[Top] [All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Minios-devel] [UNIKRAFT/LIBDNNL PATCH 2/2] Add Makefile.uk and patches

  • To: "minios-devel@xxxxxxxxxxxxxxxxxxxx" <minios-devel@xxxxxxxxxxxxxxxxxxxx>
  • From: Vlad-Andrei BĂDOIU (78692) <vlad_andrei.badoiu@xxxxxxxxxxxxxxx>
  • Date: Thu, 6 Feb 2020 09:27:17 +0000
  • Accept-language: en-US
  • Arc-authentication-results: i=1; mx.microsoft.com 1; spf=pass smtp.mailfrom=stud.acs.upb.ro; dmarc=pass action=none header.from=stud.acs.upb.ro; dkim=pass header.d=stud.acs.upb.ro; arc=none
  • Arc-message-signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector9901; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck; bh=AoELRPZjptzSsXi+M12vP81hD5nieLljfusaOY/Pwaw=; b=MaVhF44axUzV5kRx6p5CW1r3Ajv6dQThCNQv+pcEaHBlXX1WMo/7FV5gpoOrgUGiwJGGDCcCaS2xMUuPp/wqEfcZprOtHUTHTzeerPBVv8hf7KCxY5iR/WRd9YB+D35kBssWloYyF+wzkq7YOmJ1gdSHcIwsZu1FY+pG76CNkb+HgRV89/8cJVysUgGezKqSL8C+Q6swFM9MJsT0L+On2Me2b2rDLwxFrdDW3XCVX82GlEDEqZJ/5enf1CSLogxGI/STU8RYqVo4Z3oBitUfwMSr74n1uqgWF7JvG03C0BgKHknTqtPikKYkYLgjkWJ9sq3wMwJpP3v2bAuPj/eQRw==
  • Arc-seal: i=1; a=rsa-sha256; s=arcselector9901; d=microsoft.com; cv=none; b=O5I1WjuhpIRSvPQAO4xFAjfA/wa3o89ML123t75dtEjTYoIBRQvHHLSJ/fdgy1hrN1mBeF59U/MTggL/3bVeAIDbNuzEOlnx9K0PwwaELSseFpKe7lnlDzY7Vj395iGCYqEKAm8xov2DyTtGHI0NGnAk5A4mNUn572Dc2Rltha+fm3fOc4I+7TyE6frub9y+HDKMmghOa59GyTPGB+GVhjHRpls0qLPyNk1RU7ICW9gEoJUvj38CrpvQgXnT07E8jrZW8WjQ3fiqlIW62GASTuauJEs+F3K7j9qPIFIdQzruxWef7xg2XzIXc8zegIYShCy4Ro7vKV8iv6OVo7/S8w==
  • Authentication-results: spf=none (sender IP is ) smtp.mailfrom=vlad_andrei.badoiu@xxxxxxxxxxxxxxx;
  • Delivery-date: Thu, 06 Feb 2020 09:27:25 +0000
  • List-id: Mini-os development list <minios-devel.lists.xenproject.org>
  • Thread-index: AQHV3BJpRAnwTmYc8U2Z+6NFf5iQ7KgN5xeA
  • Thread-topic: [Minios-devel] [UNIKRAFT/LIBDNNL PATCH 2/2] Add Makefile.uk and patches
Thanks Felipe!

