An Interesting Difference Between AMD & NVIDIA Linux Drivers When Comparing System Usage

When running the tests recently for the NVIDIA Linux Driver 2015 Year-in-Review and How AMD's Proprietary Linux Driver Evolved In 2015, I also ran some extra tests comparing the AMD Radeon Software 15.12 and NVIDIA 358.16 proprietary drivers when looking at their CPU usage, memory consumption, and other system sensors.

From the same i7-5960X system running Ubuntu 15.10, I compared the system sensor output for the Radeon Software 15.12 and NVIDIA 358.16 drivers when testing two cards on each driver. On the Radeon side was the R9 290 and R9 Fury while on the NVIDIA side was the GeForce GTX 780 Ti and GTX 980.

While running these cards on their respective drivers, I tested Metro Last Light Redux, Unigine Valley, Team Fortress 2, BioShock Infinite, DiRT Showdown, and Tesseract. However, these results aren't about looking at the raw FPS. Rather, it's about the system vitals that were all tracked in real-time as prior to running the Phoronix Test Suite, the MONITOR=all environment variable was set for tracking all available and supported system sensors.

The reported system memory usage was noticeably higher for the AMD GPUs on the latest proprietary driver. They averaged out to memory use of ~2000MB with a peak of ~3000MB while the GeForce GPUs on the NVIDIA driver averaged out to ~1850 and a peak of ~2650. The minimum memory use was also clearly lower when using the NVIDIA proprietary driver. Of course, the NVIDIA driver performance was also faster.

In looking at the GPU usage as exposed by each of the respective driver's interfaces, it appears that the Radeon GPU usage bounces around a heck of a lot more than the NVIDIA drivers.

Similarly, the GPU frequencies / performance states seemed to bounce around a lot more on the AMD graphics cards with Catalyst than the proprietary NVIDIA driver.

When looking at the CPU usage, there wasn't a big difference between the GPUs/drivers for this Core i7 Haswell-E setup running Ubuntu Linux.

So while many Phoronix readers have been curious about CPU usage differences between Catalyst / Radeon Software and NVIDIA, there wasn't a big difference on that front. More interesting was the direct comparison of the GPU utilization and GPU frequency on these graphics cards between the two drivers: when using the proprietary Radeon driver, the utilization seemed to fluctuate much more than the NVIDIA proprietary driver, which in turn also causes the GPU performance state to change based upon the load. The system memory usage was also noticeably higher under the proprietary Radeon Linux graphics driver. Anyhow, more Linux benchmarks still forthcoming...