The bottom line here is that EP values are currently not what they once were. Back in the olden days we were able to state this is the EP values for T1, T2 etc. Even up to T6 gear we were able to give fairly sensible static values for EP so it was dead easy to compare gear you simply multiplied up the stats and compared two EP totals.
Then came WotLK and haste effects. Over the last year it has become painfully apparent that we cannot give a fixed set of EP values and hence the oft repeated mantra "You have to sim it for yourself". EP values were removed from the TTT and the mantra firmly took its place. Yet still people clinged to the notion that they could simply multiply up the stats to compare two items. Worse still people still cling to the ooohh stat X has > 2 EP I better "stack" that stat. This usually led to wild re-gemming/re-enchanting for the new stat costing hundreds of gold, often with disappointing results.
Now we are gradually coming to the realisation that doing any form of EP value stat comparison is really quite a bad idea given the current volatility of haste etc. The biggest issue is that the amount of effort it takes to work out how best to set up your gear. To setup the sim for multiple runs takes a lot of calculation. Doing this manually is a pain as it takes ages to switch around even a gem or two and re-run the results, Rawr's export helps of course and takes the tedium out of it and reduces the opportunity for errors to slip in. The other option is Rawr's optimiser and this is amazing, the issue is that the underlying calculations are decent but not as sound or as thorough as EnhSim.
So what to do? Personally I've given up entirely on using EP values. What I do is to use Rawr to predict optimisation of current gear and then CHECK the suggestion using Export to EnhSim. The ability of Rawr to hone in on a best setup is wonderful, if only EnhSim could have this feature or I could get Rawr to make a call to EnhSim as a dll so that its optimiaser was using EnhSim to calc the dps.
Perhaps this is a longer term goal? Add an option to EnhSim for a direct API call and provide a EnhSim.dll compiled from the current code. Then in Rawr I could add an option that the optimiser would make an API call with the config it wanted to test. The big issue is that it would take upwards of several hours to run. Rawr's optimiser runs several thousand possible combos of gear to find an optimal set. Running the sim a few thousand times even with a low 1000h config is going to take ages.