Reviewed-by: Vlad-Andrei Badoiu <vlad_andrei.badoiu@xxxxxxxxxxxxxxx>

On 05.02.2020 12:52, Felipe Huici wrote:
> Add dnnl port Makefile.uk and corresponding patches
>
> Signed-off-by: Felipe Huici <felipe.huici@xxxxxxxxx>
> ---
>   Makefile.uk                                   | 321 ++++++++++++++++++
>   include/cpuid.h                               | 158 +++++++++
>   include/dnnl_config.h                         |  84 +++++
>   include/dnnl_version.h                        |  36 ++
>   ...001-prevent-clobbering-barrier-macro.patch |  11 +
>   patches/0002-prevent-clobbering-B-macro.patch |  11 +
>   .../0003-prevent-clobbering-nop-macro.patch   |  11 +
>   7 files changed, 632 insertions(+)
>   create mode 100644 Makefile.uk
>   create mode 100644 include/cpuid.h
>   create mode 100644 include/dnnl_config.h
>   create mode 100644 include/dnnl_version.h
>   create mode 100644 patches/0001-prevent-clobbering-barrier-macro.patch
>   create mode 100644 patches/0002-prevent-clobbering-B-macro.patch
>   create mode 100644 patches/0003-prevent-clobbering-nop-macro.patch
>
> diff --git a/Makefile.uk b/Makefile.uk
> new file mode 100644
> index 0000000..bb25ee6
> --- /dev/null
> +++ b/Makefile.uk
> @@ -0,0 +1,321 @@
> +#  libdnnl Makefile.uk
> +#
> +#  Authors: Felipe Huici <felipe.huici@xxxxxxxxx>
> +#
> +#   Copyright (c) 2020, NEC Europe Ltd., NEC Corporation. All rights 
> reserved.
> +#
> +#  Redistribution and use in source and binary forms, with or without
> +#  modification, are permitted provided that the following conditions
> +#  are met:
> +#
> +#  1. Redistributions of source code must retain the above copyright
> +#     notice, this list of conditions and the following disclaimer.
> +#  2. Redistributions in binary form must reproduce the above copyright
> +#     notice, this list of conditions and the following disclaimer in the
> +#     documentation and/or other materials provided with the distribution.
> +#  3. Neither the name of the copyright holder nor the names of its
> +#     contributors may be used to endorse or promote products derived from
> +#     this software without specific prior written permission.
> +#
> +#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS 
> IS"
> +#  AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
> +#  IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
> +#  ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
> +#  LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
> +#  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
> +#  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
> +#  INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
> +#  CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
> +#  ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
> +#  POSSIBILITY OF SUCH DAMAGE.
> +#
> +#  THIS HEADER MAY NOT BE EXTRACTED OR MODIFIED IN ANY WAY.
> +#
> +
> +################################################################################
> +# Library registration
> +################################################################################
> +$(eval $(call addlib_s,libdnnl,$(CONFIG_LIBDNNL)))
> +
> +ifeq ($(CONFIG_LIBDNNL),y)
> +ifneq ($(CONFIG_LIBCOMPILER_RT),y)
> +$(error Require libcompiler_rt)
> +endif
> +endif
> +
> +################################################################################
> +# Sources
> +################################################################################
> +LIBDNNL_VERSION=1.2
> +LIBDNNL_URL=https://github.com/intel/mkl-dnn/archive/v$(LIBDNNL_VERSION).zip
> +LIBDNNL_PATCHDIR=$(LIBDNNL_BASE)/patches
> +$(eval $(call fetch,libdnnl,$(LIBDNNL_URL)))
> +$(eval $(call patch,libdnnl,$(LIBDNNL_PATCHDIR),mkl-dnn-$(LIBDNNL_VERSION)))
> +
> +################################################################################
> +# Helpers
> +################################################################################
> +LIBDNNL_SUBDIR=mkl-dnn-$(LIBDNNL_VERSION)
> +LIBDNNL_SRC=$(LIBDNNL_ORIGIN)/$(LIBDNNL_SUBDIR)
> +
> +################################################################################
> +# Library includes
> +################################################################################
> +LIBMKL-INCLUDES += -I$(LIBDNNL_SRC)/include      \
> +                   -I$(LIBDNNL_SRC)/src          \
> +                   -I$(LIBDNNL_SRC)/src/cpu      \
> +                   -I$(LIBDNNL_SRC)/src/cpu/gemm \
> +                   -I$(LIBDNNL_SRC)/src/common   \
> +                   -I$(LIBDNNL_BASE)/include     \
> +
> +CINCLUDES-$(CONFIG_LIBDNNL) += $(LIBMKL-INCLUDES)
> +CXXINCLUDES-$(CONFIG_LIBDNNL) += $(LIBMKL-INCLUDES)
> +
> +################################################################################
> +# Global flags
> +################################################################################
> +LIBDNNL-CONFIG_FLAGS = -DCMAKE_BUILD_TYPE=Release        \
> +                       -DDNNL_ENABLE_CONCURRENT_EXEC=OFF \
> +                       -DDNNL_LIBRARY_TYPE=STATIC     \
> +                       -DDNNL_THREADING=OMP:COMP      \
> +                       -DDNNL_USE_MKL=NONE            \
> +                       -DDNNL_VERBOSE=ON              \
> +                       -DWITH_EXAMPLE=OFF             \
> +                       -DWITH_TEST=OFF                        \
> +                    -DDNNL_ENABLE_JIT_PROFILING=OFF
> +
> +LIBDNNL_CFLAGS-y      +=  $(LIBDNNL-CONFIG_FLAGS)
> +LIBDNNL_CXXFLAGS-y    +=  $(LIBDNNL-CONFIG_FLAGS)
> +
> +################################################################################
> +# Suppress Flags
> +################################################################################
> +LIBDNNL_SUPPRESS_FLAGS-y += -Wno-unused-parameter              \
> +                            -Wno-unused-but-set-parameter      \
> +                            -Wno-unknown-pragmas               \
> +                         -Wno-extra                         \
> +                         -Wno-attributes                    \
> +                         -Wno-parentheses                   \
> +                         -fcompare-debug-second  # supress notes
> +
> +LIBDNNL_CFLAGS-y   += $(LIBDNNL_SUPPRESS_FLAGS-y) 
> -Wno-implicit-function-declaration
> +LIBDNNL_CXXFLAGS-y += $(LIBDNNL_SUPPRESS_FLAGS-y)
> +
> +################################################################################
> +# Sources
> +################################################################################
