New postgres throughput.

I'm making a benchmark for to measure the throughput of a database
 system. I'm using postgres 7.1.3 and SQL embedded in C .

 The program executes five transacctions with, SELECTs UPDATESs INSERTs,
 DELETEs and FETCHs.
 The workload consists in the intensive execution of this transacctions.

 The problem is that the more operations are executed, the more slower
 truns postgres and the transacctions are slower.

 Can you give me a solution or explication?

 Thanks.