I am looking to purchase a new laptop and have been waiting for the Richland series mobile APUs to hit the street before I buy. Specifically, I am looking to do some OpenCL and C++ AMP development and the problems I have require double precision floating point calculations. While the performance doesn't have to be amazing, I do need it to be supported by the GPU. Does anyone know if the A10-5750M or any other richland or trinity based mobile APU support double precision floating point math?
If there is no APU do any of the mobile AMD graphics solution support fp64?
Solved! Go to Solution.
I did some more digging online and came across these benchmarks and OpenCL environment variables. This is the link to the GPU on the A10-4600M with an integrated HD 7660G GPU. http://clbenchmark.com/device-environment.jsp?config=12100910 it can be seen here that it supports cl_amd_fp64 and cl_khr_fp64 OpenCL environmental variables. I don't know what the performance would be, probably low, but at least it will perform the operations.
The older Llano based APUs did not support fp64, but Trinity and I suspect Richland will. Based on this, I will be purchasing a Richland based laptop as soon as I can find one.
I did some more digging online and came across these benchmarks and OpenCL environment variables. This is the link to the GPU on the A10-4600M with an integrated HD 7660G GPU. http://clbenchmark.com/device-environment.jsp?config=12100910 it can be seen here that it supports cl_amd_fp64 and cl_khr_fp64 OpenCL environmental variables. I don't know what the performance would be, probably low, but at least it will perform the operations.
The older Llano based APUs did not support fp64, but Trinity and I suspect Richland will. Based on this, I will be purchasing a Richland based laptop as soon as I can find one.