OpenGL Performance & Perf-Per-Watt From The Radeon HD 3850 Through R9 Fury

Written by Michael Larabel in Graphics Cards on 3 June 2016 at 12:00 AM EDT. Page 6 of 6. 28 Comments.

For ending out this fun comparison are results from the most demanding Linux demos: Unigine Heaven and Valley.

Unigine Heaven was run at just 1280 x 1024 due to being so demanding on the graphics hardware. There's a big caveat worth pointing out with these numbers is the lack of tessellation on the older hardware. Nevertheless, I left these numbers in since they still are meaningful for showing the difference to the other GPUs with tessellation as well as the power efficiency.

Thanks to the beauty of OpenGL, we can still run such comparisons.

With Valley not being tessellation heavy, the data is more representative. The numbers do indicate though that there appears to still be some optimizations needed for the more open-source Linux driver stack for newer GPUs, such as pointed out by the Radeon R7 370 when compared to the GPus with more mature open-source driver support.

Before closing, a look at the system power use during all of the OpenGL benchmarks that were run for this article:

Well, that's our Radeon comparison on the latest open-source drivers (Linux 4.6 + Mesa 12.1-dev) in marking Phoronix being around for 12 years now. Hopefully you found these numbers interesting for one reason or another. It was certainly a blast pulling out some of these old cards and still seeing them fire up on a modern Linux operating system. It will also be fun to see how these numbers compare later this month to the Radeon RX 480 Polaris graphics card and I may even toss these numbers in a direct comparison to some NVIDIA hardware, including the GTX 1080 Pascal.

If you wish to see how your own Linux system(s) OpenGL performance compares to the many AMD data points in this article, simply install the Phoronix Test Suite and simply run phoronix-test-suite benchmark 1606036-HA-BIG12BIRT49 for carrying out your own automated, side-by-side performance comparison in a standardized manner to the results found within this article.

Lastly, if you'd like to support all of the Linux benchmarking work done at Phoronix.com over the years and aren't yet a member of Phoronix Premium, please consider taking advantage of our 12th birthday special or at the very minimum please do not view our site with ad-blockers. Coming up in the next few days are still some other interesting Linux hardware comparisons.

28 Comments