Given all the recent performance work that's landed recently in Mesa Git for Mesa 17.1 plus the Linux 4.11 kernel continuing to mature, in this article are some fresh benchmarks of a few Radeon GPUs with Mesa 17.1-dev + Linux 4.11 as of this week compared to some GeForce graphics cards with the latest NVIDIA proprietary driver.
For those curious how AMD's RadeonSI Gallium3D driver for GCN GPUs has evolved, here are benchmarks with two graphics cards showing how the RadeonSI Mesa performance has evolved since Mesa 11.1 going back to late 2015.
With Mesa recently landing their RadeonSI GLSL on-disk shader cache and enabling it by default plus other recent optimizations, plus in kernel-space there now being Linux 4.11-rc1 and that showing potential improvements, here are some fresh benchmarks of AMD Radeon vs. NVIDIA on Ubuntu Linux.
With the Linux 4.11 merge window now closed and the RadeonSI shader cache having landed and even turned on by default, it's a great time to run some fresh benchmarks of the open-source AMD Linux graphics driver stack. Here are some benchmark results with the latest Mesa Git code for RadeonSI Gallium3D and RADV as well as the Linux 4.11-rc1 kernel compared to Linux 4.10.
A few days ago I posted some results of surprise performance improvements for a Radeon RX 470 when testing the DRM-Next code queued for Linux 4.11. I've now tested that kernel on more systems and can confirm at least benefits more widespread for RADV's Vulkan performance.
With Vulkan turning one year old I decided to run some fresh comparison benchmarks of Mesa 17.1-dev RADV (as well as some RadeonSI OpenGL results for reference) compared to AMD's latest public hybrid driver release, the AMDGPU-PRO 16.60.
With Mesa 17.0 due to be released any day now, here are fresh benchmarks of Mesa 17.0's Git code as of Friday compared to Mesa 12.0.6, Mesa 13.0.4, and the current Mesa 17.1-devel Git master code. Not only is the i965 OpenGL driver performance being examined but also the ANV Vulkan driver present since Mesa 12.
With Mesa 17.0 due to be released in the days ahead, I've been running fresh benchmarks of this latest user-space 3D driver stack on Intel, Radeon, and Nouveau. For your viewing pleasure this Thursday are the RadeonSI benchmarks comparing the Mesa 17.0 Git code to that of the latest Mesa 13.0 branch with a few different AMD graphics cards. There are also some tests of the RADV Vulkan driver.
Mesa 17.0 is due to be released this month and is the biggest feature release we've seen in quite a while for this 3D user-space driver stack. Here's a recap of the exciting changes to find with Mesa 17.0.
Following last week's AMDGPU-PRO 16.60 hybrid driver release I delivered some early AMDGPU-PRO vs. AMDGPU+RadeonSI benchmark results using the newest driver code. After a few more days of testing, in this article is a larger OpenGL and Vulkan comparison when testing AMDGPU-PRO 16.60 and AMDGPU+RadeonSI of Mesa 17.1 + Linux 4.10 on various Radeon GPUs. On the NVIDIA side are fresh GeForce tests with the company's newest 378.09 beta driver.
Given yesterday's release of the AMDGPU-PRO 16.60 driver I've been busy running various benchmarks on this first AMD Linux hybrid driver release of 2017. A number of OpenGL benchmarks will be published this weekend compared to the latest Mesa RadeonSI Git driver while for your viewing pleasure today is a look at the Vulkan performance of AMDGPU-PRO 16.60 compared to the Linux 4.10 + Mesa 17.1-dev driver stack for Dota 2 and The Talos Principle.
For those curious about the latest Linux gaming performance numbers for the latest Linux 4.10 Git kernel plus Mesa 17.1-devel on Git master for Radeon GPUs compared to the latest NVIDIA Linux driver release (378.09 beta), here are some fresh benchmarks. A range of OpenGL and Vulkan performance tests showing the latest NVIDIA and AMD Linux graphics performance with the newest drivers as of this week.
Recently on Phoronix we've tested the re-clocking and boost support in Nouveau with the Linux 4.10 kernel and separately landing in Mesa 17.0 Git was the big Maxwell performance boost for Nouveau Gallium3D. That Gallium3D driver work improves the Maxwell open-source performance by "1.5x to 3.5x" via instruction pipelining improvements. With those latest improvements in the kernel and Mesa, how does Nouveau now compare to NVIDIA's binary Linux driver?
Last month with AMD/GPUOpen's ROCm 1.4 release they delivered on OpenCL support, albeit for this initial release all of the code is not yet open-source. I tried out ROCm 1.4 with the currently supported GPUs to see how the OpenCL performance compares to just using the AMDGPU-PRO OpenCL implementation.
In addition to Nouveau Gallium3D seeing a performance boost last week, last week Intel's Vulkan driver also seen some interesting work around HiZ. Here are some fresh benchmarks showing recent performance improvements to the Intel "ANV" Mesa Vulkan driver plus some fresh OpenGL benchmarks too.
Landing this week in Mesa 17.0-devel Git was OpenGL 4.3 for NVC0 Maxwell and a big performance boost as well for these GeForce GTX 750 / 900 series NVIDIA "Maxwell" graphics processors. Here are some before/after benchmarks of the performance improvements, which the patch cited as "1.5~3.5x better", when testing a GeForce GTX 750 Ti and GTX 980.
