Following last week's AMDGPU-PRO 16.20.3 "Beta 2" driver release of AMD's new hybrid driver stack for Linux that makes use of the AMDGPU open-source kernel DRM driver with the closed-source OpenGL driver derived from Catalyst / Radeon Software, I set out to do a fresh open vs. closed-source driver comparison. For the Radeon R9 285, R9 290, and R9 Fury, I compared the performance of this new AMDGPU-PRO driver against Mesa 11.3-devel Git and Linux 4.6 for the latest open-source driver stack.
This past week I showed how Intel Broadwell graphics are much faster with Mesa 11.3 but this new Mesa version doesn't do much for Haswell. Similar to Broadwell, Mesa 11.3 is a big win if you are on Intel's latest-generation Skylake hardware.
Not only is there now OpenGL 4.2 support for Intel's Mesa driver but also as part of the massive Mesa 11.3 release in development there are OpenGL performance improvements too, at least for newer Broadwell hardware.
Yesterday I published some Radeon DRM-Next performance tests for the Radeon DRM code queued up for the Linux 4.7 kernel merge window. Today I'm testing the DRM-Next-4.7 code for AMDGPU for looking at the performance improvements coming to this newer AMD DRM driver with the next Linux kernel cycle. The DRM-Next-4.7 performance was compared to Linux 4.6 in this article for a Tonga and Fiji graphics card.
With the AMDGPU and Radeon changes all settled down for what's proposed for Linux 4.7, this weekend I'm benchmarking both of these open-source AMD Linux Direct Rendering Manager drivers. First up is the Radeon DRM driver with various OpenGL benchmarks in conjunction with Mesa 11.3-devel.
Intel is brewing a makeover of their graphics driver stack through a large restructuring and consolidating initiative that will be formally announced in the coming weeks.
For some end-of-month benchmarks and while having a number of graphics cards out prior to being let down by Tomb Raider's Linux benchmark, here is a fresh round of OpenGL tests while using the newest Mesa 11.3-devel code on RadeonSI with AMDGPU/Radeon DRM from Ubuntu 16.04 and then compared to various Kepler/Maxwell graphics cards with the newest NVIDIA Linux driver.
Last week I published benchmarks showing Nouveau's "boost" patches offering much performance potential compared to the current state of the open-source NVIDIA Linux graphics driver but generally still not enough performance to compete with NVIDIA's proprietary Linux graphics driver. I've since carried out some fresh open-source AMD Linux results for reference to see how the NVIDIA vs. AMD GPU open-source speeds are comparing.
Karol Herbst has been one of the independent developers leading the charge to improve Nouveau re-clocking support. Within his Git tree he's been queuing up re-clocking and voltage handling improvements for this reverse-engineered NVIDIA Linux driver. He's hoping the improved re-clocking code will be ready for the Linux 4.7~4.8 kernel, but I decided to try out his Git tree this week for some benchmarking of this experimental support.
We've already looked at the GeForce 900 series performance on Nouveau with the Linux 4.6 kernel that finally has the necessary bits for supporting 3D hardware acceleration on these Maxwell GPUs. Those GTX 900 series results didn't end up being particularly exciting since there isn't yet any re-clocking support on this open-source NVIDIA Linux driver. For complementing those results are some GeForce 600/700 "Kepler" graphics card results when comparing Nouveau on Linux 4.6 with Mesa 11.3-devel and using re-clocking then compared to the latest NVIDIA proprietary driver.
For putting the open-source GeForce GTX 900 driver performance into better perspective from the results this week with the new support found on Linux 4.6 and compared to NVIDIA's proprietary driver, hare are some benchmark results when comparing the Nouveau driver performance of the GTX 900 "Maxwell" graphics cards to that of the GTX 600/700 "Kepler" graphics cards with their experimental re-clocking support.
For those curious whether the Linux 4.6 kernel is bringing any noticeable performance improvements to the AMDGPU and Radeon DRM drivers over Linux 4.5, here are some benchmarks in conjunction with using Mesa 11.3-devel built against LLVM 3.8.
With the in-development Linux 4.6 kernel there is the long-awaited NVIDIA GeForce GTX 900 series accelerated support atop the open-source Nouveau driver. While it requires using NVIDIA's signed binary blobs for the firmware, the support is now working. Here are some benchmarks on several different GTX 900 Maxwell graphics cards comparing the open-source driver performance to what's offered by NVIDIA's proprietary Linux driver.
Following this week's AMD vs. NVIDIA Linux OpenGL and Vulkan benchmarks, you may be wondering about the performance of OpenCL GPGPU performance particularly around AMD's new hybrid Linux driver stack. So for your viewing pleasure today are some OpenCL benchmarks on AMD Radeon and NVIDIA GeForce hardware using the newest drivers.
