Correct, we are looking at lm/PAR per watt, Correct?
Probably getting close to that, but still a while till it filters down to the market place.
The current Cree XP-G series produces 139lm per Watt. Higher than T5 or halide. You'll find them in Vertex products. Certainly some others by now.
This is not quite true, is the current we are watching, not voltage as LEDS are CC(constant current) devices.
You're splitting hairs, as a constant current is a stabile amperage with a flexable voltage. If you were to change both you would change the efficiency of the LED drastically.
Mmmm... well 99.9% of LED devices (well any decent devices) are driven with PWM, using a linear driving circuit would be defeating the object of trying to get better efficiency rating.
Really splitting hairs. Get me a fuzz duster!

But correct!
Totally lost here! could you please elaborate.
OK, this is the difficult part to explain. Most diy and subsequently most marketed models, placed the leds relatively evenly over the heatsink and attempted to cover most of the surface of the tank with this spread. For diy it made it easier to mount the leds, but we quickly noted that we were having light spill due to the wide angle of the led lens (120°-125°). Many opted for an optic to focus the light, few thought about placing the leds right next to each other and creating a point light source. Technically, this was tough for a diyer. Very fine work. It still is. But the optic made the looser spacing a good, if more expensive, option.
In Germany we have quite a few companies clustering the leds and placing them toward the middle of the lamp, thus giving a 120° angle. Relatively wide, but no wider than a previous halide. This option does not require optics to bundle the light.
Now, let me discuss optics a bit. They have clear advantages in that they bundle light into a much tighter beam, allowing the aquarist or other user to aim their photons! The first disadvantage is the loss of light an optic causes. They are typically plastik and are relatively rough in design (compared to a camera or telescope, in Ex). This causes a direct loss of up to 25% of the entering light. Most better optics can give between 85%-90% transparence. Still a lot of light loss. Also, when the light hits the water, we start getting reflection and refraction, which causes further light loss. Even though you are capable of focusing the light into a very tight beam, you will still have loss. Of course, this tight beam can then be hung 3ft over the tank, if that is the plan. A decision point.
Another approach is to go without an optic, which immediately gives you more light entering the water. OK, as the led light angle is wide, the lamp should be about 15cm/6" above the water surface for best performance. I like this aesthetically, a personal decision. Now, light entering water is directly refracted, this would be with or without an optic. Water will reduce a 120° angle to about a 95% angle. Sort of an automatic focus. When the light hits the glass, it is further altered, with most of it being reflected into the tank and about a 10% loss to the outside. This will depend on the height of the lamp, as we can see, with the deeper the water, the greater the light loss, to a point, as this is not a linear loss, Light does not follow a straight line in water, it is constantly reflected and refracted. If you see this plotted, you do not get a perfect cone rather a tear shaped form.
OK, what does this mean. Well, without optics you can actually get more light into the tank than with and without any spotlighting effect (when light is bundled via an optic, the cone of light is widening as it heads down. Parts of the tank will be more or less strongly lit, due to this and overlap of other optics.), which many find disturbing.
In the end, one can balance the equation/decision in saying, optics give you greater variance in hanging height, but cost quite a bit mor money. They lower the light energy in general by more than 10%. Without optics the leds must be clustered to create a point light source toward the middle axis of the tank, which assures even light distribution and minimal light spill. To take advantage of this, the lamp must hang closer to the water surface for most tanks (rule of thumb, 6"/15cm above a 24"/60cm wide tank).
Does this paint a better picture?
