[igt-dev] [PATCH i-g-t 1/4] lib/igt_matrix: Unroll and inline igt_matrix_transform()

[Chris Wilson]

Quoting Ville Syrjala (2018-06-06 21:16:39)
> From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> 
> Using the current igt_matrix for NV12 conversion ends up being
> about 4x as slow as the current non-igt_matrix based code. Unrolling
> and inlining igt_matrix_transform() improves that factor to ~1.5x.
> 
> Cc: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>
> Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
-Chris