dear amd staff,
i read opencl2.0 articles about shared virtual memory(svm) with an example using an apu.
and, i heard of using svm in opencl2.0 with a dedicated gpu will only simplify the programming, but not improve the time cost transferring data from main memory to gpu(or transferring back from), because it's just a software solution.
so, is there an amd dedicated gpu with a true hardware unified memory addressing i can purchase?
and can i say the a10 apu solution which improve the time cost efficiency as a hardware unified memory?
a decade customer