swr/rast: Adjusted avx512 primitive assembly for msvc codegen
authorAlok Hota <alok.hota@intel.com>
Fri, 25 May 2018 15:19:49 +0000 (10:19 -0500)
committerTim Rowley <timothy.o.rowley@intel.com>
Fri, 25 May 2018 15:57:02 +0000 (10:57 -0500)
commitfb20ae0374425ae3aff2a50a498c7e2b428632a4
tree0847afd4742b067085cf8fbb48dcc09f5e302ce0
parentb3360f5c8b74906187a8801d83f2c4f73f3c025e
swr/rast: Adjusted avx512 primitive assembly for msvc codegen

Optimize AVX-512 PA Assemble (PA_STATE_OPT). Reduced generated code by
about 4x, MSVC compiler was going crazy making temporaries and
split-loading inputs onto the stack unless explicit AVX-512 load ops
were added

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
src/gallium/drivers/swr/rasterizer/core/pa_avx.cpp