copy_and_blit.blit_image: fix threshold computation
authorLionel Landwerlin <lionel.g.landwerlin@intel.com>
Wed, 12 Oct 2016 18:03:31 +0000 (19:03 +0100)
committerPyry Haulos <phaulos@google.com>
Fri, 14 Oct 2016 20:40:41 +0000 (16:40 -0400)
commit161314f0ff255dfe47e280dd4945292295dab1ed
treed8a3cbf1ca1cc939b52f291f6b4978e9c40e8c58
parentef115082bbb62e6c8d71ed2ec76f99cb47725b74
copy_and_blit.blit_image: fix threshold computation

Threshold should not be ceiled, affecting all blit tests with
floating point format as destination :

dEQP-VK.api.copy_and_blit.blit_image.all_formats.*sfloat*

Also use the source texture to get the source format.

Fixes #516

Change-Id: I4c49007070306d597e36f7b79ee969ef5081d19e
external/vulkancts/modules/vulkan/api/vktApiCopiesAndBlittingTests.cpp