Thanks to AMD having released their new GPU-PRO "hybrid" Linux driver a few days ago, there is now Vulkan API support for Radeon GPU owners on Linux. This new AMD Linux driver holds much potential and the closed-source bits are now limited to user-space, among other benefits covered in dozens of Phoronix articles over recent months. With having this new driver in hand plus NVIDIA promoting their Vulkan support to the 364 Linux driver series, it's a great time for some benchmarking. Here are OpenGL and Vulkan atop Ubuntu 16.04 Linux for both AMD Radeon and NVIDIA GeForce graphics cards.
While still working on some AMD vs. NVIDIA Vulkan Linux driver benchmarks using AMD's new hybrid driver with Vulkan support, for your viewing pleasure this morning are some benchmarks comparing the new AMD GPU-PRO driver with its binary OpenGL driver against the pure open-source driver stack with the Ubuntu 16.04 AMDGPU driver and RadeonSI Gallium3D from Mesa 11.2 + LLVM 3.8.
On Friday night to much surprise, AMD published the beta version of their new hybrid Linux driver stack with Vulkan support alongside OpenCL, OpenGL, and VDPAU support. Here's some more details from my initial testing of this new driver that AMD is currently calling the Radeon Software AMD GPU-PRO Beta Driver for Linux.
This week we showed how the new AMDGPU driver stack is performing on Ubuntu 16.04 and that the recent generations of Radeon graphics cards are commonly seeing 80~90% the performance of Catalyst. However, it's important to keep in mind that aside from Catalyst being more buggy than the proprietary NVIDIA driver, the NVIDIA binary driver also tends to be more performant. So for putting the Ubuntu 16.04 open-source Radeon numbers into perspective, here are results putting them against the GeForce Kepler and Maxwell graphics cards.
Earlier this week was the How Ubuntu 16.04 Is Performing With AMDGPU/Radeon Graphics Compared To Ubuntu 14.04 With FGLRX, which showed off some interesting open-source Radeon Linux driver results but the Radeon R9 285 "Tonga" graphics card at the time couldn't be tested on Ubuntu 16.04's kernel due to a regression. That issue is fortunately now resolved in the latest Xenial Xerus kernel so here are those numbers.
With having the Core i7 5775C system out of the server racks when carrying out this week's six-way Linux distribution comparison, I carried out some additional tests of the high-performance Intel Iris Pro 6200 Graphics. This article is to serve as some fresh Iris Pro 6200 Graphics tests on Linux when upgrading to Linux 4.5, Mesa 11.3-devel Git, and switching to DRI3 rendering.
A number of Phoronix readers have been requesting some fresh AMD Kaveri Linux graphics driver benchmarks, so here you go. For your viewing pleasure today is an AMD open vs. closed-source driver comparison on Ubuntu 16.04 plus some extra runs featuring upgrades to the Linux kernel and Mesa as well as manually enabling DRI3 support.
While DRI3 appears to be in good shape with the latest X.Org Server series and Direct Rendering Infrastructure 3 is even mandated by the Intel Mesa Vulkan driver, DRI2 is still the default with the xf86-video-intel DDX driver, similar to the situation in the Radeon DDX driver as well.
Here are some fresh comparison benchmarks on Linux 4.5 and Mesa 11.2 when comparing the Radeon and Nouveau (NVIDIA) open-source Linux driver performance.
With the Linux 4.5 kernel bringing PCI Express link speed changes and other alterations to the open-source NVIDIA (Nouveau) driver, here are benchmarks comparing this unofficial open-source NVIDIA Linux driver on Linux 4.4 and Linux 4.5-rc3 compared to the proprietary NVIDIA Linux graphics driver.
With yesterday having started to run some fresh basic OpenCL benchmarks on the open-source Radeon driver given the interesting remarks by some super-computing researchers about having more hope for the open-source drivers than the proprietary Catalyst, here are some results comparing the latest open-source AMD Radeon Linux driver code to the proprietary driver.
While the AMDGPU Linux kernel driver right now exposes support for AMD GCN 1.2 GPUs and newer, like Carrizo, Fiji, and Tonga, it is possible to get GCN 1.1 Sea Islands hardware working with this driver if jumping through a few hoops. In this article are some tests of a Radeon R9 290 "Hawaii" when using the proprietary Catalyst driver, Radeon DRM driver as is the default for this card on open-source, and then using the experimental AMDGPU DRM open-source support.
With the first test release out this week for the Linux 4.5 kernel I have carried out some fresh benchmarks on different AMD Radeon graphics cards for comparing the very latest open-source driver performance against that of the proprietary AMD Linux driver. Here are how the competing AMD OpenGL Linux stacks are comparing to one another for starting off 2016.
Friday afternoon I had a call with a few AMD representatives talking about their Linux driver plans for 2016. Here's the details for those wondering about AMDGPU, Vulkan, GPUOpen, and more.
Last month I showed how AMD's open-source driver performance evolved in 2015 while today's article is looking at how the closed-source AMD / Radeon Technologies Group proprietary driver has evolved over the course of the year.
Similar to the recent open-source year end driver recaps, here's a recap of NVIDIA's binary driver activities for 2015 along with some benchmarks comparing the performance of the proprietary driver over the past year.
770 display drivers articles published on Phoronix.