AMD Phenom II X4 965 Black Edition @ 3.94GHz
4GB Corsair XMS3 DDR3 @ 1333MHz
Gigabyte MA770T UD3P Motherboard F4 BIOS
Asus Radeon HD 5850 1GB GDDR5 Graphics Card
Samsung Spinpoint F1 320GB SATA II HDD
LG 22x DVD+/-RW SATA
Arctic Freezer 64 Pro CPU Cooler
Windows Vista Home Premium
We used Furmark's stability test to record the Asus Radeon HD 5850's load temperatures. After 30 minutes of idle monitoring from a cold boot, the test was loaded and was ran until temperatures and fan speeds had appeared to have stabilised. The results were as follows.
Like all of our previous Radeon HD 5800 graphics cards, the load temperatures maxed out at 80*c with a fan duty speed of 35%. The noise levels even on our open test bed was tolerable. While the graphics card was obviously audible under load, the noise was the sound of airflow rather than motor noise. There's minimal buzzing and more importantly no whine. As said in previous reviews, this is to be expected given that Asus have not made any changes to the reference heatsink for their 5800 series graphcis cards.
Asus SmartDoctor & Overclocking
As you may or may not know, ATi's own Overdrive utility in the Catalyst Control Center offers overclocking support for the Radeon HD 5850. Aside an Auto Tune feature, it also offers manual overclocking support. The only real nuisance however is that it's maximum boundaries stand at a paltry 775MHz on the core and 1125MHz (4500MHz GDDR5 effective) on the memory. Understandable I suppose given that ATi need to keep the graphics card as far away from the current HD 5870 flagship as possible given the £100 price difference.
This is where Asus' own overclocking application comes into play. Not only does it offer frequency increases to insane heights, it is also meant to offer Voltage Increase support. The keyword however is meant as no matter what we did, we could not get the application to select a voltage higher than the HD 5850's default 1.088V.
We found our maximum stable frequency at around that of the Radeon HD 5870's defaults, at 854MHz core and 1240MHz (4960MHz effective) memory. A 129MHz / 240MHz increase shouldn't be frowned upon of course. The graphics card was returned to it's default clockspeeds for the testing that was to follow.