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/dnnl_debug.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/scratchpad.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/batch_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/binary.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/lrn.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/shuffle.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/primitive_exec_types.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/memory.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/pooling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/inner_product.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/verbose.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/utils.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/primitive_attr.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/primitive_iterator.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/engine.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/query.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/reorder.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/sum.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/convolution_pd.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/memory_desc_wrapper.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/primitive.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/rnn.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/matmul.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/resampling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/layer_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/concat.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/primitive_hashing.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/dnnl_debug_autogenerated.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/deconvolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/softmax.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/stream.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/primitive_desc.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/eltwise.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/common/memory_zero_pad.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_tbb_batch_normalization.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_f32_wino_conv_4x3.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/nchw_pooling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_dw_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_batch_normalization_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_batch_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_eltwise.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_shuffle.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_reorder.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_reorder_utils.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx2_convolution.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_x8s8s32x_1x1_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_concat.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_batch_normalization_utils.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_x8s8s32x_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_batch_normalization_s8.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_layer_normalization_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_reducer.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_eltwise.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/nspc_batch_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_lrn.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_convolution.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_x8s8s32x_1x1_convolution.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_f32_wino_conv_2x3.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_x8s8s32x_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_convolution_utils.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_eltwise_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_layer_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_sse41_1x1_conv_kernel_f32.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_lrn.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_conv_winograd_kernel_f32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_bf16_dw_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_sse41_conv_kernel_f32.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_bf16_convolution.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_x8s8s32x_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_bf16_sum.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_shuffle_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_eltwise_injector.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_pooling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_deconvolution_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_batch_normalization.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_x8s8s32x_deconvolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_barrier.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_bf16_inner_product.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx2_x8s8s32x_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_x8s8s32x_inner_product.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/bfloat16.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_bf16_conv_kernel.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_f32_wino_conv_4x3_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_bf16_convolution.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_bf16_1x1_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/simple_concat.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx2_1x1_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_sse41_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_convolution_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_softmax_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_transpose_src_utils.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_bf16_1x1_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/nhwc_pooling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_pooling_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_reorder.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx2_x8s8s32x_1x1_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_inner_product_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/simple_sum.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_sse41_1x1_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_i8i8_pooling.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_convolution_winograd.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_core_u8s8s32x_wino_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_inner_product.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_lrn_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ncsp_batch_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx2_conv_kernel_f32.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_softmax.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_softmax.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_pooling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_dw_conv_kernel_f32.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx2_1x1_conv_kernel_f32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_1x1_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_inner_product_utils.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx2_x8s8s32x_1x1_conv_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_layer_normalization.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx2_x8s8s32x_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_convolution.