Results: Euler 3D, Unigine Valley, and Power
This is a scientific application that simulates air flow over an air foil, and it heavily relies upon a robust floating point unit. It also really seems to favor Intel based parts. I test 1, 2, and 4 threads in 20 steps.
Single thread performance in this application, with the 5350, is a painful experience. 15 minutes to complete the first test! Doesn’t AMD realize that I am on a deadline here! All joking aside, this is a pretty unfair test for the 5350. HPC applications will not make their way onto this particular part.
This is the latest benchmark from the nice Unigine people that really pushes DX11 based parts. I run the benchmark at 1280×720, high quality preset, and no AA.
This application really exposes the performance differences between the much larger part and the little old Kabini. Previous benchmarks in this review show it to be within around 50% of performance of the larger part, but in this case it was just pushed a little too hard and is sitting in the 33% range. Memory bandwidth is going to play a big part in an application like this.
Now we finally come to the heart of the matter. Power is the primary goal of the Kabini design, and now we get to see if AMD delivers on that promise.
Idle numbers are not terribly impressive, but I honestly cannot see them going much lower. The 5350 is still 10 watts lower than the 6790K. Both parts exhibit aggressive clock throttling and gating. Add into that equation the inefficiency of a 500 watt power supply pushing 23 watts… we can see that the potential for a system with a much smaller and efficient power supply could potentially get down to the 12 watt range.
At load we see a much different and compelling story! The 5350 is pulling 37 watts at max. The 6790 is way up at 117 watts! Sure, overall the 5350 is about half of the performance of the 6790, but it consumes a third of the power at the wall. We would again see more gains with a smaller power supply. At least with the 6790K and its 117 watts it is getting into the higher efficiency range of this particular 500 watt power supply. 37 watts isn’t even pushing this supply in the least.