For those living on the Mesa 10.2 stable series rather than the experimental Mesa 10.3 code, there's a new point release out today.
Carl Worth of Intel released Mesa 10.2.6 as the latest bug-fix update. Mesa 10.2.6 has at least 28 bugs fixed, including many affecting core Mesa, some AMD RadeonSI fixes (affecting Hawall and Tahiti hardware), and various other fixes. Anuj Phogat contributed the most fixes at 15 followed by Marek Olšák at 4.
While Mesa is still racing towards OpenGL 4.0 compliance, another OpenGL 4.5 extension can now be crossed off the Mesa TODO list.
Some Mesa developers have already started tackling some of the easier OpenGL 4.5 extensions and today another can be crossed off the list. Thanks to Tobias Klausmann. GL_ARB_conditional_render_inverted is now supported by Mesa. The core work for GL_ARB_conditional_render_inverted is complete and is implemented currently by the Gallium3D-based Nouveau NVC0 (Fermi+), Softpipe, and LLVMpipe drivers. Support will surely come in time for mainline Mesa with this extension for the RadeonSI Gallium3D and Intel drivers.
Nouveau developer Martin Peres has published a set of ten Nouveau DRM patches working towards proper fan/power management support for NVIDIA's latest "Maxwell" GPUs.
The patches by the long-time Nouveau contributor from France adds Maxwell support for fan management, changes to fan handling, PWM fan support (Pulse Width Modulation) support, and other power/fan related fixes and improvements.
As a continuation to yesterday's brief GCC 4.9 vs. GCC 4.10 (GCC 5.0) comparison with the AMD A10 A-Series "Kaveri" APU, here's some benchmarks when using the GCC 4.10 development snapshot and trying a variety of CFLAGS/CXXFLAGS to see the current impact on their performance for a variety of Linux benchmarks.
GCC 4.10 has been under development since the 4.9.0 release near the beginning of the year. However, at the GNU Tools Cauldron it was agreed upon that GCC 4.10 will most likely become GCC 5.0 upon its release in 2015. The GCC version scheme is also being shaken up for future releases. Years ago there was talk of GCC 5.0 being modular and more like LLVM but to date there's no "killer features" of GCC 5.0 at this point in its SVN code-base.
Back on 8 August there was evidently the Catalyst 14.8 Linux driver replace to succeed the Catalyst 14.6 Beta that was last updated in mid-July. AMD didn't make any announcement about the 14.8 Linux driver update and we didn't even notice until now when a Phoronix reader stumbled across Catalyst 14.8.
Unfortunately AMD didn't publish a change-log for the Catalyst 14.8 Linux driver update so we're really not sure what (if any) significant changes made it into this latest release, but we would certainly assume there's more Linux game bug-fixes to be found in this newest version. Given it was released at the end of last week, there's not OpenGL 4.5 support expected. It's also likely too soon to expect any Linux 3.16 kernel compatibility.
With the recently released AMD A10-7800 Kaveri APU I carried out some new benchmarks comparing the open and closed-source Linux GPU driver performance for AMD with their Catalyst and RadeonSI Gallium3D solutions. When running the open-source Ubuntu driver tests, multiple versions of Mesa and the Linux kernel were used.
While the OpenGL 4.5 specification is fresh off the press and we haven't even seen the Khronos SIGGRAPH announcement yet, NVIDIA has already made public their OpenGL 4.5 beta drivers for Linux and Windows.
The NVIDIA 340.23.01 Linux driver is available today and provides beta support for OpenGL 4.5 and the OpenGL Shading Language (GLSL) 4.50 versions. For tapping all of the potential of OpenGL 4.5, a Fermi, Kepler, or Maxwell GPU is needed. This is conveniently the GeForce 400 series and newer, which is only what's supported now after NVIDIA dropped pre-Fermi support from their mainline Linux driver.