I've done pretty heavy testing on this. The DateAdd-DateDiff method has been the fastest, and has the advantage of being both backwards and forwards compatible.
The Int conversion has rounding issues and simply shouldn't be used.
The float conversion has problems with the new data types in SQL 2008, and thus isn't forwards compatible with that.
The string conversion is a performance killer (takes an average of 4-10 times as long as native date functions).
Conversion to Date is the fastest and is forwards compatible, but isn't backwards compatible.
In situations where backwards compatibility isn't an issue, conversion to Date is best. Where backwards compatibility might matter, the nested date function method is best.
- Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
Property of The Thread
"Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon