Nouveau Gallium3D Now Supports Double-Precision Floating-Point Data Types
The GL_ARB_gpu_shader_fp64 extension is about allowing GLSL shaders to use double-precision floating-point data types. Doubles with this extension can be used for inputs, outputs, and uniforms. For those wishing to learn more about this GLSL doubles support, see the OpenGL.org specification.
Up until yesterday, only the Gallium3D Softpipe driver has supported GL_ARB_gpu_shader_fp64. The Nouveau doubles support is for the NVC0 Gallium3D driver that supports NVIDIA's GeForce GTX 400 "Fermi" series and newer -- including Kepler and the in-progress Maxwell support.
The commit enabling the support following the prep work was done by prolific Nouveau/Freedreno Gallium3D contributor Ilia Mirkin.