Looks like Heatsink Factory is going to get some soon.
If you believe this review
(some don't buy the use of reference heat sources), the SLK-900 only does marginally better than the SLK-800 using the same fan.
I see the mounting system as an advantage- I was considering a Swiftech MCX462+ simply because I didn't want to worry about ~2 pounds pulling on 6 little plastic ears.
The SK-7 (which is US$25 vs. $36 for the 800 and $55 for the 900) comes in right behind the 800.
Comparing my present stock (retail) Athlon cooler which does 0.50 C/Watt vs. 0.25 C/Watt SK-7 + Tornado I see that the latter gets 1/2 the temperature rise whilst using a fan that does nearly 4x the CFM in free air (23 vs. 84).
That makes me wonder whether the surface area of the die (which is a constant) limits performance. Copper may do better than aluminum, but more air appears the largest factor.
Now running larger fans slower lets us get more CFM/dB, but there is a limit to this because no one seems to make fans that turn below maybe 1200 RPM. I have a 120MM Panaflo L1A case fan and with a rheostat adjusted so it always starts (about 6.5V when running), it's louder than an 80mm M1A controlled by a NoiseKiller (which has to have a similar minimum voltage because power-cycling the computer with the heat sinks still hot never results in a stall).
Now the 120mm is probably blowing more air than the 80, but unless I resort to PWM and growling, I can't get a quieter & slower 120mm. They just don't seem to be made.
I conclude that bigger is not better if you can't get the airflow to the minimum actually required. If an 80mm L1A at 7V on an SLK800 will keep my Athlon XP 1.4GHz cool enough, a 92mm L1A at 7V on an SLK900 is just going to be louder.
The SLK900 might be a good choice for using a duct from the case fan and reducing the total number of fans though.