diff options
author | Viet-Hoa Do <viet-hoa.do@arm.com> | 2022-06-01 11:47:14 +0100 |
---|---|---|
committer | Viet-Hoa Do <viet-hoa.do@arm.com> | 2022-11-28 16:57:42 +0000 |
commit | 03b2971ac69a86f10a1566938d1a25afee15746c (patch) | |
tree | aec7cfc047e1da278b4b71a706cda7b1b0faa158 /src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp | |
parent | 7dc0234331f2150a6b4ac5c2b49de419870f7cf5 (diff) | |
download | ComputeLibrary-03b2971ac69a86f10a1566938d1a25afee15746c.tar.gz |
Integrate SME2 kernels
* Add SME/SME2 detection.
* Integrate SME2 implementation for:
- Normal convolution
- Winograd
- Depthwise convolution
- Pooling
Resolves: COMPMID-5700
Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Change-Id: I2f1ca1d05f8cfeee9309ed1c0a36096a4a6aad5c
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8692
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp')
-rw-r--r-- | src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp b/src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp index 73abe8b945..a221aee5d8 100644 --- a/src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp +++ b/src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp @@ -29,6 +29,11 @@ namespace arm_conv { namespace winograd { namespace output_transform { +#if defined(__aarch64__) +#if defined(ARM_COMPUTE_ENABLE_SVE) && defined(ARM_COMPUTE_ENABLE_SME) +void sme_fp32_mopa_4x4_3x3(unsigned int, const float *, size_t, const float *, float *, size_t, size_t, float, float); +#endif // defined(ARM_COMPUTE_ENABLE_SVE) && defined(ARM_COMPUTE_ENABLE_SME) +#endif // defined(__aarch64__) void arm_fp32_4x4_3x3(unsigned int, const float *, size_t, const float *, float *, size_t, size_t, float, float); void arm_fp32_2x2_3x3(unsigned int, const float *, size_t, const float *, float *, size_t, size_t, float, float); void arm_fp32_2x2_5x5(unsigned int, const float *, size_t, const float *, float *, size_t, size_t, float, float); @@ -44,6 +49,9 @@ void arm_fp32_1x2_1x7(unsigned int, const float *, size_t, const float *, float static const TransformImplementation<float> transforms_fp32[] = { #if defined(__aarch64__) +#if defined(ARM_COMPUTE_ENABLE_SVE) && defined(ARM_COMPUTE_ENABLE_SME) + { IMPL(4, 4, 3, 3, sme_fp32_mopa_4x4_3x3, Unpadded), MethodConstraints::RequiresSME }, +#endif // defined(ARM_COMPUTE_ENABLE_SVE) && defined(ARM_COMPUTE_ENABLE_SME) #endif // defined(__aarch64__) { IMPL(4, 4, 3, 3, arm_fp32_4x4_3x3, Unpadded), MethodConstraints::LargerShape }, { IMPL(2, 2, 3, 3, arm_fp32_2x2_3x3, Unpadded) }, |