The Power Efficiency From A Radeon HD 4890 Through The RX 480 & R9 Fury

[phoronix]

Phoronix: The Power Efficiency From A Radeon HD 4890 Through The RX 480 & R9 Fury

This past weekend I published a number of year-end 2016 AMD Linux benchmarks on a wide-range of AMD GPUs going back many generations while using the Linux 4.9 kernel on Ubuntu along with the Mesa 13.1-development code for having the newest open-source Gallium3D drivers. Those results were very interesting and go check them out now if you haven't done so already. For this article is a sub-set of those tests carried out again while monitoring the AC power consumption, GPU temperature, and CPU utilization while also automatically calculating the performance-per-Watt.

http://www.phoronix.com/vr.php?view=23889

[Adarion]

Is it really fair to compare under 4K(+) conditions? I guess a lot of cards will struggle with memory transfers cause VRAM is always full. My CPU also becomes less efficient during an emerge compile session if the system goes into swapping or if my data media is connected via a bottleneck.

[GraysonPeddie]

Would be interesting to hook up 3 4K monitors and see how Portal runs with the 480.

Oh, I'm just way too crazy, am I? (Chuckle) 3840x3 = 11,520. Woah, that's a lot of horizontal pixels to render... I think a 21:9 monitor with a resolution of 5040x3840 would probably be more sane than having 3 monitors.

[GreatEmerald]

Hm, seems like there was a major jump in efficiency from VLIW to GCN, and again to Polaris. Pretty cool.

[donbastiano]

Would be interesting the 470 too

Can I ask how to read/interpret for example the "System Power Consumption" graphic wich is usually for me really interesting? Is the area related to something like standard deviation? Is the vertical line in the middle the aritmetic average? Are these information written somewhere in the site?

[Michael]

Would be interesting the 470 too

Can I ask how to read/interpret for example the "System Power Consumption" graphic wich is usually for me really interesting? Is the area related to something like standard deviation? Is the vertical line in the middle the aritmetic average? Are these information written somewhere in the site?

RX 470 wasn't used in this article since my only RX 470 is factory overclocked so rather skews it for cases like this with power testing.

Regarding graphs: https://en.wikipedia.org/wiki/Box_plot

[donbastiano]

RX 470 wasn't used in this article since my only RX 470 is factory overclocked so rather skews it for cases like this with power testing.

Regarding graphs: https://en.wikipedia.org/wiki/Box_plot

Thank you ALSO for both the info

[roboman2444]

Is it really fair to compare under 4K(+) conditions? I guess a lot of cards will struggle with memory transfers cause VRAM is always full. My CPU also becomes less efficient during an emerge compile session if the system goes into swapping or if my data media is connected via a bottleneck.

The 4k is probably to make sure that there isn't a CPU bottleneck on the higher end cards.

That being said, Run this test in another year or two and im sure the 480/fury will perform even better when drivers mature.

[molecule-eye]

Love these sorts of tests. Interesting to see how well the R9 285 stacks up against Polaris. I had a budget for either a new RX 460 or else a used R9 285, and I was concerned with performance-per-watt (and of course performance, period), and ended up going for the aging 285. Seems like a decent decision in the end.

[pheldens]

thanks for the test, I just ordered a passively cooled rx460 for christmas