This design you will notice can waste some stone that is caught in the eddy currents below the water blocks. This is a small and generally acceptable loss, however, if you don't want to lose those blocks, you can solve the problem by digging out the middle-lower stone block on each side and placing a block of glass one out from there, creating a new current that carries those blocks to within reach of the squares adjacent to the glass block. Just step to the side and pick them up within 5 minutes to keep them from despawning, see below:
By efficiency I presume you mean mined stone to return rate?
Mine has a very thoroughly proven 95-115 return on the 130 uses of a stone pick. And other than mousewheeling to a new pick, set your mouse in a jig, weight the button, watch a movie. I'm not even kidding.
No one should ever run out of cobblestone ever again.
All but the front can be built into another wall, this is a schematic for the fewest-materials design that is also freestanding, modify as you will. It is 5x7, and can be reduced to
Let stone represent whatever substance you use for construction.
Let the lava brick represent the places I recommend you place lava (though apparently it will work without some of them, skimp as you will)
Let cobblestone represent your product
Let water represent placed water source blocks.
Let the stick represent a ladder, recommended ingeniously by the boyf to halt water from 'wafting' you forward and backward inevitably resulting in a burned face (Absolutely requires the second ladder, I know you only have to place them every other step for climbing capability, but that isn't the purpose, just as a water/lava block. Bumping into lava does not happen anymore, but to clarify, continue to beware the 'southwest' rule regarding lava and corners as you will share a corner with lava)
This does not require or even abstractly need any obsidian. The design can be shortened by reducing the lava/product 'layers' to as short as you wish; in this case only do I recommend an obsidian as a guard against overdrilling as it would likely greatly reduce the efficiency of the device.
mostly I meant space and resource efficiency (to an extent, time efficiency, but only in that the two-block design is a fairly proven standard) As far as return rate, any design where only a single lava block borders the cobble will mathematically have the highest return probability, though perhaps higher if the lava block comes from above rather than the side... worthy of experimentation, at least... but anyway, my design should be near maximum for that, as well. Yours might be close, but the extra lava blocks in a row means a higher chance of a cobble falling into one of them.
Your design looks very intriguing as a lazy person's design, ideal for set and forget. Incidentally you could make it use only one lava source if you reduced the number of rows deep to 3. at 4 you require at least 2 lava blocks, though by no means are 4 necessary. I may try it out at some point when it comes time to refill my cobble stores (soon... I just lent over 10k cobble to a friend for one of their projects on our server.)