AMD Phenom-II X3 720 BE
XFX nForce 750a SLI
2x2GB OCZ Fatal1ty PC2-6400
1 EVGA GTX 260 55nm (GPU1) with the Thermalright T-Radē GTX
1 EVGA GTX 260 65nm (GPU2) with the Thermalright HR-03 GTX
I will be using the GPU test portion of OCCT version 3.1.0 to provide the graphics load and data logging of the core temperature. All tests were conducted with both cards installed together in SLI. GPU1 represents the 55nm card with the T-Radē, and GPU2 is the 65nm card with the HR-03. First I ran a baseline test to record how the 55nm card with stock cooler performed in comparison to the 65nm card with HR-03.
GPU1 with the stock cooler recorded an idle temp of 52°C and averages 81°C under load, with a peak of 84°C. GPU2 with the HR-03 installed registers only 41°C at idle, and averages just 60°C under load, with a peak at 65°C. That's basically a 10°/20°C temperature difference at idle/load between the stock and Thermalright cooler, which matches what was observed previously during the HR-03 GTX review.
After installing the T-Radē GTX onto GPU1 and attempting to insert the card back into the system, it was discovered that the second 92mm fan had an interference issue with the motherboard SATA ports. This board has six ports, three of which had been blocked before by the stock cooler, and now with the T-Radē and both fans in place all six would be inaccessible. To remedy this I first removed one of the 92mm fans. I became concerned during the subsequent test however when the temperature skyrocketed past 80°C, and quickly placed the unmounted fan loosely back into position at the right side of the cooler as pictured above.
As you can see from the graph, the idle temp began at a respectable 45°C, however by six or seven minutes into the test GPU1's temp had exceeded 80°C at which point I reconnected the second 92mm fan. This brought the average load temp back down to 73°C but still peaked at 79°C. When comparing the thermal radiation given off by the T-Radē vs. the HR-03, the heatpipes on both coolers were hot to the touch, as were the fins on the T-Radē. However the fins on the HR-03 were only warm. The dual 80mm fans on the HR-03 are run at just +7v for reduced noise, and the overall airflow between those and the T-Radē's dual 92mm fans seem to be comparable, but I suspect the HR-03 may have more fin surface area or else a better design. Determined to see if the T-Radē GTX performance could be improved, I decided to install a different fan.
The unit pictured above is a 120x38mm NMB fan, rated at 130cfm and 50dBA. Obviously you would not want to run this monster at full tilt for daily use, unless your computer happens to be located in the next zip code. But for purposes of this test I wanted to see if a significant increase in airflow would result in an equally impressive boost in performance for the T-Radē. After all, the fans that Thermalright includes with the cooler seem to be tailored more for quiet operation than tons of airflow. Although mounting holes are located in the center of the bracket for installing 120mm fans, I had to offset it to clear the SATA ports on the motherboard.
Now with three times the amount of air blowing through the T-Radē fins, the idle temp has been reduced by 5° to 40°C, beating out the HR-03 by a single degree. Load temperature averaged just 55°C, with a peak of only 57°C, a whopping 20°C difference from the results obtained with the original fans and even managing to beat the HR-03 by several degrees. Granted it took installing a hairdryer over the fins to do it, but this shows that there is definitely more performance to be had from the T-Radē GTX by simply increasing the amount of airflow.
Let's wrap up this review with some final thoughts and conclusion.