COMPMID-1021: CPUInfo refactoring.

Removed CPUTarget in favor of the CPUModel type.
CPUInfo now holds a vector of N CPUs.
CPUInfo autoinitialise upon construction with 1 GENERIC CPU.
CPPScheduler fills CPUInfo's vector upon construction (runtime).
IScheduler has a single CPUInfo obj and ThreadInfo always gets a pointer to it (avoid copying the vector)

Change-Id: I30f293258c959c87f6bac5eac8b963beb6a4d365
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/124626
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>

4 files changed, 12 insertions, 21 deletions

diff --git a/src/runtime/NEON/functions/NEGEMM.cpp b/src/runtime/NEON/functions/NEGEMM.cpp
index c8cba8a174..18e6e919c3 100644
--- a/src/runtime/NEON/functions/NEGEMM.cpp
+++ b/src/runtime/NEON/functions/NEGEMM.cpp
@@ -65,7 +65,8 @@ void NEGEMM::configure(const ITensor *a, const ITensor *b, const ITensor *c, ITe
     // Check if we need to reshape the matrix B only on the first run
     _reshape_b_only_on_first_run      = gemm_info.reshape_b_only_on_first_run();
     _run_vector_matrix_multiplication = a->info()->dimension(1) < 2;
-    const bool run_optimised          = setup_assembly_kernel(a, b, c, d, alpha, beta, _workspace, _memory_group, _asm_glue);
+
+    const bool run_optimised = a->info()->data_type() == DataType::F32 && (c == nullptr || beta == 0.f) && setup_assembly_kernel(a, b, d, alpha, beta, _workspace, _memory_group, _asm_glue);
 
     // Check if the first input tensor is a vector.
     // If so, all the kernels for reshaping the tensors can be skipped
diff --git a/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp b/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp
index b2dd0227a5..cdbd32373a 100644
--- a/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp
+++ b/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp
@@ -271,8 +271,7 @@ void NEGEMMConvolutionLayer::configure(const ITensor *input, const ITensor *weig
     const unsigned int fixed_point_position = input->info()->fixed_point_position();
     const ITensor     *biases_to_use        = (_append_bias) ? biases : nullptr;
 
-    bool run_optimised =
-        (NEScheduler::get().cpu_info().CPU == CPUTarget::ARMV7 && dt == DataType::F32) || (NEScheduler::get().cpu_info().CPU >= CPUTarget::ARMV8 && dt == DataType::F32);
+    bool run_optimised = dt == DataType::F32;
 
     // Reshape weights if needed
     if(run_optimised)
@@ -369,8 +368,10 @@ void NEGEMMConvolutionLayer::configure(const ITensor *input, const ITensor *weig
     // Configure matrix multiply
     if(run_optimised)
     {
-        run_optimised = setup_assembly_kernel(&_input_im2col_reshaped, weights, nullptr, &_gemm_output, 1.f, 0.f, _workspace, _memory_group, _asm_glue);
-        ARM_COMPUTE_ERROR_ON_MSG(run_optimised == false, "setup_assembly_kernel failed.");
+        if(!setup_assembly_kernel(&_input_im2col_reshaped, weights, &_gemm_output, 1.f, 0.f, _workspace, _memory_group, _asm_glue))
+        {
+            ARM_COMPUTE_ERROR("setup_assembly_kernel failed.");
+        }
     }
     else
     {
@@ -450,17 +451,10 @@ Status NEGEMMConvolutionLayer::validate(const ITensorInfo *input, const ITensorI
     std::unique_ptr<ITensorInfo> reshaped_weights = weights->clone();
     bool                         optimised_kernel = false;
 
-#if defined(__arm__)
-    if(NEScheduler::get().cpu_info().CPU == CPUTarget::ARMV7 && dt == DataType::F32)
-    {
-        optimised_kernel = true;
-    }
-#elif defined(__aarch64__)
-    if(NEScheduler::get().cpu_info().CPU >= CPUTarget::ARMV8 && dt == DataType::F32)
+    if(dt == DataType::F32)
     {
         optimised_kernel = true;
     }
-#endif /* defined(__arm__) || defined(__aarch64__) */
 
     // Reshape weights if needed
     if(optimised_kernel)
diff --git a/src/runtime/NEON/functions/NEGEMMLowpAssemblyMatrixMultiplyCore.cpp b/src/runtime/NEON/functions/NEGEMMLowpAssemblyMatrixMultiplyCore.cpp
index a02872c3d5..27dd6c51d7 100644
--- a/src/runtime/NEON/functions/NEGEMMLowpAssemblyMatrixMultiplyCore.cpp
+++ b/src/runtime/NEON/functions/NEGEMMLowpAssemblyMatrixMultiplyCore.cpp
@@ -58,13 +58,13 @@ void NEGEMMLowpAssemblyMatrixMultiplyCore::configure(const ITensor *a, const ITe
     {
         case DataType::S8:
         {
-            run_optimised = setup_assembly_kernel(a, b, nullptr, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_signed);
+            run_optimised = setup_assembly_kernel(a, b, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_signed);
             break;
         }
         case DataType::QASYMM8:
         case DataType::U8:
         {
-            run_optimised = setup_assembly_kernel(a, b, nullptr, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_unsigned);
+            run_optimised = setup_assembly_kernel(a, b, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_unsigned);
             break;
         }
         default:
diff --git a/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp b/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp
index b0a7bdd7bc..7372c6ca57 100644
--- a/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp
+++ b/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp
@@ -62,13 +62,13 @@ void NEGEMMLowpMatrixMultiplyCore::configure(const ITensor *a, const ITensor *b,
     {
         case DataType::S8:
         {
-            _dot_product_path = setup_assembly_kernel(a, b, nullptr, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_signed);
+            _dot_product_path = setup_assembly_kernel(a, b, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_signed);
             break;
         }
         case DataType::QASYMM8:
         case DataType::U8:
         {
-            _dot_product_path = setup_assembly_kernel(a, b, nullptr, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_unsigned);
+            _dot_product_path = setup_assembly_kernel(a, b, output, 1.f, 1.f, _workspace, _memory_group, _asm_glue_unsigned);
             break;
         }
         default:
@@ -156,10 +156,6 @@ void NEGEMMLowpMatrixMultiplyCore::configure(const ITensor *a, const ITensor *b,
         _tmp_a.allocator()->allocate();
         _tmp_b.allocator()->allocate();
     }
-    else
-    {
-        _workspace.allocator()->allocate();
-    }
 
     if(_a_offset != 0)
     {