[VKCTS1.3.8] likely case issue: dEQP-VK.draw.renderpass.implicit_sample_shading.sample*

[NPUer-ZMT]

According to my test and the document ' Vulkan-cts-1.3.8/external/vulkancts/modules/vulkan/draw
/vktDrawSampleAttributeTests.cpp(line480~490) ', this case group mean to store value from bufferData to result, which records times of calling atom_add. If the result was checked to be expectedCounter(64), case would pass. But I found it was 16 and case failed. Nothing wrong with driver or compiler, maybe it's a case issue that doesn't make a 4-time-loop for recording calling times?

Thanks

[NPUer-ZMT]

I'm sorry to tell, mentioned bug came from my fault, not VKCTS1.3.8 case issue. This case group set the multisamping.sampleShaingEnable(as VK_FALSE) to test sample shading frature, which my HW doesn't support. Failed to set 4 sample per pixcel(when VK_SAMPLE_COUNT_4_BIT) during PS stage, finally causing result only got 1/4(16) of expectedCounter(64).

Thanks.

[NPUer-ZMT]

Dear Khronos devers:
We have supported Multi-sample Feature now, but this case group still met something wrong.
The multisamping.sampleShaingEnable was set as VK_FALSE inside cases, but wasn't turned on when VK_SAMPLE_COUNT_4_BIT was supported.
Can you help check it ?

Thanks.