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_avx512_common_1x1_convolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_isa_traits.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_deconvolution.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_sum.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/cpu_engine.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_pool_kernel.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm_inner_product.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/ref_lrn.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_uni_lrn_kernel_f32.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/resampling/simple_resampling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/resampling/ref_resampling.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/resampling/cpu_resampling_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/binary/cpu_binary_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/binary/jit_uni_binary.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/binary/ref_binary.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/binary/jit_uni_i8i8_binary.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/matmul/gemm_x8s8s32x_matmul.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/matmul/cpu_matmul_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/matmul/gemm_bf16_matmul.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/matmul/gemm_f32_matmul.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/matmul/ref_matmul.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/rnn_utils.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/cell_gru_lbr.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/ref_rnn.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/ref_postgemm_rnn.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/cpu_rnn_list.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/ref_postgemm_lstm.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/ref_postgemm_gru_lbr.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/ref_postgemm_gru.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/cell_gru.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/rnn/cell_common.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_utils/jit_utils.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/jit_utils/jitprofiling/jitprofiling.c
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/jit_utils/linux_perf/linux_perf.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/gemm_info.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/gemm_pack.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/gemm_driver.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/gemm.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/gemv_driver.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_gemm_s8u8s32_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_kernel_gemv_s8x8s32_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_sum_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_sum_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_sum_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_sum_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_sum_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_gemv_s8x8s32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_sum_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_sum_bn_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_gemm_s8u8s32_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx2_u8_copy_sum_bn_kern.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/ref_gemm_s8x8s32.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/simple_gemm_s8s8s32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/s8x8s32/jit_avx512_core_u8_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/f32/gemm_utils_f32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx512_core_f32_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx2_f32_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx2_f32_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_f32_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_f32_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_f32_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx2_kernel_sgemm_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx512_core_f32_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_kernel_sgemm_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_f32_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx2_f32_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_gemv_t_f32_kern.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_gemv_t_f32_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_f32_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_f32_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_f32_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx512_core_f32_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_gemm_f32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx512_common_gemm_f32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx2_f32_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_kernel_b0_sgemm_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_kernel_b0_sgemm_kern.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/cpu/gemm/f32/ref_gemm_f32.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_sse41_f32_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx512_core_f32_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/f32/jit_avx_kernel_sgemm_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/bf16/jit_avx512_core_s16_copy_an_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/bf16/jit_avx512_core_s16_copy_bt_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/bf16/jit_avx512_core_s16_copy_at_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/bf16/jit_avx512_core_gemm_bf16bf16f32_kern.cpp
> +LIBDNNL_SRCS-y += 
> $(LIBDNNL_SRC)/src/cpu/gemm/bf16/jit_avx512_core_s16_copy_bn_kern.cpp
> +LIBDNNL_SRCS-y += $(LIBDNNL_SRC)/src/compute/dispatch.cpp
> +
> diff --git a/include/cpuid.h b/include/cpuid.h
> new file mode 100644
> index 0000000..cb75258
> --- /dev/null
> +++ b/include/cpuid.h
> @@ -0,0 +1,158 @@
> +/* Taken from FreeBSD */
> +
> +/*===---- cpuid.h - X86 cpu model detection 
> --------------------------------===
> + *
> + * Permission is hereby granted, free of charge, to any person obtaining a 
> copy
> + * of this software and associated documentation files (the "Software"), to 
> deal
> + * in the Software without restriction, including without limitation the 
> rights
> + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
> + * copies of the Software, and to permit persons to whom the Software is
> + * furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice shall be included in
> + * all copies or substantial portions of the Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 
> THE
> + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 
> FROM,
> + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
> + * THE SOFTWARE.
