Three images at distinct 1 ev steps are enough to create an HDR image. The amount of range is not really relevant compared to the response curve of the sensor (which Pentax knows very well). Most HDR software require at least 3 exposures to be able to reconstruct the response curve to know where to "line up" the three histograms and combine them (because "1 ev" apart is never exactly that amount). Again, this is something that in-camera firmware knows very, very well.
Also, it is easier to keep the images lined up to avoid the appearance of "ghosts" or slight misregistration "blur" if you have fewer images. The HDR feature in the K-7does not align the images like off-camera software, and no matter how stable your tripod and head, there will always be some shake when the mirror goes up and down. From what I've seen and read, the range of the images is probably 2 ev apart and computed from the raw files (as it takes 10-12 seconds to compute!). Finally, converting to JPEG first would actually take *more* time as the frames are held in buffer as raw after each shot anyways.
I got a good idea from this article by Jack Howard:
http://www.adorama.com/alc/blogarticle/11608 But I'll know more after I try out the K-7 tomorrow.