diff options
author | Michalis Spyrou <michalis.spyrou@arm.com> | 2021-06-07 14:23:57 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2021-06-23 12:25:50 +0000 |
commit | 20fca524baf99402f742ce38c538f2fd07d5fff9 (patch) | |
tree | b63d98383d1ba22bb3ca59d393e4ab9d47a9c762 /src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp | |
parent | 1d359279e22874121def2ce4bfdb633d94ea5ade (diff) | |
download | ComputeLibrary-20fca524baf99402f742ce38c538f2fd07d5fff9.tar.gz |
Create core library using high priority operators
A smaller core library is created using a subset of the operators.
Changed the structure of filelist.json in order to include more
information about the kernels and make the selection easier.
Resolves: COMPMID-4514
Change-Id: I079ca7d8e64346174eebdd13b834e1dd4dc36ca2
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5786
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/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp')
-rw-r--r-- | src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp | 23 |
1 files changed, 14 insertions, 9 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 fdb36fc1d1..6ba7c78e97 100644 --- a/src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp +++ b/src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp @@ -33,13 +33,13 @@ #include "depthwise_implementation_constraints.hpp" #if defined(__aarch64__) -#if defined(__ARM_FEATURE_SVE) +#if defined(ARM_COMPUTE_ENABLE_SVE) #include "kernels/sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst.hpp" #include "kernels/sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst.hpp" #include "kernels/sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst.hpp" #include "kernels/sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst.hpp" #include "kernels/sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst.hpp" -#endif // defined(__ARM_FEATURE_SVE) +#endif // defined(ARM_COMPUTE_ENABLE_SVE) #include "kernels/a64_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst.hpp" #include "kernels/a64_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst.hpp" #include "kernels/a64_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst.hpp" @@ -83,12 +83,13 @@ namespace static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = { #if defined(__aarch64__) -#if defined(__ARM_FEATURE_SVE) +#if defined(ARM_COMPUTE_ENABLE_SVE) { DepthwiseMethod::DEPTHFIRST, "sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst", constraint(is_supported<sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst>, - has_no_channel_multiplier), + has_no_channel_multiplier, + cpu_has_sve), cycle_estimate<sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { return new DepthwiseDepthfirst<sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst>(args); @@ -98,7 +99,8 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst", constraint(is_supported<sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst>, - has_no_channel_multiplier), + has_no_channel_multiplier, + cpu_has_sve), cycle_estimate<sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { return new DepthwiseDepthfirst<sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst>(args); @@ -108,7 +110,8 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst", constraint(is_supported<sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst>, - has_no_channel_multiplier), + has_no_channel_multiplier, + cpu_has_sve), cycle_estimate<sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { return new DepthwiseDepthfirst<sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst>(args); @@ -118,7 +121,8 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst", constraint(is_supported<sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst>, - has_no_channel_multiplier), + has_no_channel_multiplier, + cpu_has_sve), cycle_estimate<sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { return new DepthwiseDepthfirst<sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst>(args); @@ -128,13 +132,14 @@ static const DepthwiseImplementation<__fp16, __fp16> depthwise_fp16_methods[] = DepthwiseMethod::DEPTHFIRST, "sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst", constraint(is_supported<sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst>, - has_no_channel_multiplier), + has_no_channel_multiplier, + cpu_has_sve), cycle_estimate<sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst>, [] (const DepthwiseArgs &args, const Nothing &) -> DepthwiseCommon<__fp16, __fp16, __fp16> * { return new DepthwiseDepthfirst<sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst>(args); }, }, -#endif // defined(__ARM_FEATURE_SVE) +#endif // defined(ARM_COMPUTE_ENABLE_SVE) #if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) { DepthwiseMethod::DEPTHFIRST, |