>>Yes, looks like it. All my efforts of optimizing the code led to decreasing the speed so far. Or it seems so.
>
>It's ulikely that you can speed up some complex code w/o first finding out where bottleneck is.

I looked at the Coverage Profiler results (posted in my reply to Cetin) and nothing sprang at me except for 8K calles of HexToChar (57 GUID calls on 6 records).

07/26/2006 03:52:37 PM Started Load New Trans for Batch # 3138
07/26/2006 03:52:44 PM Finished to fix addresses.
07/26/2006 03:52:47 PM Running LoadNewTrans for 4318 records.
07/26/2006 05:01:29 PM Started to commit changes.
07/26/2006 05:13:03 PM Committed all changes.
07/26/2006 05:23:12 PM Finished with Load New Trans in 5435 sec.

The 10 minutes at the end is ConvertMemoToPDF for 4318 records.