diff options
author | Michael Tyler <michael.tyler@arm.com> | 2023-07-07 12:01:32 +0100 |
---|---|---|
committer | michael.tyler <michael.tyler@arm.com> | 2023-07-13 12:21:29 +0000 |
commit | 4c30de056afe8680b42723b26a2241811715b989 (patch) | |
tree | 4f522a816a5ea1b58b51226eb685c786096f30e3 /src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp | |
parent | c8e1617807ef1985a39d8f8f5f69c113b758494d (diff) | |
download | ComputeLibrary-4c30de056afe8680b42723b26a2241811715b989.tar.gz |
Enable premultiplication for depthwise convolution
with fp16 and quantized types
Resolves: COMPMID-6337
Change-Id: I81542e51c9c0329f202ac8452f173b138e51a0f6
Signed-off-by: Michael Tyler <michael.tyler@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/9883
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp')
-rw-r--r-- | src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp | 54 |
1 files changed, 46 insertions, 8 deletions
diff --git a/src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp b/src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp index 3b76e52206..8fef6f8ae0 100644 --- a/src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp +++ b/src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp @@ -66,9 +66,47 @@ namespace depthwise { namespace { +#if defined(__aarch64__) +#if defined(ENABLE_FP16_KERNELS) && defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) + bool prefer_premultiply(const DepthwiseArgs &args) { + if ((args.stride_rows != args.stride_cols) || (args.kernel_rows != args.kernel_cols)) + { + return false; + } + + unsigned int threshold; + + if (args.stride_rows == 1 && args.kernel_rows == 3) + { + threshold = 30; + } + else if (args.stride_rows == 1 && args.kernel_rows == 5) + { + threshold = 31; + } + else if (args.stride_rows == 2 && args.kernel_rows == 3) + { + threshold = 11; + } + else if (args.stride_rows == 2 && args.kernel_rows == 5) + { + threshold = 19; + } else + { + return false; + } + + return args.channel_multiplier <= threshold; + } + template <class Strategy> unsigned int cycle_estimate(const DepthwiseArgs &args, const Nothing &) { + if (args.channel_multiplier > 1 && !prefer_premultiply(args)) + { + return std::numeric_limits<unsigned int>::max(); + } + // First-pass: compute the number of output pixels which will be computed. return arm_gemm::roundup(args.output_rows, Strategy::output_rows) * arm_gemm::roundup(args.output_cols, Strategy::output_cols) * @@ -90,13 +128,18 @@ namespace ); } -#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) + unsigned int multiplier_cycle_estimate(const DepthwiseArgs &args, const Nothing &) + { + return prefer_premultiply(args)? std::numeric_limits<unsigned int>::max() : 0; + } + unsigned int not_preferred(const DepthwiseArgs &, const Nothing &) __attribute__ ((unused)); unsigned int not_preferred(const DepthwiseArgs &, const Nothing &) { return std::numeric_limits<unsigned int>::max(); } -#endif // defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) +#endif // defined(ENABLE_FP16_KERNELS) && defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) +#endif // defined(__aarch64__) } static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = { @@ -108,7 +151,6 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sme2_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst", constraint(is_supported<sme2_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst>, - has_no_channel_multiplier, cpu_has_sme2), cycle_estimate<sme2_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { @@ -120,7 +162,6 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sme2_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst", constraint(is_supported<sme2_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst>, - has_no_channel_multiplier, cpu_has_sme2), cycle_estimate<sme2_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { @@ -132,7 +173,6 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sme2_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst", constraint(is_supported<sme2_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst>, - has_no_channel_multiplier, cpu_has_sme2), cycle_estimate<sme2_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { @@ -144,7 +184,6 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sme2_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst", constraint(is_supported<sme2_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst>, - has_no_channel_multiplier, cpu_has_sme2), cycle_estimate<sme2_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { @@ -156,7 +195,6 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sme2_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst", constraint(is_supported<sme2_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst>, - has_no_channel_multiplier, cpu_has_sme2), cycle_estimate<sme2_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { @@ -291,7 +329,7 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "a64_fp16_nhwc_generic_with_multiplier_output2x8_mla_depthfirst", constraint(cpu_has_fp16, has_channel_multiplier), - nullptr, + multiplier_cycle_estimate, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { auto kern = new a64_fp16_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst(args.cpu_info); auto strat = new GenericDepthfirstMultiplierStrategy<__fp16>(kern, args); |