BENCHMARKS I ran the following three tests. The time duration was chosen after
monitoring each test to see where the temperatures leveled off under load. The
card's ambient temperatures are also reported. The machine was shut off for 10
minutes between each of the three main tests.
UPDATE: Ambient room
temperature was 20°C and the ambient case temperatures were 25°C at
idle and 28°C under load.
UPDATE: I had already removed the stock thermal compounds in
my review of the BFG 6800 Ultra OC. At that time I then replaced with Arctic
Silver 2 on the GPU core and Ceramique on the memory modules. So, the "stock"
comparisons below are actually better than the original factory compounds.
I also tested at stock speeds of
425/1100 and overclocked speeds of 460/1250. However, there was no appreciable
difference in temperatures. The Silencer NV 5 did an excellent job at
keeping even a well overclocked GPU and memory at the same temperatures found at
Far Cry v1.3 - Research Map - 1600x1200 - 4xAA/8xAF - looped 10 times
I recorded the temperatures with the NVIDIA Forceware Control Panel. I also
used SpeedFan v4.20
(after making the necessary +10°C offset for my GPU core
temp sensor reading) as a secondary measurement for comparison.
Tests were taken with my case side panel both on and
off. I have four Vantec Stealth 80mm case fans (one topside blowhole, two rear
exhaust and one side intake). Also, the fans were left running when the side
panel was removed. Yes, I need a front intake, I know...and my wire management
is in a state of transition. :)
REMINDER: The 6800
Ultra stock cooling is actually quite good. So, I wasn't expecting the near 20°C
drop in load temperatures that some have been reporting on the 6800 GT. The 6800
GT stock cooling is clearly not as good as 6800 Ultra's. Let's find out how much
of an improvement the Arctic Cooling Silencer NV 5 can make over the
already excellent 6800 Ultra stock cooling.
Case Opened Results
In the results above you can see that there's a nice 4°C
drop for the GPU at idle in Windows. We see a bit more cooling occurring in
RTHDRIBL and Far Cry. My goal is to always achieve < 70°C under load which I
reached easily with the side panel off.
Case Closed Results
First of all, the results above clearly show that I need to improve the
cooling within my case when closed. I've never been a stickler for case cooling
but I'm leaning more that way now. This case cooling issue of mine has a nice
side effect though. It really helps to demonstrate how well the Arctic Cooling
Silencer NV 5 performs.
We're about the same idling in Windows. What's impressive is the 8°C
and 11°C reduction in temperatures with RTHDRIBL and Far Cry respectively. Also
worth noting is the 3°C reduction in ambient temperatures whereas the there was
a slight increase with the side panel off.
SpeedFan provides it's own graphing over time for the
temperature sensors that you select. I thought I'd include these results as
well. It's always nice to be able to see the trend over time. Below is a
screenshot of the entire window and then following are just the graph portions.
SpeedFan Temperature/Time Graph
Even though I didn't include overclocked results in the graphs
above, I thought I'd include them here for reference.