NVIDIA GeForce RTX 3070 Ti Linux Performance

With the Strange Brigade running on Linux via Steam Play, in this instance the RTX 3070 Ti performance nearly matched the Radeon RX 6800 XT.

Those wishing to see even more of these Linux gaming benchmarks across the tested GPUs can go to this OpenBenchmarking.org result file.

Across all of the tests conducted, the GeForce RTX 3070 Ti had an average power draw of 226 Watts and a peak of 289 Watts, right in line with the reported 290 Watt graphics power. This 226 Watt average is very similar again to the TITAN RTX which was delivering similar performance in these benchmarks. The RTX 3070 (non-Ti) meanwhile saw an average power draw of 170 Watts across these tests.

When taking the geometric mean of all those benchmarks from the OpenBenchmarking.org result file above, the RTX 3070 Ti performance comes down to being 8% faster than the conventional GeForce RTX 3070. This put the RTX 3070 Ti performance overall right behind the TITAN RTX and RTX 2080 Ti. Compared to the RTX 2070 SUPER, there was a 38% generational improvement in performance or 60% against the RTX 2070 non-SUPER.

When it comes to the Vulkan ray-tracing performance, NVIDIA easily wins there right now. Not only does NVIDIA offer better performance but the AMD Radeon Vulkan ray-tracing support is rather limited. The Vulkan ray-tracing support was just recently introduced to the Radeon Software "PRO" packaged driver on Linux and isn't yet supported by their official AMDVLK open-source Vulkan driver nor the Mesa RADV Vulkan driver. So to even use Vulkan ray-tracing with AMD on Linux right now leaves only the rather unpopular closed-source Vulkan driver component and even then there are various bugs in the early implementation. As shown by the geometric mean of those Vulkan ray-tracing benchmarks, the performance is quite poor.

See this OpenBenchmarking.org page for those Vulkan ray-tracing benchmarks with the limited selection right now of Quake II RTX and Ray Tracing In Vulkan for Vulkan RT Linux benchmarks.

From these OpenBenchmarking.org links you are also able to generate your own performance-per-dollar metrics based on your local/available pricing and other analysis.

Coming up this week still is a thorough look at the GeForce RTX 3070 Ti / RTX 3080 Ti compute performance alongside other RTX 20 and RTX 30 series graphics cards. Overall the RTX 3070 Ti performance is coming in right where one would expect with the RTX 3070 Ti offering more CUDA cores, slightly higher clocks, and GDDR6X video memory compared to the non-Ti RTX 3070. Thanks to NVIDIA for providing the review samples for today's testing.