topic PROC MIXED optimization in Statistical Procedures https://communities.sas.com/t5/Statistical-Procedures/PROC-MIXED-optimization/m-p/57777#M2691 <HTML><HEAD></HEAD><BODY><P>Hello all</P><P></P><P>I want to fit a neste ANOVA model, with random effects, to a large dataset. This dataset has 263376 rows and 7 columns: 1 response and 6 predictors.</P><P></P><P>First, I tried to use PROC MIXED, but it is very slow. Then, I learnt that I could use PROC HPMIXED, which is way faster, but it still can't give me the answers I want in less than 24 hours.</P><P></P><P>Do you have any advice for me? I'm using a 64 bit Red Hat Entreprise machine, with 12 cores and up to 64GB of RAM, but I'm afraid SAS isn't using any parallelization for PROC HPMIXED.</P></BODY></HTML> Thu, 01 Dec 2011 01:45:27 GMT mnunes 2011-12-01T01:45:27Z PROC MIXED optimization https://communities.sas.com/t5/Statistical-Procedures/PROC-MIXED-optimization/m-p/57777#M2691 <HTML><HEAD></HEAD><BODY><P>Hello all</P><P></P><P>I want to fit a neste ANOVA model, with random effects, to a large dataset. This dataset has 263376 rows and 7 columns: 1 response and 6 predictors.</P><P></P><P>First, I tried to use PROC MIXED, but it is very slow. Then, I learnt that I could use PROC HPMIXED, which is way faster, but it still can't give me the answers I want in less than 24 hours.</P><P></P><P>Do you have any advice for me? I'm using a 64 bit Red Hat Entreprise machine, with 12 cores and up to 64GB of RAM, but I'm afraid SAS isn't using any parallelization for PROC HPMIXED.</P></BODY></HTML> Thu, 01 Dec 2011 01:45:27 GMT https://communities.sas.com/t5/Statistical-Procedures/PROC-MIXED-optimization/m-p/57777#M2691 mnunes 2011-12-01T01:45:27Z