Read performance from multivalued UV table

[datastage]

I have some jobs that are reading an UniVerse tables with multivalued columns and load to SQL Server after the transformer breaks out multivalues into multiple columns. Displayed by the performance when having the ODBC stage set to 'Insert new or update existing' I changed it to 'Clear the table before insert' to compare times. I got the same load times and then began to think the slow performance was at the UV stage. It appears the tables with more multivalues were the ones that were running longest.

Is it common to expect slower read performance from UV when the table has many multivalues? If there any performance tweaking to improve the read from UV?