diff options
author | Aleksandr Nikolaev <aleksandr.nikolaev@arm.com> | 2020-06-25 12:25:52 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2020-07-16 16:02:50 +0000 |
commit | a084b46835d20fdfe6e590b91b7ca64fba3542df (patch) | |
tree | e4abe49cc10f3406dcaad9d3dfbbc22cc54476a6 /src/core | |
parent | f738fe6b6e059916294c48b942952c261569df18 (diff) | |
download | ComputeLibrary-a084b46835d20fdfe6e590b91b7ca64fba3542df.tar.gz |
[ONCPUML-97]: Implement "int8" support for 2D decomposition at high core counts
Interleaved2d functionality was extended to uint8 and int8 kernels.
Change-Id: If78facbce56e9ec7b2f4c23436af0bd5db7f7b69
Signed-off-by: Aleksandr Nikolaev <aleksandr.nikolaev@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3467
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core')
-rw-r--r-- | src/core/NEON/kernels/arm_gemm/gemm_int8.cpp | 19 | ||||
-rw-r--r-- | src/core/NEON/kernels/arm_gemm/gemm_qint8.cpp | 10 | ||||
-rw-r--r-- | src/core/NEON/kernels/arm_gemm/gemm_quint8.cpp | 10 | ||||
-rw-r--r-- | src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp | 19 | ||||
-rw-r--r-- | src/core/NEON/kernels/assembly/arm_gemm.hpp | 1 | ||||
-rw-r--r-- | src/core/NEON/kernels/assembly/gemm_common.hpp | 2 |
6 files changed, 54 insertions, 7 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp index d1d137e090..3ee47492db 100644 --- a/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp +++ b/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp @@ -28,6 +28,7 @@ #include "gemm_hybrid.hpp" #include "gemm_implementation.hpp" #include "gemm_interleaved.hpp" +#include "gemm_interleaved_pretransposed_2d.hpp" #include "kernels/a64_gemm_s16_12x8.hpp" #include "kernels/a64_gemm_s8_12x8.hpp" @@ -107,15 +108,29 @@ static const GemmImplementation<int8_t, int32_t> gemm_s8_methods[] = { [](const GemmArgs &args) { return new GemmHybrid<hybrid_s8s32_dot_16x4, int8_t, int32_t>(args); } }, { + GemmMethod::GEMM_INTERLEAVED_2D, + "gemm_s8_12x8_2d", + [](const GemmArgs &args) { return args._ci->has_dotprod(); }, + [](const GemmArgs &args) { return (args._maxthreads >= 8) && (args._Msize >= 8) && (args._Nsize >= 8); }, + [](const GemmArgs &args) { return new GemmInterleavedPretransposed2d<gemm_s8_12x8, int8_t, int32_t>(args); } +}, +{ GemmMethod::GEMM_INTERLEAVED, - "gemm_s8_12x8", + "gemm_s8_12x8_1d", [](const GemmArgs &args) { return args._ci->has_dotprod(); }, nullptr, [](const GemmArgs &args) { return new GemmInterleaved<gemm_s8_12x8, int8_t, int32_t>(args); } }, { + GemmMethod::GEMM_INTERLEAVED_2D, + "gemm_s8_4x4_2d", + nullptr, + [](const GemmArgs &args) { return (args._maxthreads >= 8) && (args._Msize >= 8) && (args._Nsize >= 8); }, + [](const GemmArgs &args) { return new GemmInterleavedPretransposed2d<gemm_s8_4x4, int8_t, int32_t>(args); } +}, +{ GemmMethod::GEMM_INTERLEAVED, - "gemm_s8_4x4", + "gemm_s8_4x4_1d", nullptr, nullptr, [](const GemmArgs &args) { return new GemmInterleaved<gemm_s8_4x4, int8_t, int32_t>(args); } diff --git a/src/core/NEON/kernels/arm_gemm/gemm_qint8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_qint8.cpp index 67f28d38e2..04cac6095c 100644 --- a/src/core/NEON/kernels/arm_gemm/gemm_qint8.cpp +++ b/src/core/NEON/kernels/arm_gemm/gemm_qint8.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019 Arm Limited. + * Copyright (c) 2019-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -75,6 +75,14 @@ static const GemmImplementation<int8_t, int8_t, Requantize32> gemm_qint8_methods [](const GemmArgs &args, const Requantize32 &) { return args._Nsize<=256 && args._Ksize>128; }, [](const GemmArgs &args, const Requantize32 &qp) { return new GemmHybridQuantized<hybrid_s8s32_dot_16x4, int8_t, int8_t>(args, qp); } }, +/** QUANTIZE_WRAPPER_2D enables 2D parallelisation hint for IScheduler in NEGEMMAssemblyDispatch */ +{ + GemmMethod::QUANTIZE_WRAPPER_2D, + "quantized_wrapper_2d", + nullptr, + [](const GemmArgs &args, const Requantize32 &) { return (args._maxthreads >= 8) && (args._Msize >= 8) && (args._Nsize >= 8);}, + [](const GemmArgs &args, const Requantize32 &qp) { return new QuantizeWrapper<int8_t, int8_t, int32_t>(args, qp); } +}, { GemmMethod::QUANTIZE_WRAPPER, "quantized_wrapper", diff --git a/src/core/NEON/kernels/arm_gemm/gemm_quint8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_quint8.cpp index b9e2bf6c26..7b08041005 100644 --- a/src/core/NEON/kernels/arm_gemm/gemm_quint8.cpp +++ b/src/core/NEON/kernels/arm_gemm/gemm_quint8.