> + *
> + 
> *===-----------------------------------------------------------------------===
> + */
> +
> +#if !(__x86_64__ || __i386__)
> +#error this header is for x86 only
> +#endif
> +
> +/* Features in %ecx for level 1 */
> +#define bit_SSE3        0x00000001
> +#define bit_PCLMULQDQ   0x00000002
> +#define bit_DTES64      0x00000004
> +#define bit_MONITOR     0x00000008
> +#define bit_DSCPL       0x00000010
> +#define bit_VMX         0x00000020
> +#define bit_SMX         0x00000040
> +#define bit_EIST        0x00000080
> +#define bit_TM2         0x00000100
> +#define bit_SSSE3       0x00000200
> +#define bit_CNXTID      0x00000400
> +#define bit_FMA         0x00001000
> +#define bit_CMPXCHG16B  0x00002000
> +#define bit_xTPR        0x00004000
> +#define bit_PDCM        0x00008000
> +#define bit_PCID        0x00020000
> +#define bit_DCA         0x00040000
> +#define bit_SSE41       0x00080000
> +#define bit_SSE42       0x00100000
> +#define bit_x2APIC      0x00200000
> +#define bit_MOVBE       0x00400000
> +#define bit_POPCNT      0x00800000
> +#define bit_TSCDeadline 0x01000000
> +#define bit_AESNI       0x02000000
> +#define bit_XSAVE       0x04000000
> +#define bit_OSXSAVE     0x08000000
> +#define bit_AVX         0x10000000
> +#define bit_RDRAND      0x40000000
> +
> +/* Features in %edx for level 1 */
> +#define bit_FPU         0x00000001
> +#define bit_VME         0x00000002
> +#define bit_DE          0x00000004
> +#define bit_PSE         0x00000008
> +#define bit_TSC         0x00000010
> +#define bit_MSR         0x00000020
> +#define bit_PAE         0x00000040
> +#define bit_MCE         0x00000080
> +#define bit_CX8         0x00000100
> +#define bit_APIC        0x00000200
> +#define bit_SEP         0x00000800
> +#define bit_MTRR        0x00001000
> +#define bit_PGE         0x00002000
> +#define bit_MCA         0x00004000
> +#define bit_CMOV        0x00008000
> +#define bit_PAT         0x00010000
> +#define bit_PSE36       0x00020000
> +#define bit_PSN         0x00040000
> +#define bit_CLFSH       0x00080000
> +#define bit_DS          0x00200000
> +#define bit_ACPI        0x00400000
> +#define bit_MMX         0x00800000
> +#define bit_FXSR        0x01000000
> +#define bit_SSE         0x02000000
> +#define bit_SSE2        0x04000000
> +#define bit_SS          0x08000000
> +#define bit_HTT         0x10000000
> +#define bit_TM          0x20000000
> +#define bit_PBE         0x80000000
> +
> +/* Features in %ebx for level 7 sub-leaf 0 */
> +#define bit_FSGSBASE    0x00000001
> +#define bit_SMEP        0x00000080
> +#define bit_ENH_MOVSB   0x00000200
> +
> +/* PIC on i386 uses %ebx, so preserve it. */
> +#if __i386__
> +#define __cpuid(__level, __eax, __ebx, __ecx, __edx) \
> +    __asm("  pushl  %%ebx\n" \
> +          "  cpuid\n" \
> +          "  mov    %%ebx,%1\n" \
> +          "  popl   %%ebx" \
> +        : "=a"(__eax), "=r" (__ebx), "=c"(__ecx), "=d"(__edx) \
> +        : "0"(__level))
> +
> +#define __cpuid_count(__level, __count, __eax, __ebx, __ecx, __edx) \
> +    __asm("  pushl  %%ebx\n" \
> +          "  cpuid\n" \
> +          "  mov    %%ebx,%1\n" \
> +          "  popl   %%ebx" \
> +        : "=a"(__eax), "=r" (__ebx), "=c"(__ecx), "=d"(__edx) \
> +        : "0"(__level), "2"(__count))
> +#else
> +#define __cpuid(__level, __eax, __ebx, __ecx, __edx) \
> +    __asm("cpuid" : "=a"(__eax), "=b" (__ebx), "=c"(__ecx), "=d"(__edx) \
> +                  : "0"(__level))
> +
> +#define __cpuid_count(__level, __count, __eax, __ebx, __ecx, __edx) \
> +    __asm("cpuid" : "=a"(__eax), "=b" (__ebx), "=c"(__ecx), "=d"(__edx) \
> +                  : "0"(__level), "2"(__count))
> +#endif
> +
> +static __inline int __get_cpuid (unsigned int __level, unsigned int *__eax,
