1:1 is not always best, but it's usually recommended so you can eliminate variables while OCing the CPU/FSB.
You can get better results w/ 450FSB and a 5:6 or 4:5 ratio w/ tight timings and tight Performance Level (tRD).
This article explains why.
http://www.anandtech.com/mb/showdoc.aspx?i=3208&p=9
Interesting article. It is an article for the x48 chipset though. May be different with hte newer p45.
Reading page 8 out of the article...
How do they come up with that equation? I do not see any reference to white papers from intel or jdec or well, anyone thats has any authority on the issue.
Actually I think the equation is wrong on page 8. Otherwise you would need to run FASTER at cas 3 to make it possible. Not slower. If you loosen cas to 4 instead of 3, Its still not possible (1 is not greater then 2) You would actually need to tighten the cas latency to 2.5 or 2 to make trd5 stable at 400mhz and a 1ns command rate.
Of course my math could be completely wrong. but its not
There was an old formula running around somehwere that showed exactly what minimum ram timings could be and it was not all complicated like that. DFI street had it I think...
EDIT: Actually this proves my initial doubt. that X increases with clock speed. The make no mention of whst exactly it is. If anything it should be the other way around (the slower the speed the larger the number) to make their equation work correctly. But in reality. Speed should not change anything except, well speed.
If they are going to come up with this X= whatever than they need to say why.
LOL!!
It may be hard to believe that 450 could be faster than 550, but the key is what tRD is set to. If tRD is too loose you loose all the benefits of the higher speed b/c you have to wait more cycles. It's like having a Ferrari in stop & go traffic.
This is true and we learned it overclocking ATI video cards IIRC. (the x1900 series it became popular to try and crank down the TRD.. )
EDIT EDIT
http://forums.anandtech.com/messageview.aspx?catid=28&threadid=2149827 Now that makes much more sense and adding +2 all the time instead of changing it based on clock speed makes much more sense. (And also does not end up with a result that requires you to run unacheivably tight cas at low trd settings)