Fix 16-bit storage IO f32-to-f16 conversion tests
authorAlexander Galazin <alexander.galazin@arm.com>
Wed, 17 Oct 2018 07:46:34 +0000 (09:46 +0200)
committerAlexander Galazin <Alexander.Galazin@arm.com>
Thu, 18 Oct 2018 07:20:39 +0000 (03:20 -0400)
commitf0f82066bc94bb6082deb7734b18a9ed18908df5
tree484bb8ae63a85da01c670a13f978c13dfd3bf0df
parent13f7cae9596e7e321432f3f384ab48a0ca74f28d
Fix 16-bit storage IO f32-to-f16 conversion tests

The tests were applying a rounding mode decoration
to OpFConvert that was used by OpReturnValue.
The extension doesn't allow this. The only allowed
successor for suc instruction is OpStore to specific
storage classes.

Components: Vulkan

VK-GL-CTS issue: 1424

Affects:
dEQP-VK.spirv_assembly.instruction.graphics.16bit_storage.input_output_float_32_to_16.*

Change-Id: Ica7b0c1da79225500124fc596859a7a04f086d52
external/vulkancts/modules/vulkan/spirv_assembly/vktSpvAsm16bitStorageTests.cpp
external/vulkancts/modules/vulkan/spirv_assembly/vktSpvAsmGraphicsShaderTestUtil.cpp