diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2017-12-14 17:53:39 +0000 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:42:33 +0000 |
| commit | 08c5a06e2b49df0d7912deedd6d26d2c603cfe58 (patch) | |
| tree | 0742d970b5dff2354f917063ad11980af97dee93 /arm_compute/core/NEON/kernels/arm64 | |
| parent | 941cd706bea1847ae89e4ee13f144fc51050ad1f (diff) | |
| download | ComputeLibrary-08c5a06e2b49df0d7912deedd6d26d2c603cfe58.tar.gz | |
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>
Diffstat (limited to 'arm_compute/core/NEON/kernels/arm64')
5 files changed, 9 insertions, 7 deletions
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__*/ |