sergiotarxz/nuttx-apps
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
f4a41c6ed4
nuttx-apps/mlearning/cmsis/libcmsisnn/Makefile

108 lines
6.2 KiB
Makefile
Raw Normal View History

Add support for CMSIS NN the CMSIS NN software library is a collection of efficient neural network kernels developed to maximize the performance and minimize the memory footprint of neural networks on Cortex-M processor cores. Project https://github.com/ARM-software/CMSIS_5 The library is divided into a number of functions each covering a specific category: Convolution Functions Activation Functions Fully-connected Layer Functions SVDF Layer Functions Pooling Functions Softmax Functions Basic math Functions The library has separate functions for operating on different weight and activation data types including 8-bit integers (q7_t) and 16-bit integers (q15_t). The descrition of the kernels are included in the function description. More information https://www.keil.com/pack/doc/CMSIS/NN/html/index.html Project license : Apache 2.0 License https://github.com/ARM-software/CMSIS_5/blob/develop/LICENSE.txt Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-01-26 17:34:58 +01:00
############################################################################
# apps/mlearning/cmsis/libcmsisnn/Makefile
#
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements. See the NOTICE file distributed with
# this work for additional information regarding copyright ownership. The
# ASF licenses this file to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance with the
# License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
#
############################################################################
include $(APPDIR)/Make.defs
CMSIS5_NN := ../CMSIS_5/CMSIS/NN/Source
mlearning: cmsis: enable FPU support In case of HW with FPU we can benefit from the FPU support. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-02-24 13:18:38 +01:00
ifeq ($(CONFIG_ARCH_FPU),y)
CFLAGS += -D__FPU_PRESENT=1U
CXXFLAGS += -D__FPU_PRESENT=1U
endif
Add support for CMSIS NN the CMSIS NN software library is a collection of efficient neural network kernels developed to maximize the performance and minimize the memory footprint of neural networks on Cortex-M processor cores. Project https://github.com/ARM-software/CMSIS_5 The library is divided into a number of functions each covering a specific category: Convolution Functions Activation Functions Fully-connected Layer Functions SVDF Layer Functions Pooling Functions Softmax Functions Basic math Functions The library has separate functions for operating on different weight and activation data types including 8-bit integers (q7_t) and 16-bit integers (q15_t). The descrition of the kernels are included in the function description. More information https://www.keil.com/pack/doc/CMSIS/NN/html/index.html Project license : Apache 2.0 License https://github.com/ARM-software/CMSIS_5/blob/develop/LICENSE.txt Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-01-26 17:34:58 +01:00
CSRCS += $(CMSIS5_NN)/ActivationFunctions/arm_nn_activations_q15.c
CSRCS += $(CMSIS5_NN)/ActivationFunctions/arm_relu6_s8.c
CSRCS += $(CMSIS5_NN)/ActivationFunctions/arm_relu_q15.c
CSRCS += $(CMSIS5_NN)/ActivationFunctions/arm_nn_activations_q7.c
CSRCS += $(CMSIS5_NN)/ActivationFunctions/arm_relu_q7.c
CSRCS += $(CMSIS5_NN)/SVDFunctions/arm_svdf_s8.c
CSRCS += $(CMSIS5_NN)/SoftmaxFunctions/arm_softmax_u8.c
CSRCS += $(CMSIS5_NN)/SoftmaxFunctions/arm_softmax_q15.c
CSRCS += $(CMSIS5_NN)/SoftmaxFunctions/arm_softmax_q7.c
CSRCS += $(CMSIS5_NN)/SoftmaxFunctions/arm_softmax_s8.c
CSRCS += $(CMSIS5_NN)/SoftmaxFunctions/arm_softmax_with_batch_q7.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_s8.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_q15_opt.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_q7_opt.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_mat_q7_vec_q15_opt.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_mat_q7_vec_q15.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_q7.c
CSRCS += $(CMSIS5_NN)/FullyConnectedFunctions/arm_fully_connected_q15.c
CSRCS += $(CMSIS5_NN)/ConcatenationFunctions/arm_concatenation_s8_y.c
CSRCS += $(CMSIS5_NN)/ConcatenationFunctions/arm_concatenation_s8_w.c
CSRCS += $(CMSIS5_NN)/ConcatenationFunctions/arm_concatenation_s8_x.c
CSRCS += $(CMSIS5_NN)/ConcatenationFunctions/arm_concatenation_s8_z.c
CSRCS += $(CMSIS5_NN)/PoolingFunctions/arm_avgpool_s8.c
CSRCS += $(CMSIS5_NN)/PoolingFunctions/arm_pool_q7_HWC.c
