|Reduce Sapphire Toxic Radeon HD4870 Fan Noise
The Sapphire Toxic Radeon HD4870 512MB graphics card was recently reviewed at Bigbruin.com and it definitely proved itself as a high end card ready for any 3D gaming action you could throw at it. The main downside was that the fan on the funky looking cooler was noisier than what I considered to be acceptable. Considering that the GPU temperature on the Sapphire Toxic Radeon HD4870 ran 19 to 31 degrees Celsius cooler than the GPU on a stock Sapphire Radeon HD4870, it seemed like a better balance between noise and cooling could be found.
Just before publishing the review of the Toxic Radeon HD4870, the representative from Sapphire sent an e-mail with two files attached; ATI WinFlash v2.0.15 utility and BIOS file 85VCTOXI.008. The BIOS file was developed specifically to address complaints about noise with the Toxic Radeon HD4870, and the flash utility provides a simple, Windows-based means of applying it. I finally had an opportunity to install this BIOS file, and decided to document the results.
A system with the following components was used to test the thermal results of the Sapphire Toxic Radeon HD4870 512MB graphics card before and after the BIOS flash:
» ASUS P5E64 WS Evolution X48 ATX motherboard
» Intel Core 2 Duo E8400 dual core processor (at 3.2GHz)
» G.Skill 4GB DDR3-1600 Pi Series dual channel memory kit (at 1600MHz and 7-7-7-18)
» OCZ Technology Vendetta 2 CPU cooler (set to run quietly at 900RPM)
» Maxtor MaxLine III 250GB SATA 3Gbps hard drive
» TSST Super WriteMaster optical drive
» Nesteq EECS 700 Watt ultra quiet power supply
» Windows Vista Home Premium (64-Bit)
CPUID's HWMonitor 1.11 (64-Bit) was installed to monitor the temperature of the GPU, and a Radio Shack digital sound level meter (Model: 33-2055) was used to try to monitor the noise produced.
The idle condition was established by letting the system sit at the Window's desktop with nothing running but HWMonitor. Once the temperature stabilized, the current reading (labeled as "Value") provided by HWMonitor for the GPU was recorded. The load condition was established by running FutureMark's 3DMark Vantage in three consecutive loops in "High" mode, and after that ATI Tool version 0.27 Beta 3 was allowed to stress the card for and additional 30 minutes. The maximum reading (labeled as "Max") provided by HWMonitor for the GPU was then recorded. Of interest is that the maximum temperature occured during the ATI Tool session, and was five degrees higher than the maximum achieved by 3DMark Vantage alone. An ambient temperature of 21 degrees Celsius was maintained thorughout testing.
The noise level meter proved to be of little use during testing, as it was not sensitive enough to provide any useful information. Even with the meter's sensor placed about 12 inches from the fan, it only wound up being able to register a reading while the card was in the load condition, prior to the BIOS update. The minimum value the meter will display is 50 dBA, and this level apparently wasn't exceeded at all after the BIOS update, and only under the load condition prior to the BIOS update (hitting 55 dBA). Therefore, more subjective terms will be used to describe the noise output.