Why does tightening mem speed help with prime95

I was trying to find the max fsb. I was testing 211 fsb and had to tighten my memory timing to 6-3-3-2.5, the max rated speed for my memory at 200mhz, from 11-4-4-3 in order to have Prime95 pass. With the 6-3-3-2.5, P95 was good for at least 55 minutes but with 11-4-4-3 and 8-4-4-3, P95 failed after 2 minutes. This does not make sense since the lax time should have made the oc more stable. But in this case, I needed to tighten the timing in order to have stability. Can anyone explain this? Thanks.

What kind of memory chips are they? I can see something like this happening with BH-5/6. Sometimes memory doesn't like really really lax timings - it messes it up. I think that problem is a little more common with AMD systems. For example, BH-5 WON'T run CAS 3, it's sloppy at CAS 2.5, and happy at CAS 2.

Same with my CH-5; the highest FSB possible was attained at timings 2-8-4-4 but that was no mhz significant over 2-5-3-2

~t0m