COMPMID-420, COMPMID-414 - Port CLConvolutionLayer and CLFullyConnectedLayer to use 8 bit fixed point

Change-Id: I1cb1b4d7711ad7b569ee691e13a5df1b3430292b
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79565
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/src/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.cpp b/src/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.cpp
index 75c1a6e..a7ca6f2 100644
--- a/src/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.cpp
+++ b/src/core/CL/kernels/CLGEMMMatrixAccumulateBiasesKernel.cpp
@@ -53,6 +53,10 @@
 
     std::set<std::string> build_opts;
     build_opts.insert(("-DDATA_TYPE=" + get_cl_type_from_data_type(accum->info()->data_type())));
+    if(accum->info()->data_type() == DataType::QS8)
+    {
+        build_opts.emplace("-DFIXED_POINT_POSITION=" + support::cpp11::to_string(accum->info()->fixed_point_position()));
+    }
 
     // Create kernel
     _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("gemm_accumulate_biases", build_opts));