View Single Post
  #19 (permalink)  
Old July 15, 2009, 05:46 AM
SKYMTL's Avatar
SKYMTL SKYMTL is offline
HardwareCanuck Review Editor
 
Join Date: Feb 2007
Location: Montreal
Posts: 11,616
Default

Believe it or not, I tested with Furmark and LinX and I got about 10-15% less power consumption than with my test. One of the reasons these reviews were put off for so long is that I tested EVERY combination of benching programs to find the highest power consumption.

I think the main issue is that Furmark needs CPU cycles to run and if the CPU is off running LinX, the GPUs get bottlenecked a bit. One the other hand, the 3DMark Batch Size test seems to use next to no CPU power when running at high res and high instances of AA.

Another interesting thing I logged was the fact that as Prime95 runs, power consumption gradually INCREASES.

Here are the results I got with the Extreme Load test. These were all peaks after 30 minutes with the Corsair HX1000:

LinX + FurMark: 1048W
LinX + 3DMark Batch: 1159W
Prime95 + Furmark: 1051W
Prime95 + 3DMark: 1166W

I couldn't believe it either but I went over the test again and agian with the same results.

To be honest with you though, the absolute highest load you can put on your GPU without worrying about the CPU getting in the way is the last benchmark 3DMark Vangate runs (I think it's Perlin Noise). Unfortunately, the way Vantage runs loops is a bit of a dog's breakfast with the benchmark reloading every time instead of being loaded from memory like 06 does. That means the load will not be applied in a linear fashion which is an issue for comparing efficiency and ripple.

Last edited by SKYMTL; July 15, 2009 at 05:53 AM.
Reply With Quote