Update GEMM assembly kernels

- Introduce Fp32 kernels with internal calculations in Bfloat16 when
fast_mode is enabled
- Improve kernel selection heuristics

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I68a9e7e862b6fd2721b46e0d7cc791091c4ab279
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5965
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/utils.hpp b/src/core/NEON/kernels/arm_gemm/utils.hpp
index e648ce2..4ba03da 100644
--- a/src/core/NEON/kernels/arm_gemm/utils.hpp
+++ b/src/core/NEON/kernels/arm_gemm/utils.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -37,6 +37,29 @@
 namespace arm_gemm {
 
 template<typename T>
+std::string get_type_name() {
+#ifdef __GNUC__
+    std::string s = __PRETTY_FUNCTION__;
+
+    auto start = s.find("cls_");
+
+    if (start==std::string::npos) {
+        return "(unknown)";
+    }
+
+    for(size_t x = start+4; x<s.size(); x++) {
+        if (s[x] == ';' || s[x] == ']') {
+            return s.substr(start+4, x-(start+4));
+        }
+    }
+
+    return "(unknown)";
+#else
+    return "(unsupported)";
+#endif
+}
+
+template<typename T>
 inline T iceildiv(const T a, const T b) {
     return (a + b - 1) / b;
 }