COMPMID-750: Fix assembly kernel interfaces

Assembly kernel interfaces were wrongly translating the layout of the
input matrices. Boolean flags transform0 and transform1 do not match the
actual interface of the gemm assembly code which expects transpose0 and
transposed1.

Change-Id: Ia4df65a533834647fa63e78e8c897924793949df
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/113410
Tested-by: BSG Visual Compute Jenkins server to access repositories on http://mpd-gerrit.cambridge.arm.com <bsgcomp@arm.com>
Reviewed-by: Pablo Tello <pablo.tello@arm.com>

7 files changed, 24 insertions, 22 deletions

diff --git a/arm_compute/core/NEON/kernels/NEGEMMAssemblyBaseKernel.h b/arm_compute/core/NEON/kernels/NEGEMMAssemblyBaseKernel.h
index 9e0fe8059b..1090dd5b0a 100644
--- a/arm_compute/core/NEON/kernels/NEGEMMAssemblyBaseKernel.h
+++ b/arm_compute/core/NEON/kernels/NEGEMMAssemblyBaseKernel.h
@@ -36,7 +36,7 @@ class NEGEMMAssemblyBaseKernel : public INEKernel
 public:
     /** Constructor */
     NEGEMMAssemblyBaseKernel()
-        : _input0(nullptr), _input1(nullptr), _output(nullptr), _workspace(nullptr), _alpha(1.f), _beta(0.f), _transform_0(true), _transform_1(true)
+        : _input0(nullptr), _input1(nullptr), _output(nullptr), _workspace(nullptr), _alpha(1.f), _beta(0.f), _is_transposed_0(false), _is_transposed_1(false)
     {
     }
 
@@ -55,22 +55,22 @@ public:
      *
      * The computed function is C = a * AxB + b * C.
      *
-     * @param[in]     input0      Input tensor containing the Matrix A. Data types supported: F32
-     * @param[in]     input1      Input tensor containing the Matrix B. Data types supported: same as @p input0
-     * @param[in,out] output      Output tensor to store the result of matrix multiplication. If @p beta is not zero the values are multiplied by @p beta before the result is accumulated. Otherwise the values are overwritten by the result. Data types supported: same as @p input0.
-     * @param[out]    workspace   Space for intermediate results.
-     * @param[in]     alpha       Weight of the matrix product
-     * @param[in]     beta        Weight of the accumulation.
-     * @param[in]     transform_0 If true the kernel will transform @p input0 prior to the multiplication.
-     * @param[in]     transform_1 If true the kernel will transform @p input1 prior to the multiplication.
+     * @param[in]     input0          Input tensor containing the Matrix A. Data types supported: F32
+     * @param[in]     input1          Input tensor containing the Matrix B. Data types supported: same as @p input0
+     * @param[in,out] output          Output tensor to store the result of matrix multiplication. If @p beta is not zero the values are multiplied by @p beta before the result is accumulated. Otherwise the values are overwritten by the result. Data types supported: same as @p input0.
+     * @param[out]    workspace       Space for intermediate results.
+     * @param[in]     alpha           Weight of the matrix product
+     * @param[in]     beta            Weight of the accumulation.
+     * @param[in]     is_transposed_0 (Optional)True if @p input0 is transposed else false. (Defaults to false)
+     * @param[in]     is_transposed_1 (Optional)True if @p input1 is transposed else false. (Defaults to false)
      */
-    void configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha = 1.f, float beta = 0.f, bool transform_0 = true, bool transform_1 = true)
+    void configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha = 1.f, float beta = 0.f, bool is_transposed_0 = false, bool is_transposed_1 = false)
     {
-        internal_configure(input0, input1, output, workspace, alpha, beta, transform_0, transform_1);
+        internal_configure(input0, input1, output, workspace, alpha, beta, is_transposed_0, is_transposed_1);
     }
 
 protected:
-    virtual void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) = 0;
+    virtual void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool _is_transposed_0, bool _is_transposed_1) = 0;
 
