Published: June 22, 2018 | Source: Basemark | Author: Mark Campbell

Basemark GPU Performance Review

Conclusion

Basemark GPU offers us an interesting look into the world of graphical APIs, allowing us to get a glimpse into how modern development tools can enable higher levels of performance on today’s hardware.

In most cases, we found that the Kronos Group’s Vulkan API can offer a lot more performance than the OpenGL API it replaces, especially on Radeon hardware. At 1080p Medium settings, we found that our RX Vega 56 was able to offer a 3x boost in performance when using the Vulkan API, showcasing how limiting the ageing OpenGL standard can be. At the same settings, our Nvidia-based GTX 1080 test system was able to offer a performance boost of almost 85%, clearly showcasing the benefits of the modern Vulkan API.

Using the benchmark’s default 4K High settings, we found that our Nvidia hardware offered seemingly identical performance when using both OpenGL and Vulkan, suggesting that there is no performance difference when the benchmark is GPU compute limited. Even so, AMD Radeon graphics cards continued to offer a significant boost in performance, though this was only enough to bring AMD’s offerings in-line with their Nvidia counterparts.

In general, Nvidia’s Geforce graphics hardware offered higher performance levels than their Radeon equivalents, though this factor could be due to Basemark’s apparent lack of work with AMD during Basemark’s development. In their official press release, Basemark explicitly mentions working with “Imagination Technologies, Intel, NVIDIA, Qualcomm and Renesas”, with no mention of either AMD or Radeon.

Another factor that is worth mentioning is that AMD Radeon graphics cards did not run this benchmark correctly when using AMD’s latest Radeon Software Adrenalin 18.6.1 drivers, forcing us to roll back to version 18.5.1. It is strange for a dedicated benchmarking utility to break when using AMD’s most recent drivers, as this should be something that Basemark should have checked/validated before launch.

Will we use this benchmark for future graphics card testing? Probably not, though we do look forward to having a more in-depth look at the tool when DirectX 12 support is added to the utility. If this test cannot run reliably on AMD/Radeon’s latest driver releases, we cannot consider Basemark GPU as a worthy addition to our standard graphics card test suite, though at least we have a useful tool that showcases the benefits of the Vulkan API.

You can join the discussion on Basemark’s new Basemark GPU benchmarking utility on the OC3D Forums.