1 // Copyright 2015 Google Inc. All Rights Reserved. 2 // 3 // Licensed under the Apache License, Version 2.0 (the "License"); 4 // you may not use this file except in compliance with the License. 5 // You may obtain a copy of the License at 6 // 7 // http://www.apache.org/licenses/LICENSE-2.0 8 // 9 // Unless required by applicable law or agreed to in writing, software 10 // distributed under the License is distributed on an "AS IS" BASIS, 11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 // See the License for the specific language governing permissions and 13 // limitations under the License. 14 15 // kernel_default.h: Chooses default GEMM and GEMV kernels for the 16 // host platform. 17 18 #ifndef GEMMLOWP_INTERNAL_KERNEL_DEFAULT_H_ 19 #define GEMMLOWP_INTERNAL_KERNEL_DEFAULT_H_ 20 21 #include "../public/bit_depth.h" 22 #include "common.h" 23 24 namespace gemmlowp { 25 26 enum class KernelFamily { Gemm, Gemv }; 27 28 template <KernelFamily Family, int ProductBits> 29 struct DefaultKernelImpl : DefaultKernelImpl<Family, ProductBits + 1> { 30 static_assert(ProductBits <= 16, "Bit depth too large"); 31 }; 32 33 template <KernelFamily Family, typename BitDepthParams> 34 struct DefaultKernel 35 : DefaultKernelImpl<Family, BitDepthParams::LhsBitDepth::kBits + 36 BitDepthParams::RhsBitDepth::kBits> {}; 37 38 } // end namespace gemmlowp 39 40 #define GEMMLOWP_SET_DEFAULT_KERNEL(op, max_product_bits, kernel) \ 41 namespace gemmlowp { \ 42 template <> \ 43 struct DefaultKernelImpl<KernelFamily::op, max_product_bits> : kernel {}; \ 44 } 45 46 #if defined GEMMLOWP_NEON_32 47 #include "kernel_neon.h" 48 GEMMLOWP_SET_DEFAULT_KERNEL(Gemm, 16, NEON_32_Kernel12x4Depth2) 49 GEMMLOWP_SET_DEFAULT_KERNEL(Gemm, 12, 50 NEON_32_Kernel12x4Depth2Assuming12BitProducts) 51 GEMMLOWP_SET_DEFAULT_KERNEL(Gemv, 16, NEONKernel4Nx1Depth2<3>) 52 #elif defined GEMMLOWP_NEON_64 53 #include "kernel_neon.h" 54 GEMMLOWP_SET_DEFAULT_KERNEL(Gemm, 16, NEON_64_Kernel12x8Depth2) 55 GEMMLOWP_SET_DEFAULT_KERNEL(Gemv, 16, NEONKernel4Nx1Depth2<3>) 56 #elif defined GEMMLOWP_SSE4_32 57 #include "kernel_SSE.h" 58 GEMMLOWP_SET_DEFAULT_KERNEL(Gemm, 16, SSE4_32_Kernel4x4Depth2) 59 GEMMLOWP_SET_DEFAULT_KERNEL(Gemv, 16, SSE4_32_Kernel4x4Depth2) 60 #elif defined GEMMLOWP_SSE4_64 61 #include "kernel_SSE.h" 62 GEMMLOWP_SET_DEFAULT_KERNEL(Gemm, 16, SSE4_64_Kernel12x4Depth2) 63 GEMMLOWP_SET_DEFAULT_KERNEL(Gemv, 16, SSE4_64_Kernel12x4Depth2) 64 #else 65 #include "kernel_reference.h" 66 namespace gemmlowp { 67 typedef ReferenceKernel<KernelFormat<KernelSideFormat<CellFormat<4, 4>, 2>, 68 KernelSideFormat<CellFormat<4, 4>, 2> > > 69 DefaultReferenceKernel; 70 } 71 GEMMLOWP_SET_DEFAULT_KERNEL(Gemm, 16, DefaultReferenceKernel) 72 GEMMLOWP_SET_DEFAULT_KERNEL(Gemv, 16, DefaultReferenceKernel) 73 #endif 74 75 #endif // GEMMLOWP_INTERNAL_KERNEL_DEFAULT_H_ 76