> +                                 unsigned int *__ebx, unsigned int *__ecx,
> +                                 unsigned int *__edx) {
> +    __cpuid(__level, *__eax, *__ebx, *__ecx, *__edx);
> +    return 1;
> +}
> +
> +static __inline int __get_cpuid_max (unsigned int __level, unsigned int 
> *__sig)
> +{
> +    unsigned int __eax, __ebx, __ecx, __edx;
> +#if __i386__
> +    int __cpuid_supported;
> +
> +    __asm("  pushfl\n"
> +          "  popl   %%eax\n"
> +          "  movl   %%eax,%%ecx\n"
> +          "  xorl   $0x00200000,%%eax\n"
> +          "  pushl  %%eax\n"
> +          "  popfl\n"
> +          "  pushfl\n"
> +          "  popl   %%eax\n"
> +          "  movl   $0,%0\n"
> +          "  cmpl   %%eax,%%ecx\n"
> +          "  je     1f\n"
> +          "  movl   $1,%0\n"
> +          "1:"
> +        : "=r" (__cpuid_supported) : : "eax", "ecx");
> +    if (!__cpuid_supported)
> +        return 0;
> +#endif
> +
> +    __cpuid(__level, __eax, __ebx, __ecx, __edx);
> +    if (__sig)
> +        *__sig = __ebx;
> +    return __eax;
> +}
> diff --git a/include/dnnl_config.h b/include/dnnl_config.h
> new file mode 100644
> index 0000000..a4d58b0
> --- /dev/null
> +++ b/include/dnnl_config.h
> @@ -0,0 +1,84 @@
> +/*******************************************************************************
> +* Copyright 2019 Intel Corporation
> +*
> +* Licensed 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.
> +*******************************************************************************/
> +
> +#ifndef DNNL_CONFIG_H
> +#define DNNL_CONFIG_H
> +
> +#include "dnnl_types.h"
> +
> +/// @cond DO_NOT_DOCUMENT_THIS
> +
> +// All symbols shall be internal unless marked as DNNL_API
> +#if defined _WIN32 || defined __CYGWIN__
> +#define DNNL_HELPER_DLL_IMPORT __declspec(dllimport)
> +#define DNNL_HELPER_DLL_EXPORT __declspec(dllexport)
> +#else
> +#if __GNUC__ >= 4
> +#define DNNL_HELPER_DLL_IMPORT __attribute__((visibility("default")))
> +#define DNNL_HELPER_DLL_EXPORT __attribute__((visibility("default")))
> +#else
> +#define DNNL_HELPER_DLL_IMPORT
> +#define DNNL_HELPER_DLL_EXPORT
> +#endif
> +#endif
> +
> +#ifdef DNNL_DLL
> +#ifdef DNNL_DLL_EXPORTS
> +#define DNNL_API DNNL_HELPER_DLL_EXPORT
> +#else
> +#define DNNL_API DNNL_HELPER_DLL_IMPORT
> +#endif
> +#else
> +#define DNNL_API
> +#endif
> +
> +#if defined(__GNUC__)
> +#define DNNL_DEPRECATED __attribute__((deprecated))
> +#elif defined(_MSC_VER)
> +#define DNNL_DEPRECATED __declspec(deprecated)
> +#else
> +#define DNNL_DEPRECATED
> +#endif
> +
> +/// @endcond
> +
> +// clang-format off
> +
> +// DNNL CPU threading runtime
> +#define DNNL_CPU_THREADING_RUNTIME DNNL_RUNTIME_SEQ
> +
> +// DNNL CPU engine runtime
> +#define DNNL_CPU_RUNTIME DNNL_RUNTIME_SEQ
> +
> +// DNNL GPU engine runtime
> +#define DNNL_GPU_RUNTIME DNNL_RUNTIME_NONE
> +
> +// clang-format on
> +
> +#if defined(DNNL_CPU_RUNTIME) && defined(DNNL_GPU_RUNTIME)
> +#if (DNNL_CPU_RUNTIME == DNNL_RUNTIME_NONE) \
> +        || (DNNL_CPU_RUNTIME == DNNL_RUNTIME_OCL)
> +#error "Unexpected DNNL_CPU_RUNTIME"
> +#endif
> +#if (DNNL_GPU_RUNTIME != DNNL_RUNTIME_NONE) \
> +        && (DNNL_GPU_RUNTIME != DNNL_RUNTIME_OCL)
> +#error "Unexpected DNNL_GPU_RUNTIME"
> +#endif
> +#else
> +#error "BOTH DNNL_CPU_RUNTIME and DNNL_GPU_RUNTIME must be defined"
> +#endif
> +
> +#endif
> diff --git a/include/dnnl_version.h b/include/dnnl_version.h
> new file mode 100644
> index 0000000..57715b8
> --- /dev/null
> +++ b/include/dnnl_version.h
> @@ -0,0 +1,36 @@
> +/*******************************************************************************
> +* Copyright 2019 Intel Corporation
> +*
> +* Licensed 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.