With word of Fedora switching away from using the Intel X.Org driver in favor of the generic xf86-video-modesetting driver, following in the steps laid by Debian/Ubuntu, there is fresh discussions over features and any performance impact of xf86-video-modesetting vs. xf86-video-intel DDX drivers. As such, here are some fresh 2D and 3D benchmarks.
For those curious about the performance difference if upgrading to third-party PPAs from Ubuntu 16.10 when using a modern AMD Radeon graphics card with the open-source driver stack, here are some fresh numbers.
Continuing on from this weekend's open-source Nouveau vs. closed-source NVIDIA Linux driver performance are results now added in with showing AMD's open-source vs. closed-source driver performance with the same tests.
Earlier this week I posted some benchmarks showing the open-source NVIDIA (Nouveau) driver performance on Linux 4.10 with the new NvBoost capability for finally being able to hit the "boost" clock frequencies with Kepler graphics cards when using this reverse-engineered driver. While the manual re-clocking and enabling NvBoost is able to increase the Nouveau driver's performance, how do these results compare to using the closed-source NVIDIA Linux driver? These benchmarks answer that question.
As mentioned earlier when posting some fresh AMD Kaveri vs. Intel Linux graphics benchmarks, I have some fresh AMD A10-7850K "Kaveri" APU numbers with running the latest Ubuntu 16.10 + Linux 4.10 + Mesa 13.1-dev stack on many of my benchmarking systems in the basement server room. With having an A10-7850K Kaveri system running with the latest Linux open-source driver code, I figured I'd compare it to some of my older Kaveri results.
After the Nouveau DRM driver updates didn't make it for the Linux 4.9 merge window, this open-source NVIDIA graphics kernel driver saw significant updates for Linux 4.10. Nouveau in Linux 4.10 has atomic mode-setting, DP MST support, a LED driver for controlling the cards that have the illuminated "GeForce" logo, NvBoost support for hitting the higher boost frequencies on supported cards, and many other changes. Here are some fresh benchmarks of Nouveau with the Linux 4.10 kernel.
With the Linux 4.10 kernel there remains experimental Kconfig switches for being able to build the Linux kernel with GCN 1.0 "Southern Islands" and GCN 1.1 "Sea Islands" support in the newer AMDGPU DRM driver rather than the mature Radeon DRM driver. For your viewing pleasure today are benchmarks of a few GCN 1.0/1.1 GPUs when testing the Linux 4.10 Git kernel with Radeon DRM and then the experimental AMDGPU DRM driver while both kernel drivers were tested in conjunction with the same Mesa 13.1-dev snapshot as of this week.
Last week I published a 31-way Linux graphics card comparison with an assortment of both NVIDIA GeForce and Radeon graphics cards using the latest Linux drivers. I also published a variety of Vulkan benchmarks. In those tests the open-source Radeon driver stack was used given that's what AMD is endorsing these days for Linux gamers with AMDGPU-PRO not even working on all modern Linux distributions. But for those curious how AMDGPU-PRO compares to those big result data-sets, here are those -PRO results to share today.
For those curious how the latest open-source AMDGPU+RadeonSI driver code is comparing to yesterday's AMDGPU-PRO 16.50 release, here are some fresh OpenGL Linux driver benchmarks from a few AMD graphics cards.
The embargo has just expired on the Radeon Software Crimson ReLive Edition release, which had leaked out a few days ago elsewhere. While Crimson ReLive Edition is a big update for AMD Windows users, on the Linux side there's less to talk about but it's still a rather big release. I've been testing the AMDGPU-PRO 16.50 release the past two weeks and overall it's a sizable update for those using this hybrid AMD Linux driver.
It's been a while since publishing any fresh Intel Core i7 5775C benchmarks, the socketed Broadwell CPU with Iris Pro 6200 graphics, since normally it's busy in the daily benchmarking churn of the server room for Phoronix Test Suite / OpenBenchmarking.org / LinuxBenchmarking.com efforts. But with having been doing some maintenance on that system this week and loading a clean install of Ubuntu 16.10, I did some fresh benchmarks of the Iris Pro 6200 graphics using Mesa 13.1-dev and Linux 4.9, including a look at the OpenGL vs. Vulkan performance for the Iris Pro graphics.
If you are planning to upgrade your graphics card in a Linux system this holiday season, here are some fresh benchmarks of several different AMD Radeon and NVIDIA GeForce graphics cards with various Linux gaming tests under Ubuntu. The AMD tests were done both with the latest RadeonSI Gallium3D stack as well as the AMDGPU-PRO hybrid driver.
Anyone who read any number of my reviews from a few years ago, when I interned here at Phoronix, should know that I have been a fan of AMD and their open source efforts for a very long time. I remember the years of trying to get Catalyst to work under Arch or Fedora, usually only to have it blow up in my face. I remember the struggle holding back kernel and X server updates, hoping that none of those updates contained security fixes that were pertinent to me.
As some more exciting benchmarks to carry out this US holiday week, here are benchmarks of all major Mesa releases from Mesa 11.0 from mid 2015 through the latest Mesa 13.1-dev code as of this week. Additionally, the latest AMDGPU-PRO numbers are provided too for easy comparison of how the open-source AMD GCN 3D driver performance has evolved over the past year. It's a huge difference!
819 display drivers articles published on Phoronix.