Temperature and Overclocking Testing
Cooler Testing Methods
To best gage the quality of the GPU cooler under review, GPU temperature was taken with the graphics card idle and under load. To replicate GPU idle conditions, the system was rebooted and allowed to sit idle for 30 minutes. To replicate a stress graphics load, FurMark was run over a 30 minute period running at 1280×1024 resolution and 4x MSAA in Burn-in Test mode. After each run, the system was shut down and allowed to rest for 10 minutes to cool down. This procedure was repeated a total of six times for each cooler – 3 times for the stock speed runs, and 3 times for the overclocked speed runs.
Temperature measurements were taken directly from the GPU thermistors using TechPowerUp GPU-Z v0.7.2. For both the idle and load temperatures, the highest recorded value in the application were used for the run. Note that the temperature values are reported as deltas rather than absolute temperatures with the delta value reported calculated as GPU temperature – ambient temperature.
Stock Temperature Testing
The graphics card temperature testing was conducted at stock speeds for the EVGA GTX 670 FTW card in two configurations – with the stock EVGA cooler and with the XSPC Razor GTX680 water block installed.
Just at stock speeds, the temperature difference is dramatic between the stock EVGA cooler and the XSPC Razor GTX680 water block. At idle speeds, the observed differences a full 8C drop in temperature with the card idling a mere 2C above ambient with the water block installed. The difference was found to be even greater under load with the card temperature measured at only 21C over ambient. Compared to the 55C over ambient temps measured using the stock cooler, that is quite a difference.
Overclocked Temperature Testing
Using the EVGA PrecisionX v4.20 overclocking software, the graphics card was overclocked to its highest stable settings with the XSPC Razor GTX680 water bock installed on the card. For details on the overclocked settings used, please see the overclocking section below.
Even with the added voltage and speed of the GPU and memory, the graphics card temperatures did not exceed 26C over ambient under load and 3C over ambient at idle. The measured load temperatures illustrate just how well the block is able to cool the card with the overclocked temperatures exceeding the stock speed temperatures by a 5C delta over ambient. XSPC definitely did their homework with the design of the Razor GTX680 water block.
While the a water block does not in any way directly influence the overclockability of the graphics card, it is an effect heat dissipater. This results in lower component heat, leading to better overclocking potential of the graphics card. This indirect relationship between card cooling and overclocking seemed to be the case with the XSPC Razor GTX680 water block. Overclock testing was performed in conjunction with the XSPC water block to demonstrate the card's increased performance potential.
With the help of EVGA's PrecisionX software (v4.20), we were able to get the GTX 670 FTW card running with a +100MHz GPU speed and a +650MHz memory speed. The actual running speeds after boost clock adjustments came out to 1228MHz GPU speed and 1879Mhz memory speed under load. For this overclock, the Power Target value was set to 145% with the GPU voltage set to 1.175V. We were able to get the card running at a +165MHz setting on the GPU and +750MHz on the memory, but it would not stabilize. Graphics card stability was tested by performing a full run through the Unigine Heaven benchmark at max settings and 1920×1080 resolution without crash or artifacting. Once the Heaven benchmark run stabilized, the card was torture testing over an eight hour period with FurMark running at 1280×1024 resolution and 4x MSAA in Burn-in Test mode.
EVGA PrecisionX profile settings
- GPU CLK OFFSET – +100MHz
- MEM CLK OFFSET – +650MHz
- POWER TARGET – 145%
- GPU VOLTAGE – 1.175V
- GPU Boost / K-Boost Clock Speed – 1228MHz
- Memory Speed – 1879MHz
- GPU voltage – 1.175V