> +*******************************************************************************/
> +
> +#ifndef DNNL_VERSION_H
> +#define DNNL_VERSION_H
> +
> +// clang-format off
> +
> +/// Major version
> +#define DNNL_VERSION_MAJOR 1
> +
> +/// Minor version
> +#define DNNL_VERSION_MINOR 2
> +
> +/// Patch version
> +#define DNNL_VERSION_PATCH 0
> +
> +/// Git commit hash
> +#define DNNL_VERSION_HASH  "N/A"
> +
> +// clang-format on
> +
> +#endif
> diff --git a/patches/0001-prevent-clobbering-barrier-macro.patch 
> b/patches/0001-prevent-clobbering-barrier-macro.patch
> new file mode 100644
> index 0000000..5a44492
> --- /dev/null
> +++ b/patches/0001-prevent-clobbering-barrier-macro.patch
> @@ -0,0 +1,11 @@
> +--- a/src/cpu/cpu_barrier.hpp        2020-02-01 22:15:52.999902062 +0100
> ++++ b/src/cpu/cpu_barrier.hpp        2020-02-01 22:16:03.487787545 +0100
> +@@ -22,6 +22,8 @@
> + #include "jit_generator.hpp"
> + #include "utils.hpp"
> +
> ++#undef barrier
> ++
> + namespace dnnl {
> + namespace impl {
> + namespace cpu {
> diff --git a/patches/0002-prevent-clobbering-B-macro.patch 
> b/patches/0002-prevent-clobbering-B-macro.patch
> new file mode 100644
> index 0000000..0ae33b0
> --- /dev/null
> +++ b/patches/0002-prevent-clobbering-B-macro.patch
> @@ -0,0 +1,11 @@
> +--- a/src/common/tag_traits.hpp      2020-02-02 18:20:18.349160812 +0100
> ++++ b/src/common/tag_traits.hpp      2020-02-02 18:20:32.421006289 +0100
> +@@ -22,6 +22,8 @@
> + #include "c_types_map.hpp"
> + #include "utils.hpp"
> +
> ++#undef _B
> ++
> + namespace dnnl {
> + namespace impl {
> +
> diff --git a/patches/0003-prevent-clobbering-nop-macro.patch 
> b/patches/0003-prevent-clobbering-nop-macro.patch
> new file mode 100644
> index 0000000..27f7267
> --- /dev/null
> +++ b/patches/0003-prevent-clobbering-nop-macro.patch
> @@ -0,0 +1,11 @@
> +--- a/src/cpu/xbyak/xbyak.h  2020-02-04 15:12:32.586176051 +0100
> ++++ b/src/cpu/xbyak/xbyak.h  2020-02-04 15:12:42.758058038 +0100
> +@@ -72,6 +72,8 @@
> + #include <iostream>
> + #endif
> +
> ++#undef nop
> ++
> + // #define XBYAK_DISABLE_AVX512
> +
> + //#define XBYAK_USE_MMAP_ALLOCATOR
_______________________________________________
Minios-devel mailing list
Minios-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/minios-devel

 

©2013 Xen Project, A Linux Foundation Collaborative Project. All Rights Reserved.
Linux Foundation is a registered trademark of The Linux Foundation.
Xen Project is a trademark of The Linux Foundation.

Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.