volk VS ncnn

Im learning Vulkan in attempt to create a renderer for a video game I plan on making. Following the spec https://registry.khronos.org/vulkan/specs/1.0/html/vkspec.html, I've reached the point to where I can create and destroy the VkInstance. I'm using GLFW for the windowing and was originally using for the Vulkan stuff, but turned to Volk https://github.com/zeux/volk to load Vulkan in the end. Reflecting that change, I replace the vulkan.h includes with volk.h, defined the platform macro for my platform, and unlinked Vulkan in my Makefile.

Do you link libVulkan-1.so statically in your project? On Android, only Vulkan 1.0 functions are exposed via this lib. Performance wise and to use any Vulkan functions above 1.0 you need to load all your Vulkan functions by yourself or (even better) use a loader library like volk

no problem! This library might help you too https://github.com/zeux/volk

You also should not link to Vulkan directly (-lvulkan) but retrieve the function pointers to the Vulkan functions by yourself or using a loder library like volk.

You can't statically link vulkan, you need to load vulkan-1.dll from the driver installation. You can use something like volk for this https://github.com/zeux/volk

Use a Vulkan loader like Volk. https://github.com/zeux/volk

For developing with Vulkan all you need are the Vulkan Headers and some kind of static loader to link against or you use a dynamic loader like volk.

I am using VMA as well in my renderer and I am not having any issues with it. As I am using Volk (https://github.com/zeux/volk) to locate and load all Vulkan function pointers. Thats why I need to tell VMA to not load any functions statically:

ncnn uses Vulkan for GPU acceleration, I've seen it used in a few projects to get AMD hardware support.

https://github.com/Tencent/ncnn

They're using NCNN to package the model. Have a look. https://github.com/Tencent/NCNN

Hi. Here is my prefered android app for realtime objet detection: https://github.com/nihui/ncnn-android-nanodet ; https://github.com/Tencent/ncnn contains a lot of android demo app for a lot of models.

ncnn home page: https://github.com/Tencent/ncnn

ncnn

Yes, Android devices tend to have bigger RAMs, making running 1024x1024 possible (this is not possible at all on iPhones, which could peak around 5GiB memory with my current implementation, some serious engineering required to bring that down on iPhone devices). The problem is I am not sure about speed. I would likely switch to NCNN (https://github.com/Tencent/ncnn) as the backend which have a decent Vulkan computing kernel support. It is definitely a possibility and there is a path to do that.

As the leading open-source inference framework in China and in the world, what we like are its almost zero cost cross-platform capability, high inference speed, and minimal deployment volume. (Project address: https://github.com/Tencent/ncnn)

If you’re not planning on training the neural net on an embedded device and just do inference, this might interest you: https://github.com/Tencent/ncnn