https://github.com/yywyz/OpenCL-Programming-Examples
https://blog.csdn.net/zhouxuanyuye/category_7567322.html
https://blog.csdn.net/zhouxuanyuye/article/details/80037295
https://blog.csdn.net/zhouxuanyuye/article/details/80445076
A guide to convolution arithmetic for deep learning
https://arxiv.org/pdf/1603.07285v1.pdf
opencl:kernel中两种向量类型转换(convert_T,as_typen)的主要区别
https://blog.csdn.net/10km/article/details/51171911
https://registry.khronos.org/OpenCL/sdk/1.2/docs/man/xhtml/convert_T.html
https://registry.khronos.org/OpenCL/sdk/1.2/docs/man/xhtml/as_typen.html
http://man.opencl.org/dataTypes.html
https://stackoverflow.com/questions/28904114/opencl-sending-float3-host-to-device
https://community.khronos.org/t/typedef-cl-float4-cl-float3-cl-platform-h/2530
They shouldn’t and they won’t. cl_float3 is identical in size, alignment, and behavior to cl_float4.
It’s really just there to let you write self-documenting code. You can just as well use cl_float4 and forget that cl_float3 ever existed. It’s a convenience but the opencl types such as float and double are aligned to powers of 2 (cl_float2, cl_float4, cl_float8, and cl_float16 which are aligned to 8, 16, 32, and 64 bytes respectively).
Typedef is just saying “use this type but call it another name.” What’s being done with “typedef cl_float4 cl_float3” is that you’re being allowed to say your variable is a type of cl_float3 but to the machine, it still sees it as a cl_float4. No matter what you call it, the machine will still treat it the same.
https://www.codenong.com/17361274/
https://community.amd.com/t5/archives-discussions/cl-float3/td-p/333687
In a kernel it must be an error, not in the host program, as the spec only defines the kernel language definition. From the spec:
"float3 pos;
pos.z = 1.0f; // is legal
pos.w = 1.0f; // is illegal"
This only is illegal for a float3, not a cl_float3.
On a side note, the header file that defines cl_float3 is straight from Khronos, not something that we create. So if it is behavior you want changed, you need to go through Khronos to get it changed.
https://github.com/KhronosGroup/OpenCL-Headers/blob/main/CL/cl_platform.h
/* cl_float3 is identical in size, alignment and behavior to cl_float4. See section 6.1.5. */ typedef cl_float4 cl_float3;