My DP2s, before I moved on to the Merrill version, at 100% was a tad sharper than my E=M5 at 50%, but roughly comparable. That's 4.9 x 3 MP vs. 16 MP.
That doesn't mean I'm about to ditch my E-M5. But having one spatial pixel (x 3 colors) does certainly have a resolution advantage over 4 pixels representing one unit of spatial resoution (2 green, 1 blue, and 1 red) through interpellation. Champions of the bayer sensor against the foveon (I'm not sure why there has to be a war!) argue with the way Sigma counts pixels (46 for the Merril), but the 15 megapixels in a Bayer sensor are not all devoted to resolution either, showing once again that counting MP is a side show. The MFT charts on the DP2 Merrill is better than almost anything out there, across the frame, but again, for photographers numbers become more an interesting way of rationalizing what you see than anything useful in themselves. I would not buy a camera based solely on its specs; I want to see images from it, and preferably be able to download some raw files.
In the end it's the image, as it a has always been, and the DPs can give you quite special results. I'll leave it to the engineers to figure out how to make that happen; I'm a mathematically incompetent student of Anglo-Saxon, English poetry, and music, so I'll never do more than pick up the lingo, only partially understanding it. But I like to think I can make the camera I have work to my advantage. And when the language moves from pixels, signal to noise ratios and all that jazz, to tonal values and scale, color, "drawing", etc. I'm on terra firma.