussion stats
  • 1 reply
  • 466 views
  • 0 likes
  • 2 in conversation