I notice the GT's are clocked above 600 Mhz for the GPU core.
My GTS is stock-clocked at 550, and I'm running it at 580 with a 900 Mhz (1,800 DDR) memory clock.
I'm using the HR-03-Plus with an 80x15mm fan -- but it's also ducted with air being sucked from the HR-03 fins by a 120x38mm rear-exhaust fan. The 80mm pushes air from the GPU-component side of the board blowing directly on the GPU's heatsink and the heatpipe "hot ends," while the 120mm fan pulls air from the opposite (top) side of the VGA and from narrow apertures near the edge of the motherboard.
Running "heat-up" mode in the ATITool "3D View" mode, the "GPU_Diode" temperature goes up to about 63C with a 74F room-ambient. That corresponds to a "GPU" temperature of 55C.
For control purposes, I measured my stock-cooler temperatures with S&M's vga-heat-up sequence and an hour of gaming, respectively, and the "GPU" temperature would post as 70+C at 75F room-ambient. Temperatures were measured with Everest or nTune[ nvidia monitor], as those coming from S&M are known to be skewed.
My ducted HR-03 only trumps the unducted Zalman VF-1000 by 4C degrees using a very similar eVGA 8800 GTS model -- at similar stock settings (48C-load for mine; 52C-load for the Zalman). So I'm guessing that some airflow enhancements to the Zalman might reap at least a few-degrees-C in improvement, also.
So for reasons related to potential PCI-E or PCI slot interference, I would pick the Zalman VF-1000 as an alternative to the HR-03-Plus -- as needed. The biggest problem with the HR-03 is the size of the fins and case-volume and slot-clearance consumed by them; the biggest advantage of the HR-03 is that the fins can be deployed in two different orientations on either side of the VGA card.
But given the measurements I quoted here, either of the two coolers provides the near-equivalent advantage in thermal resistance and therefore cooling.