COMPMID-3930: Update CLGEMM heuristic for fp16. Mali-G76

- Since the GEMM kernel can now work without padding, the heuristic
  requires to be fine-tuned to exploit this feature
- The heuristic affects Mali-G76 FP16 only

Change-Id: Ia430627f02131ad956ce2219b80c83c8e7cabaf2
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4284
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>

2 files changed, 64 insertions, 3 deletions

diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp
index 73b90568f5..7c6efe3f11 100644
--- a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp
+++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp
@@ -72,7 +72,7 @@ CLGEMMKernelType CLGEMMKernelSelectionBifrost::select_kernel(const CLGEMMKernelS
     static std::map<DataType, FunctionExecutorPtr> gemm_g76_configs =
     {
         { DataType::F32, &CLGEMMKernelSelectionBifrost::g76_f32 },
-        { DataType::F16, &CLGEMMKernelSelectionBifrost::default_f16 },
+        { DataType::F16, &CLGEMMKernelSelectionBifrost::g76_f16 },
         { DataType::QASYMM8, &CLGEMMKernelSelectionBifrost::default_q8 },
         { DataType::QASYMM8_SIGNED, &CLGEMMKernelSelectionBifrost::default_q8 },
         { DataType::QSYMM8, &CLGEMMKernelSelectionBifrost::default_q8 },
@@ -188,12 +188,10 @@ CLGEMMKernelType CLGEMMKernelSelectionBifrost::g76_f32(unsigned int m, unsigned
     {
         return CLGEMMKernelType::NATIVE_V1;
     }
-
     if(m == 1)
     {
         return CLGEMMKernelType::RESHAPED_ONLY_RHS;
     }
-
     if(k <= 496)
     {
         if(n <= 544)
@@ -239,6 +237,68 @@ CLGEMMKernelType CLGEMMKernelSelectionBifrost::g76_f32(unsigned int m, unsigned
     }
 }
 
+CLGEMMKernelType CLGEMMKernelSelectionBifrost::g76_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant)
+{
+    ARM_COMPUTE_UNUSED(b);
+
+    if (!is_rhs_constant)
+    {
+        return CLGEMMKernelType::NATIVE_V1;
+    }
+
+    if (m == 1)
+    {
+        return CLGEMMKernelType::RESHAPED_ONLY_RHS;
+    }
+
+    const float r_mn = static_cast<float>(m) / static_cast<float>(n);
+    const float r_nk = static_cast<float>(n) / static_cast<float>(k);
+
+    if(k <= 212)
+    {
+        return CLGEMMKernelType::RESHAPED_ONLY_RHS;
+    }
+    else
+    {
+        if(r_nk <= 0.4990234375f)
+        {
+            if(k <= 1392)
+            {
+                return CLGEMMKernelType::RESHAPED_ONLY_RHS;
+            }
+            else
+            {
+                if(m <= 325)
+                {
+                    return CLGEMMKernelType::RESHAPED_ONLY_RHS;
+                }
+                else
+                {
+                    return CLGEMMKernelType::RESHAPED;
+                }
+            }
+        }
+        else
+        {
+            if(k <= 471)
+            {
+                return CLGEMMKernelType::RESHAPED_ONLY_RHS;
+            }
+            else
+            {
+                if(r_mn <= 0.04475911520421505f)
+                {
+                    return CLGEMMKernelType::RESHAPED;
+                }
+                else
+                {
+                    return CLGEMMKernelType::RESHAPED_ONLY_RHS;
+                }
+            }
+        }
+    }
+}
+
 CLGEMMKernelType CLGEMMKernelSelectionBifrost::g71_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant)
 {
     ARM_COMPUTE_UNUSED(b);
diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h
index a495b48301..e3cc8e4a27 100644
--- a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h
+++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h
@@ -45,6 +45,7 @@ public:
 
 private:
     CLGEMMKernelType g76_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant);
+    CLGEMMKernelType g76_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant);
     CLGEMMKernelType g71_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant);
     CLGEMMKernelType default_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant);
     CLGEMMKernelType default_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant);