Half-float texture interpolation tolerance is too small.

[platform/upstream/VK-GL-CTS.git] / external / vulkancts / modules / vulkan / texture / vktTextureFilteringExplicitLodTests.cpp

diff --git a/external/vulkancts/modules/vulkan/texture/vktTextureFilteringExplicitLodTests.cpp b/external/vulkancts/modules/vulkan/texture/vktTextureFilteringExplicitLodTests.cpp
index cf35765..979103e 100644 (file)
--- a/external/vulkancts/modules/vulkan/texture/vktTextureFilteringExplicitLodTests.cpp
+++ b/external/vulkancts/modules/vulkan/texture/vktTextureFilteringExplicitLodTests.cpp
@@ -614,7 +614,7 @@ TestStatus TextureFilteringTestInstance::verify (void)
        int                                                                                             warningCount            = 0;
        const tcu::TextureFormat                                                tcuFormat                       = mapVkFormat(m_imParams.format);
        std::vector<de::SharedPtr<tcu::FloatFormat>>    strictPrecision         = getPrecision(m_imParams.format, 0);
-       std::vector<de::SharedPtr<tcu::FloatFormat>>    relaxedPrecision        = tcuFormat.type == tcu::TextureFormat::HALF_FLOAT ? getPrecision(m_imParams.format, -3) : getPrecision(m_imParams.format, -2);
+       std::vector<de::SharedPtr<tcu::FloatFormat>>    relaxedPrecision        = tcuFormat.type == tcu::TextureFormat::HALF_FLOAT ? getPrecision(m_imParams.format, -6) : getPrecision(m_imParams.format, -2);
        const bool                                                                              allowRelaxedPrecision   = (tcuFormat.type == tcu::TextureFormat::HALF_FLOAT || tcuFormat.type == tcu::TextureFormat::SNORM_INT8) &&
                (m_samplerParams.minFilter == VK_FILTER_LINEAR || m_samplerParams.magFilter == VK_FILTER_LINEAR);