CSRCS += $(CMSIS5_NN)/PoolingFunctions/arm_max_pool_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q15_fast.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_wrapper_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_conv_wrapper_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_mat_mult_kernel_q7_q15.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q15_fast_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_conv_s8_opt.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q15_basic.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_mat_mult_kernel_q7_q15_reordered.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_mat_mult_kernel_s8_s16.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q7_RGB.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_conv_3x3_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_conv_u8_basic_ver1.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_separable_conv_HWC_q7.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q7_fast.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_1_x_n_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_1x1_s8_fast.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q7_basic.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_depthwise_conv_s8_core.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_separable_conv_HWC_q7_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_mat_mult_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q7_fast_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_depthwise_conv_s8.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_1x1_HWC_q7_fast_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_HWC_q7_basic_nonsquare.c
libcmsisnn: add the new CHW functionality The patch for CHW functionality has been added as a result we can compile the new functions Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-02-23 12:10:44 +01:00
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_CHW_f32_basic_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_CHW_q7_basic_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_convolve_CHW_q15_basic_nonsquare.c
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_CHW_mat_mult_kernel_q7_q15.c
Add support for CMSIS NN the CMSIS NN software library is a collection of efficient neural network kernels developed to maximize the performance and minimize the memory footprint of neural networks on Cortex-M processor cores. Project https://github.com/ARM-software/CMSIS_5 The library is divided into a number of functions each covering a specific category: Convolution Functions Activation Functions Fully-connected Layer Functions SVDF Layer Functions Pooling Functions Softmax Functions Basic math Functions The library has separate functions for operating on different weight and activation data types including 8-bit integers (q7_t) and 16-bit integers (q15_t). The descrition of the kernels are included in the function description. More information https://www.keil.com/pack/doc/CMSIS/NN/html/index.html Project license : Apache 2.0 License https://github.com/ARM-software/CMSIS_5/blob/develop/LICENSE.txt Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-01-26 17:34:58 +01:00
CSRCS += $(CMSIS5_NN)/ConvolutionFunctions/arm_nn_mat_mult_kernel_s8_s16_reordered.c
CSRCS += $(CMSIS5_NN)/BasicMathFunctions/arm_elementwise_mul_s8.c
CSRCS += $(CMSIS5_NN)/BasicMathFunctions/arm_elementwise_add_s8.c
CSRCS += $(CMSIS5_NN)/ReshapeFunctions/arm_reshape_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nntables.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_q7_to_q15_with_offset.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_q7_to_q15_reordered_no_shift.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_vec_mat_mult_t_svdf_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_vec_mat_mult_t_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_q7_to_q15_no_shift.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_mult_q15.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_depthwise_conv_nt_t_padded_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_mult_q7.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_mat_mul_core_4x_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_mat_mult_nt_t_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_accumulate_q7_to_q15.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_depthwise_conv_nt_t_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_mat_mul_core_1x_s8.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_nn_add_q7.c
CSRCS += $(CMSIS5_NN)/NNSupportFunctions/arm_q7_to_q15_reordered_with_offset.c
cmsis: build CMSIS support as a module CMSIS should compile as a module to provide the necessary support for the dnn test application Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-02-23 12:13:14 +01:00
MODULE = $(CONFIG_CMSIS_NN)
Add support for CMSIS NN the CMSIS NN software library is a collection of efficient neural network kernels developed to maximize the performance and minimize the memory footprint of neural networks on Cortex-M processor cores. Project https://github.com/ARM-software/CMSIS_5 The library is divided into a number of functions each covering a specific category: Convolution Functions Activation Functions Fully-connected Layer Functions SVDF Layer Functions Pooling Functions Softmax Functions Basic math Functions The library has separate functions for operating on different weight and activation data types including 8-bit integers (q7_t) and 16-bit integers (q15_t). The descrition of the kernels are included in the function description. More information https://www.keil.com/pack/doc/CMSIS/NN/html/index.html Project license : Apache 2.0 License https://github.com/ARM-software/CMSIS_5/blob/develop/LICENSE.txt Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2022-01-26 17:34:58 +01:00
include $(APPDIR)/Application.mk
Reference in New Issue Copy Permalink