From 7cd26d4a1b14bc4bf7c61496803416ab3d84791f Mon Sep 17 00:00:00 2001
From: Georgios Pinitas <georgios.pinitas@arm.com>
Date: Wed, 9 Jan 2019 18:35:17 +0000
Subject: COMPMID-1867: Add NEON/SVE GEMM Hybrid kernels.

Change-Id: Ib40a9921e7f9a6a8be6c38872d6b3a0f24ed0cd3
Reviewed-on: https://review.mlplatform.org/515
Reviewed-by: Anthony Barbier <Anthony.barbier@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
---
 src/core/NEON/kernels/arm_gemm/transform.hpp | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

(limited to 'src/core/NEON/kernels/arm_gemm/transform.hpp')

diff --git a/src/core/NEON/kernels/arm_gemm/transform.hpp b/src/core/NEON/kernels/arm_gemm/transform.hpp
index e422b91c83..0330783a0b 100644
--- a/src/core/NEON/kernels/arm_gemm/transform.hpp
+++ b/src/core/NEON/kernels/arm_gemm/transform.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2018 ARM Limited.
+ * Copyright (c) 2017-2019 ARM Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -81,11 +81,14 @@ struct TransformImpl {
                     }
                 }
                 // "row" tail - row is out of range so fill with zeros always.
-                for (int row = 0; row < blank_rows; row++) {
-                    for (int col=0; col < (fill_cols + blank_cols); col++) {
-                        *out++ = static_cast<TOut>(0);
-                    }
+                TOut zeroval = static_cast<TOut>(0);
+                int pads = blank_rows * (fill_cols + blank_cols);
+
+                for (int i=0; i<pads; i++) {
+                    out[i] = zeroval;
                 }
+
+                out += pads;
             }
         }
     }
-- 
cgit v1.2.1