Help !!!!!!

Hello all, Here is the issue. I have a huge insert/update/delete operation and i'm trying to do some benchmark. When i have five client programs running the same stufffs processes are taking 6-8 seconds for handling 100 records which consists of approx 1200 deletes.h 1200 updates and about 200 insertions. So i bumped up the number of processes which does the same job but with different values and i expected up a reduction in the overall processinfg time, But this is what happened. For the first 500 hunderds records it came down dramatically to 1-2 seconds then the onstat -u is showing B--PR-- for almost all the userthreads and the timings bumped upto 11-13 seconds and it maintians the same timing till the end of the process. I has serious buffwaits and a corresponing latch waits showing in the onstat -p outputs. I am totally stuck now.Any body had a similar problem before. I do apprecitate all comments and solutions IFX 7.31.UC2 BUFFERS 150000 LRUS 124CLEANER 20 CHPINTVL 300 I tried all the cases using LRUS and BUFFERS. I found that it is not really the waiting for buffers. Disks are tuned pretty well and also the data are distributed pretty well. I don;t have much lock waits. Also i don't have any buffer overflow or lock overflow Thanks Dhanesh