Closed billhollings closed 1 year ago
I think having atomics on workgroup memory is essential. Please reconsider putting an effort into fixing this instead of flagging off.
Based on that error message, it looks like the problem isn't workgroup variables per se as opposed to the fact that MoltenVK isn't passing down the buffer backing an image being accessed atomically. The names of those tests corroborate this.
Based on that error message, it looks like the problem isn't workgroup variables per se as opposed to the fact that MoltenVK isn't passing down the buffer backing an image being accessed atomically. The names of those tests corroborate this.
Ah. Good observation. I guess I didn't delve into the error message much before identifying this as an issue.
Should be fixable in MoltenVK then? That's good.
This has been fixed in MoltenVK.
Some CTS tests fail when using atomic operations on workgroup local variables.
This might be fixable with effort within the implementations.
In the meantime, these use cases can be disabled in CTS and reported as Not Supported, by adding a
VkPhysicalDevicePortabilitySubsetFeaturesKHR:: shaderAtomicOperationLocal
capability.This affects 2 CTS tests: