Archive of May 2016

Mesa 11.3/12.0 RC Is Coming Closer

The Release Candidate for Mesa 11.3/12.0 is coming closer. There's plenty of fixes, new features and performance improvements for this new release.

Most interesting additions/fixes:

There's also a huge pile of i965 fixes and additions, bringing Intel's driver to OpenGL 4.2 compliance.

Stay tuned for more informations when the first RC gets released!

Mesa Updated To 11.2.2

I've uploaded Mesa 11.2.2 to, enjoy!


Boyuan Zhang (1):
  radeon/uvd: alignment fix for decode message buffer<br/>

<br/> Brian Paul (2):<br/>

  st/mesa: fix sampler view leak in st_DrawAtlasBitmaps()<br/>
  gallium/util: initialize pipe_framebuffer_state to zeros<br/>

<br/> Chad Versace (1):<br/>

  dri: Fix robust context creation via EGL attribute<br/>

<br/> Egbert Eich (1):<br/>

  dri2: Check for dummyContext to see if the glx_context is valid<br/>

<br/> Emil Velikov (6):<br/>

  docs: add sha256 checksums for 11.2.1<br/>
  docs: update the sha256 checksums for 11.2.1<br/>
  cherry-ignore: remove duplicate commit<br/>
  cherry-ignore: ignore the GetSamplerParameterIuiv{EXT,OES} fixups<br/>
  Update version to 11.2.2<br/>
  docs: add release notes for 11.2.2<br/>

<br/> Eric Anholt (4):<br/>

  vc4: Fix subimage accesses to LT textures.<br/>
  vc4: Add support for rendering to cube map surfaces.<br/>
  vc4: Fix tests for format supported with nr_samples == 1.<br/>
  vc4: Make sure we recompile when sample_mask changes.<br/>

<br/> Frederic Devernay (1):<br/>

  glapi: fix _glapi_get_proc_address() for mangled function names<br/>

<br/> Ilia Mirkin (2):<br/>

  nvc0: fix retrieving query results into buffer for timestamps<br/>
  nouveau/video: properly detect the decoder class for availability checks<br/>

<br/> Jason Ekstrand (1):<br/>

  i965/fs: Properly report regs_written from SAMPLEINFO<br/>

<br/> Jonathan Gray (1):<br/>

  egl/x11: authenticate before doing chipset id ioctls<br/>

<br/> Jose Fonseca (1):<br/>

  winsys/sw/xlib: use correct free function for xlib_dt->data<br/>

<br/> Kenneth Graunke (3):<br/>

  i965: Fix clear code for ignoring colormask for XRGB formats on Gen9+.<br/>
  glsl: Convert lower_vec_index_to_swizzle to a rvalue visitor.<br/>
  glsl: Lower vector_extracts to swizzles after lower_vector_derefs.<br/>

<br/> Leo Liu (1):<br/>

  radeon/uvd: fix tonga feedback buffer size<br/>

<br/> Marek Olšák (1):<br/>

  st/mesa: fix blit-based GetTexImage for non-finalized textures<br/>

<br/> Nicolai Hähnle (5):<br/>

  gallium/radeon: handle failure when mapping staging buffer<br/>
  st/glsl_to_tgsi: reduce stack explosion in recursive expression visitor<br/>
  gallium/radeon: fix crash in r600_set_streamout_targets<br/>
  radeonsi: correct NULL-pointer check in si_upload_const_buffer<br/>
  radeonsi: work around an MSAA fast stencil clear problem<br/>

<br/> Oded Gabbay (4):<br/>

  r600g/radeonsi: send endian info to format translation functions<br/>
  r600g: set endianess of 16/32-bit buffers according to do_endian_swap<br/>
  r600g: use do_endian_swap in color swapping functions<br/>
  r600g: use do_endian_swap in texture swapping function<br/>

<br/> Patrick Rudolph (1):<br/>

  r600g: fix and optimize tgsi_cmp when using ABS and NEG modifier<br/>

<br/> Roland Scheidegger (3):<br/>

  llvmpipe: (trivial) initialize src1_alpha var to NULL<br/>
  gallivm: fix bogus argument order to lp_build_sample_mipmap function<br/>
  gallivm: make sampling more robust against bogus coordinates<br/>

<br/> Samuel Pitoiset (6):<br/>

  gk110/ir: do not overwrite def value with zero for EXCH ops<br/>
  gk110/ir: make use of IMUL32I for all immediates<br/>
  nvc0/ir: fix wrong emission of (a OP b) OP c<br/>
  gk110/ir: add emission for (a OP b) OP c<br/>
  nvc0: reduce GL_MAX_3D_TEXTURE_SIZE to 2048 on Kepler+<br/>
  st/glsl_to_tgsi: fix potential crash when allocating temporaries<br/>

<br/> Stefan Dirsch (1):<br/>

  dri3: Check for dummyContext to see if the glx_context is valid<br/>

<br/> Topi Pohjolainen (2):<br/>

  i965/blorp/gen7: Prepare re-using for gen8<br/>
  i965/blorp: Use 8k chunk size for urb allocation<br/>

<br/> WuZhen (3):<br/>

  tgsi: initialize stack allocated struct<br/>
  winsys/sw/dri: use correct free function for dri_sw_dt->data<br/>
  android: enable dlopen() on all architectures</blockquote></p>

Optimized Tessellation Comes To Mesa

Bas Nieuwenhuizen has published a set of patches which should be able to deliver dramatically better tessellation performance with Mesa on RadeonSI cards.

This patchset implements offchip tessellation after which we can finally process more than one patch per wave without decreasing tessmark scores.

For tessmark this improves performance by ~20% for the x32 case and ~80% for the x64 case. x8 and x16 have roughly the same performance as before. Unigine heaven gets 43 fps compared to 28 before (roughly +50%). Amdgpu-pro gets 44 fps for heaven. For Shadow of Mordor the performance changes from 28 fps to 40 fps (roughly +40%).

Note: These values have been archived with a Tonga GPU.

Convenient Apt-Pinning Package For

I've built a package to provide a convenient way to protect you from accidentally dist-upgrading my Gallium-Nine drivers away and replacing them with debian's stock Mesa. Just install with

# apt-get install apt-pin-mesa
and keep dist-upgrading.

Y U Do Dis, Feral?

Liam Dawe has just posted some shocking benchmark results for the newly ported game "Tomb Raider (2013)". There seems to be a huge performance gap!

He also uploaded a → side-by-side comparsion so you can see it for yourself.

(→ Source)