1. 17 Apr, 2016 1 commit
  2. 23 Feb, 2015 3 commits
  3. 19 Dec, 2014 1 commit
  4. 13 Dec, 2014 1 commit
    • Dylan Baker's avatar
      generated_tests: Except error for os.makedirs · b59ff71e
      Dylan Baker authored
      Currently we check for the existence of a directory, and if it doesn't
      exist we create it. This would be fine, except that multiple generators
      can create the same directory trees. When os.makedirs tries to create a
      tree that already exists it raises and OSError exception, with the errno
      set to 17 (the File Exists Error). Since between the time that a check
      is made and the time the os.makedirs completes a second generator can
      create the directory tree we can have exceptions raised sporadically,
      especially as the number of threads being used to generate goes up.
      
      We maintain the check for existence before trying to create the
      directories because os.makedirs is slow, so not even trying is better
      than trying in vain.
      Signed-off-by: default avatarDylan Baker <dylanx.c.baker@intel.com>
      b59ff71e
  5. 24 Nov, 2014 5 commits
  6. 01 Jul, 2014 1 commit
  7. 22 May, 2014 1 commit
  8. 26 Aug, 2013 1 commit
  9. 29 Jul, 2013 1 commit
    • Dylan Baker's avatar
      gen_builtin_packing_tests.py PEP8 compliance · a448e409
      Dylan Baker authored
      Remaining PEP8 violations:
      247, 392: These are part of a mako template that is embedded in the
      		  python file. A better solution than wrapping them would be to
      		  split them out of the python and put them in their own mako
      		  files.
      
      v2: - Fix one of the issues left in v1
          - Revert to tabluar spacing in tables
          - Fix indent in two functions per Chad
          - Revert glsl_es_300 function to be formatted more like it was
            originally
      v3: - Revert some changes made as a result of false positives from an
            old version of the pep8 tool
      Signed-off-by: Dylan Baker's avatarDylan Baker <baker.dylan.c@gmail.com>
      Reviewed-by: default avatarChad Versace <chad.versace@linux.intel.com>
      a448e409
  10. 20 Feb, 2013 1 commit
    • Paul Berry's avatar
      ARB_shading_language_packing: fix exact tests · c4993595
      Paul Berry authored
      Commit 78435fef (ARB_shading_language_packing: Allow some imprecision
      in unpackUnorm4x8) inadvertently introduced a regression to the
      unpackHalf2x16 tests.  In those tests, the "exact" uniform wasn't
      being initialized, resulting in the tests being run in inexact
      fashion.
      
      Surprisingly, this actually produced a failure, because inexact
      testing uses "distance(actual, expect) < 0.00001" as the pass/fail
      criterion, and this doesn't work when actual == expect == Infinity
      (because distance(Inf, Inf) == NaN).
      
      We always intended to test unpackHalf2x16 in exact fashion anyhow, so
      the fix is to go ahead and initialize the "exact" uniform to 1 for
      this case.
      Reviewed-by: Matt Turner's avatarMatt Turner <mattst88@gmail.com>
      c4993595
  11. 05 Feb, 2013 2 commits
    • Matt Turner's avatar
      gen_builtin_packing_tests.py: Remove stray tab · 14da245f
      Matt Turner authored
      Caused the shader_runner tests generated from that template to be
      indented.
      Reviewed-by: Paul Berry's avatarPaul Berry <stereotype441@gmail.com>
      14da245f
    • Matt Turner's avatar
      ARB_shading_language_packing: Allow some imprecision in unpackUnorm4x8 · 78435fef
      Matt Turner authored
      The unpack builtins are implemented using floating-point division, which
      is implemented by reciprocal+multiply. As it so happens, 1.0/255.0 falls
      almost exactly between two representable 32-bit floats and leads to bad
      rounding. None of the other divisors (127, 32767, 65535) have this
      property.
      
      Paul was uncomfortable with allowing inexact results in only
      unpackUnorm4x8, so allow the results of the builtins (except
      unpackHalf2x16) to be within some small epsilon (0.00001) of the actual
      value. Some values like 0.0 and 1.0 should be calculated exactly, so
      handle them specially.
      78435fef
  12. 28 Jan, 2013 1 commit
    • Matt Turner's avatar
      ARB_shading_language_packing: Generate tests for builtin packing functions · 2f9badcf
      Matt Turner authored
      This uses the existing pack/unpack testing infrasturcture for GLSL ES
      3.0 and adds support for testing pack/unpack 4x8 operations.
      
      Generate the following test files:
      	{const,vs,fs}-{pack,unpack}{Snorm,Unorm}{2x16,4x8}.shader_test
      	{const,vs,fs}-{pack,unpack}Half2x16.shader_test
      
      Note that division is performed by multiplication by the reciprocal, and
      1/255 falls almost exactly in between two representable float32 values,
      leading to bad rounding errors. 1/127, 1/32767, and 1/65535 do not have
      this behavior.
      
      As a result, the {fs,vs}-unpackUnorm4x8 tests currently fail. the const-
      version passes, since the divisions are calculated by the CPU. A
      follow-on patch will allow some tolerance in precision.
      Reviewed-by: default avatarChad Versace <chad.versace@linux.intel.com>
      2f9badcf
  13. 24 Jan, 2013 1 commit