     const ITensor *_input0;
     const ITensor *_input1;
@@ -78,8 +78,8 @@ protected:
     ITensor       *_workspace;
     float          _alpha;
     float          _beta;
-    bool           _transform_0;
-    bool           _transform_1;
+    bool           _is_transposed_0;
+    bool           _is_transposed_1;
 };
 } // namespace arm_compute
 #endif /*__ARM_COMPUTE_NEGEMMASSEMBLYBASE_H__*/
diff --git a/arm_compute/core/NEON/kernels/arm32/NEGEMMAArch32Kernel.h b/arm_compute/core/NEON/kernels/arm32/NEGEMMAArch32Kernel.h
index 597acca439..7564f6a0e1 100644
--- a/arm_compute/core/NEON/kernels/arm32/NEGEMMAArch32Kernel.h
+++ b/arm_compute/core/NEON/kernels/arm32/NEGEMMAArch32Kernel.h
@@ -38,7 +38,7 @@ public:
     void run(const Window &window, const ThreadInfo &info) override;
 
 protected:
-    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) override;
+    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1) override;
 };
 } // namespace arm_compute
 #endif /*__ARM_COMPUTE_NEGEMMAARCH32KERNEL_H__*/
diff --git a/arm_compute/core/NEON/kernels/arm64/NEGEMMAArch64Kernel.h b/arm_compute/core/NEON/kernels/arm64/NEGEMMAArch64Kernel.h
index 77431d2bc8..5c29a825c2 100644
--- a/arm_compute/core/NEON/kernels/arm64/NEGEMMAArch64Kernel.h
+++ b/arm_compute/core/NEON/kernels/arm64/NEGEMMAArch64Kernel.h
@@ -38,7 +38,7 @@ public:
     void run(const Window &window, const ThreadInfo &info) override;
 
 protected:
-    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) override;
+    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1) override;
 };
 } // namespace arm_compute
 #endif /*__ARM_COMPUTE_NEGEMMAARCH64KERNEL_H__*/
diff --git a/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64A53Kernel.h b/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64A53Kernel.h
index 33cd2d42d0..8e9783720e 100644
--- a/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64A53Kernel.h
+++ b/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64A53Kernel.h
@@ -43,10 +43,11 @@ public:
     void run(const Window &window, const ThreadInfo &info) override;
 
 protected:
-    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) override;
+    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1) override;
 
 private:
-    using NEGEMMLowpAArch64A53 = void(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1, const Window &window,
+    using NEGEMMLowpAArch64A53 = void(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1,
+                                      const Window     &window,
                                       const ThreadInfo &info);
     NEGEMMLowpAArch64A53 *_func;
 };
diff --git a/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64Kernel.h b/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64Kernel.h
index a93df033de..3829d5e1d6 100644
--- a/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64Kernel.h
+++ b/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64Kernel.h
@@ -44,10 +44,11 @@ public:
     void run(const Window &window, const ThreadInfo &info) override;
 
 protected:
-    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) override;
+    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1) override;
 
 private:
-    using NEGEMMLowpAArch64 = void(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1, const Window &window,
+    using NEGEMMLowpAArch64 = void(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0,
+                                   bool is_transposed_1, const Window &window,
                                    const ThreadInfo &info);
     NEGEMMLowpAArch64 *_func;
 };
diff --git a/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64V8P4Kernel.h b/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64V8P4Kernel.h
index b03e5fa1a2..b94499392e 100644
--- a/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64V8P4Kernel.h
+++ b/arm_compute/core/NEON/kernels/arm64/NEGEMMLowpAArch64V8P4Kernel.h
@@ -51,7 +51,7 @@ public:
     static Status validate(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output);
 
 protected:
-    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) override;
+    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1) override;
 };
 } // namespace arm_compute
 #endif /* ARM_COMPUTE_AARCH64_V8_2 */
diff --git a/arm_compute/core/NEON/kernels/arm64/NEHGEMMAArch64FP16Kernel.h b/arm_compute/core/NEON/kernels/arm64/NEHGEMMAArch64FP16Kernel.h
index 9480a6a5d0..5671d99b85 100644
--- a/arm_compute/core/NEON/kernels/arm64/NEHGEMMAArch64FP16Kernel.h
+++ b/arm_compute/core/NEON/kernels/arm64/NEHGEMMAArch64FP16Kernel.h
@@ -38,7 +38,7 @@ public:
     void run(const Window &window, const ThreadInfo &info) override;
 
 protected:
-    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool transform_0, bool transform_1) override;
+    void internal_configure(const ITensor *input0, const ITensor *input1, ITensor *output, ITensor *workspace, float alpha, float beta, bool is_transposed_0, bool is_transposed_1) override;
 };
 } // namespace arm_compute
 #endif /*__ARM_COMPUTE_NEHGEMMAARCH64FP16KERNEL_H__*/