COMPMID-662: Integrated the new a64_s8_gemm_12x8 + dot product kernel into ACL.

Change-Id: Id8f919e486a132fc58346c9f84fccbeeb83d19b3
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/94233
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
diff --git a/arm_compute/core/NEON/kernels/assembly/profiler.hpp b/arm_compute/core/NEON/kernels/assembly/profiler.hpp
index d2f8ba9..f7a1d1c 100644
--- a/arm_compute/core/NEON/kernels/assembly/profiler.hpp
+++ b/arm_compute/core/NEON/kernels/assembly/profiler.hpp
@@ -31,6 +31,7 @@
 private:
     static const int maxevents = 10000;
     unsigned long times[maxevents];
+    unsigned long units[maxevents];
     int events[maxevents];
     int currentevent;
     int countfd;
@@ -45,35 +46,38 @@
         close(countfd);
         int tots[5];
         unsigned long counts[5];
+        unsigned long tunits[5];
         const char * descs[] = { "Prepare A", "Prepare B", "Kernel", "Merge" };
 
         for (int i=1; i<5; i++) {
             tots[i] = 0;
             counts[i] = 0;
+            tunits[i] = 0;
         }
 
         printf("Profiled events:\n");
         for (int i=0; i<currentevent; i++) {
-            printf("%10s: %ld\n", descs[events[i]-1], times[i]);
             tots[events[i]]++;
             counts[events[i]] += times[i];
+            tunits[events[i]] += units[i];
         }
 
-        printf("%20s  %9s %9s %9s\n", "", "Events", "Total", "Average");
+        printf("%20s  %9s %9s %9s %12s %9s\n", "", "Events", "Total", "Average", "Bytes/MACs", "Per cycle");
         for (int i=1; i<5; i++) {
-            printf("%20s: %9d %9ld %9ld\n",descs[i-1],tots[i],counts[i],counts[i]/tots[i]);
+            printf("%20s: %9d %9ld %9ld %12lu %9.2f\n",descs[i-1],tots[i],counts[i],counts[i]/tots[i],tunits[i],(float)tunits[i]/counts[i]);
         }
     }
 
     template <typename T>
-    void operator() (int i, T func) {
+    void operator() (int i, unsigned long u, T func) {
         if (currentevent==maxevents) {
             func();
         } else {
+            events[currentevent] = i;
+            units[currentevent] = u;
             start_counter(countfd);
             func();
             long long cycs = stop_counter(countfd);
-            events[currentevent] = i;
             times[currentevent++] = cycs;
         }
     }
@@ -84,7 +88,7 @@
 class profiler {
 public:
     template <typename T>
-    void operator() (int i, T func) {
+    void operator() (int i, unsigned long u, T func) {
         func();
     }
 };
@@ -95,3 +99,5 @@
 #define PROFILE_PREPB 2
 #define PROFILE_KERNEL 3
 #define PROFILE_MERGE 4
+
+