GBE: fix a bug in int64 to float conversion.
authorZhigang Gong <zhigang.gong@intel.com>
Fri, 6 Jun 2014 06:26:48 +0000 (14:26 +0800)
committerZhigang Gong <zhigang.gong@intel.com>
Mon, 9 Jun 2014 07:14:16 +0000 (15:14 +0800)
commit2c4160bf81540a693bc00fcb69c1cb1ef993f282
tree9f28936cfbd25ca03c962fb6e247a1346b9c9cce
parent681aa2bc2d5df275b1e1dea6406769333868a917
GBE: fix a bug in int64 to float conversion.

When copy those pure 32bit int to float destination, we
should enable the mask. Otherwise, we may destroy the
value in inactive lanes.

Signed-off-by: Zhigang Gong <zhigang.gong@intel.com>
Reviewed-by: Yang Rong <rong.r.yang@intel.com>
backend/src/backend/gen_context.cpp