We almost always have some form of split/multivariate testing underway for client's/partners. Even so we have never come close to testing every variation, let alone for a given site.
It doesn't wind up making sense to test every possible variation of price (aka segment), because that means that you wind up with too many 'segments' to compare.
When are split testing you need to make sure that you have enough people in each segment to give you confidence in the validity of the test.
Here is a spreadsheet that enables you to calculate how much traffic to send to a given segment (price/page etc) in order to let you judge the 'winner' based on how confident you want to be that the results of your test aren't just a 'fluke'.
Depending on your conversion ratio (or estimated ratios) it usually turns out that you only want to devote enough traffic/time to test a few prices.
Don't forget to that the traffic you send to the test segments that perform worse than the 'winner' is costing you money.