At 11:05 PM 6/01/2006 +0700, you wrote:

>I just tried the simple tests firebird super server between using
>CPUAffinityMask at value 2 (use the 2nd CPU) and at value 3 (use both
>CPUS) on AMD Athlon64 X2 3800+.
>
>I thought at the CPUAffinityMask = 3 the performance would be drop, but it
>didn't happened. The both results is same. Is it just a abnormal
>case? If it's true, why the three test has the same results?

1. Changes to firebird.conf don't take effect until the server is shut
down and restarted.

2. The "see-saw" effect occurs when one processor reaches near
capacity: Windows swings the entire process over to the next
processor...it hits 100% and immediately swings back....etc.

3. How are you actually testing what's going on?

4. I was talking to a CEO in Canberra last year, whose company distributes
a POS system with a Firebird SS back-end as a hardware/software bundle. He
said he had one favourite hardware setup with four CPUs that worked
smoothly with affinity set to all four. He agreed to do a write-up for
us. Unfortunately, it hasn't appeared yet so I can't give any more
information. Perhaps by coincidence you have the same hardware and any
performance problems you perceive are unrelated to SMP.

./heLen