From 229757be70e7884b1a64972c2bdf9919075401a2 Mon Sep 17 00:00:00 2001
From: Gian Marco Iodice <gianmarco.iodice@arm.com>
Date: Sun, 15 Nov 2020 16:06:10 +0000
Subject: COMPMID-3975: Update CLGEMM heuristic for Mali-G52 F16

- Update the GEMM heuristic for Mali-G52
- Fix the GEMV heuristic on Mali-G76

Change-Id: I895153d02d09e73c51ecf158e2ae43f81f6d7b05
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4425
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
---
 .../CLGEMMReshapedKernelConfigurationBifrost.cpp       | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

(limited to 'src/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfigurationBifrost.cpp')

diff --git a/src/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfigurationBifrost.cpp b/src/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfigurationBifrost.cpp
index 70992974a3..e5c89f89af 100644
--- a/src/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfigurationBifrost.cpp
+++ b/src/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfigurationBifrost.cpp
@@ -64,7 +64,7 @@ std::pair<GEMMLHSMatrixInfo, GEMMRHSMatrixInfo> CLGEMMReshapedKernelConfiguratio
     static std::map<DataType, ConfigurationFunctionExecutorPtr> gemm_configs_G52 =
     {
         { DataType::F32, &CLGEMMReshapedKernelConfigurationBifrost::configure_G52_f32 },
-        { DataType::F16, &CLGEMMReshapedKernelConfigurationBifrost::configure_G7x_f16 },
+        { DataType::F16, &CLGEMMReshapedKernelConfigurationBifrost::configure_G52_f16 },
         { DataType::QASYMM8, &CLGEMMReshapedKernelConfigurationBifrost::configure_G7x_u8 },
         { DataType::QSYMM8, &CLGEMMReshapedKernelConfigurationBifrost::configure_G7x_u8 },
         { DataType::QASYMM8_SIGNED, &CLGEMMReshapedKernelConfigurationBifrost::configure_G7x_u8 },
@@ -263,6 +263,22 @@ std::pair<GEMMLHSMatrixInfo, GEMMRHSMatrixInfo> CLGEMMReshapedKernelConfiguratio
     }
 }
 
+std::pair<GEMMLHSMatrixInfo, GEMMRHSMatrixInfo> CLGEMMReshapedKernelConfigurationBifrost::configure_G52_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b)
+{
+    ARM_COMPUTE_UNUSED(k);
+
+    const float workload = (static_cast<float>(m) * static_cast<float>(n) * static_cast<float>(b)) / 20.0f;
+
+    if(workload <= 232.8000f)
+    {
+        return configure_lhs_rhs_info(m, n, 2, 4, 4, 4, 4, true, true, true, false, false);
+    }
+    else
+    {
+        return configure_lhs_rhs_info(m, n, 4, 4, 4, 4, 4, true, true, true, false, false);
+    }
+}
+
 std::pair<GEMMLHSMatrixInfo, GEMMRHSMatrixInfo> CLGEMMReshapedKernelConfigurationBifrost::configure_G76_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b)
 {
     ARM_COMPUTE_UNUSED(k);
-- 
cgit v1.2.1