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019 Arm Limited. + * Copyright (c) 2019-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -75,6 +75,14 @@ static const GemmImplementation<uint8_t, uint8_t, Requantize32> gemm_quint8_meth [](const GemmArgs &args, const Requantize32 &) { return args._Nsize<=256 && args._Ksize>128; }, [](const GemmArgs &args, const Requantize32 &qp) { return new GemmHybridQuantized<hybrid_u8u32_dot_16x4, uint8_t, uint8_t>(args, qp); } }, +/** QUANTIZE_WRAPPER_2D enables 2D parallelisation hint for IScheduler in NEGEMMAssemblyDispatch */ +{ + GemmMethod::QUANTIZE_WRAPPER_2D, + "quantized_wrapper_2d", + nullptr, + [](const GemmArgs &args, const Requantize32 &) { return (args._maxthreads >= 8) && (args._Msize >= 8) && (args._Nsize >= 8);}, + [](const GemmArgs &args, const Requantize32 &qp) { return new QuantizeWrapper<uint8_t, uint8_t, uint32_t>(args, qp); } +}, { GemmMethod::QUANTIZE_WRAPPER, "quantized_wrapper", diff --git a/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp index d5a9e585b5..caab2e2cc2 100644 --- a/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp +++ b/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp @@ -27,6 +27,7 @@ #include "gemm_common.hpp" #include "gemm_implementation.hpp" #include "gemm_interleaved.hpp" +#include "gemm_interleaved_pretransposed_2d.hpp" #include "gemm_hybrid.hpp" #include "kernels/a64_gemm_u16_12x8.hpp" @@ -107,15 +108,29 @@ static const GemmImplementation<uint8_t, uint32_t> gemm_u8_methods[] = { [](const GemmArgs &args) { return new GemmHybrid<hybrid_u8u32_dot_16x4, uint8_t, uint32_t>(args); } }, { + GemmMethod::GEMM_INTERLEAVED_2D, + "gemm_u8_12x8_2d", + [](const GemmArgs &args) { return args._ci->has_dotprod(); }, + [](const GemmArgs &args) { return (args._maxthreads >= 8) && (args._Msize >= 8) && (args._Nsize >= 8) ; }, + [](const GemmArgs &args) { return new GemmInterleavedPretransposed2d<gemm_u8_12x8, uint8_t, uint32_t>(args); } +}, +{ GemmMethod::GEMM_INTERLEAVED, - "gemm_u8_12x8", + "gemm_u8_12x8_1d", [](const GemmArgs &args) { return args._ci->has_dotprod(); }, nullptr, [](const GemmArgs &args) { return new GemmInterleaved<gemm_u8_12x8, uint8_t, uint32_t>(args); } }, { + GemmMethod::GEMM_INTERLEAVED_2D, + "gemm_u8_4x4_2d", + nullptr, + [](const GemmArgs &args) { return (args._maxthreads >= 8) && (args._Msize >= 8) && (args._Nsize >= 8); }, + [](const GemmArgs &args) { return new GemmInterleavedPretransposed2d<gemm_u8_4x4, uint8_t, uint32_t>(args); } +}, +{ GemmMethod::GEMM_INTERLEAVED, - "gemm_u8_4x4", + "gemm_u8_4x4_1d", nullptr, nullptr, [](const GemmArgs &args) { return new GemmInterleaved<gemm_u8_4x4, uint8_t, uint32_t>(args); } diff --git a/src/core/NEON/kernels/assembly/arm_gemm.hpp b/src/core/NEON/kernels/assembly/arm_gemm.hpp index 1a613e2160..e3e15478fb 100644 --- a/src/core/NEON/kernels/assembly/arm_gemm.hpp +++ b/src/core/NEON/kernels/assembly/arm_gemm.hpp @@ -42,6 +42,7 @@ enum class GemmMethod GEMM_INTERLEAVED, GEMM_INTERLEAVED_2D, QUANTIZE_WRAPPER, + QUANTIZE_WRAPPER_2D, GEMM_HYBRID_QUANTIZED }; diff --git a/src/core/NEON/kernels/assembly/gemm_common.hpp b/src/core/NEON/kernels/assembly/gemm_common.hpp index 8feecf440e..e9e56842c7 100644 --- a/src/core/NEON/kernels/assembly/gemm_common.hpp +++ b/src/core/NEON/kernels/assembly/gemm_common.hpp @@ -77,7 +77,7 @@ public: return false; } - /** Main execute member fucntion + /** Main execute member function * @param [in] work_range specifies the range of work we want to be computed, total range defined by get_window_size() * @param [in] thread_locator where are we inside of the thread space * @naram [in] threadid a unique threadid |