hi, I was using opencl adreno gpu sdk to implement math algorithm, where I called API
float4 qcom_convolve_imagef(image2d_t image, sampler_t sampler, float2 coord, qcom_weight_image_t weight_image);In the examples, the API seems only work on image data type uint8. I want to apply this API on image data type float. I tried this API on float type, the convolution result is wrong. I wonder if there is restriction on the image data type of API, and if yes, if there are any other API that qualcomm provide can support image type float? thanks in advance.