jgbit / vuda

VUDA is a header-only library based on Vulkan that provides a CUDA Runtime API interface for writing GPU-accelerated applications.
MIT License
865 stars 35 forks source link

Activating assert (the size of the buffer should always be a multiple of the atom size) #18

Closed ant0nT closed 4 years ago

ant0nT commented 4 years ago

Hi, there! I've tried to run vuda usage example but I've gotten SIGABRT in ./inc/state/logicaldevice.inl file (row #667). hardware: Intel UHD Graphics 630 and Radeon Pro 560X software: macOS Mojave, Vulkan SDK 1.2.141 with moltenVK count_noncoherentatomsize = 2048 dst_ptr->m_size = 2000 without assert it executes successfully.

jgbit commented 4 years ago

Hello, thanks for trying out VUDA. I have revisited the documentation for VkMappedMemoryRange -- the asserts for InvalidateMappedMemoryRanges and FlushMappedMemoryRanges should now be correct. Let me know. Thank you.

ant0nT commented 4 years ago

Hello again! I've updated VUDA and now usage example is working fine. Thank you very much!