--- /dev/null
--- /dev/null
++volk (1.3-3) unstable; urgency=medium
++
++ * update to v1.3-23-g0109b2e
++ * update debian/libvolk1-dev.abi.tar.gz.amd64
++ * Add breaks/replaces gnuradio (<=3.7.2.1) (LP: #1614235)
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 04 Feb 2018 13:12:21 -0500
++
++volk (1.3-2) unstable; urgency=medium
++
++ * update to v1.3-16-g28b03a9
++ apps: fix profile update reading end of lines
++ qa: lower tolerance for 32fc_mag to fix issue #96
++ * include upstream master patch to sort input files
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 27 Aug 2017 13:44:55 -0400
++
++volk (1.3-1) unstable; urgency=medium
++
++ * New upstream release
++ * The index_max kernels were named with the wrong output datatype. To
++ fix this there are new kernels that return a 32u (int32_t) and the
++ existing kernels had their signatures changed to return 16u (int16_t).
++ * The output to stdout and stderr has been shuffled around. There is no
++ longer a message that prints what VOLK machine is being used and the
++ warning messages go to stderr rather than stdout.
++ * The 32fc_index_max kernels previously were only accurate to the SSE
++ register width (4 points). This was a pretty serious and long-lived
++ bug that's been fixed and the QA updated appropriately.
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 02 Jul 2016 16:30:47 -0400
++
++volk (1.2.2-2) unstable; urgency=medium
++
++ * update to v1.2.2-11-g78c8bc4 (to follow gnuradio maint branch)
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 19 Jun 2016 14:44:15 -0400
++
++volk (1.2.2-1) unstable; urgency=medium
++
++ * New upstream release
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Fri, 08 Apr 2016 00:12:10 -0400
++
++volk (1.2.1-2) unstable; urgency=medium
++
++ * Upstream patches:
++ Fix some CMake complaints
++ The fix for compilation with cmake 3.5
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Wed, 23 Mar 2016 17:47:54 -0400
++
++volk (1.2.1-1) unstable; urgency=medium
++
++ * New upstream release
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 07 Feb 2016 19:38:32 -0500
++
++volk (1.2-1) unstable; urgency=medium
++
++ * New upstream release
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 24 Dec 2015 20:28:13 -0500
++
++volk (1.1.1-5) experimental; urgency=medium
++
++ * update to v1.1.1-22-gef53547 to support gnuradio 3.7.9
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Fri, 11 Dec 2015 13:12:55 -0500
++
++volk (1.1.1-4) unstable; urgency=medium
++
++ * more lintian fixes
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Wed, 25 Nov 2015 21:49:58 -0500
++
++volk (1.1.1-3) unstable; urgency=medium
++
++ * Lintian fixes Pre-Depends
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 19 Nov 2015 21:24:27 -0500
++
++volk (1.1.1-2) unstable; urgency=medium
++
++ * Note that libvolk1-dev replaces files in gnuradio-dev versions <<3.7.8
++ (Closes: #802646) again. Thanks Andreas Beckmann.
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Fri, 13 Nov 2015 18:45:49 -0500
++
++volk (1.1.1-1) unstable; urgency=medium
++
++ * New upstream release
++ * New architectures exist for the AVX2 and FMA ISAs.
++ * The profiler now generates buffers that are vlen + a tiny amount and
++ generates random data to fill buffers. This is intended to catch bugs
++ in protokernels that write beyond num_points.
++ * Note that libvolk1-dev replaces files in earlier gnuradio-dev versions
++ (Closes: #802646)
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 01 Nov 2015 18:45:43 -0500
++
++volk (1.1-4) unstable; urgency=medium
++
++ * update to v1.1-12-g264addc
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 29 Sep 2015 23:41:50 -0400
++
++volk (1.1-3) unstable; urgency=low
++
++ * drop dh_acc to get reproducible builds
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Fri, 11 Sep 2015 22:57:06 -0400
++
++volk (1.1-2) unstable; urgency=low
++
++ * use dh-acc
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Mon, 07 Sep 2015 15:45:20 -0400
++
++volk (1.1-1) unstable; urgency=medium
++
++ * re-organize package naming convention
++ * New upstream release tag v1.1
++ New architectures exist for the AVX2 and FMA ISAs. Along
++ with the build-system support the following kernels have
++ no proto-kernels taking advantage of these architectures:
++
++ * 32f_x2_dot_prod_32f
++ * 32fc_x2_multiply_32fc
++ * 64_byteswap
++ * 32f_binary_slicer_8i
++ * 16u_byteswap
++ * 32u_byteswap
++
++ QA/profiler
++ -----------
++
++ The profiler now generates buffers that are vlen + a tiny
++ amount and generates random data to fill buffers. This is
++ intended to catch bugs in protokernels that write beyond
++ num_points.
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Wed, 26 Aug 2015 09:22:48 -0400
++
++volk (1.0.2-2) unstable; urgency=low
++
++ * Use SOURCE_DATE_EPOCH from the environment, if defined,
++ rather than current date and time to implement volk_build_date()
++ (embedding build date in a library does not help reproducible builds)
++ * add watch file
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 15 Aug 2015 17:43:15 -0400
++
++volk (1.0.2-1) unstable; urgency=medium
++
++ * Maintenance release 24 Jul 2015 by Nathan West
++ * The major change is the CMake logic to add ASM protokernels. Rather
++ than depending on CFLAGS and ASMFLAGS we use the results of VOLK's
++ built in has_ARCH tests. All configurations should work the same as
++ before, but manually specifying CFLAGS and ASMFLAGS on the cmake call
++ for ARM native builds should no longer be necessary.
++ * The 32fc_s32fc_x2_rotator_32fc generic protokernel now includes a
++ previously implied header.
++ * Finally, there is a fix to return the "best" protokernel to the
++ dispatcher when no volk_config exists. Thanks to Alexandre Raymond for
++ pointing this out.
++ * with maint branch patch:
++ kernels-add-missing-include-arm_neon.h
++ * removed unused build-dependency on liboil0.3-dev (closes: #793626)
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Wed, 05 Aug 2015 00:43:40 -0400
++
++volk (1.0.1-1) unstable; urgency=low
++
++ * Maintenance Release v1.0.1 08 Jul 2015 by Nathan West
++ This is a maintenance release with bug fixes since the initial release of
++ v1.0 in April.
++
++ * Contributors
++
++ The following authors have contributed code to this release:
++
++ Doug Geiger doug.geiger@bioradiation.net
++ Elliot Briggs elliot.briggs@gmail.com
++ Marcus Mueller marcus@hostalia.de
++ Nathan West nathan.west@okstate.edu
++ Tom Rondeau tom@trondeau.com
++
++ * Kernels
++
++ Several bug fixes in different kernels. The NEON implementations of the
++ following kernels have been fixed:
++
++ 32f_x2_add_32f
++ 32f_x2_dot_prod_32f
++ 32fc_s32fc_multiply_32fc
++ 32fc_x2_multiply_32fc
++
++ Additionally the NEON asm based 32f_x2_add_32f protokernels were not being
++ used and are now included and available for use via the dispatcher.
++
++ The 32f_s32f_x2_fm_detect_32f kernel now has a puppet. This solves QA seg
++ faults on 32-bit machines and provide a better test for this kernel.
++
++ The 32fc_s32fc_x2_rotator_32fc generic protokernel replaced cabsf with
++ hypotf for better Android support.
++
++ * Building
++
++ Static builds now trigger the applications (volk_profile and
++ volk-config-info) to be statically linked.
++
++ The file gcc_x86_cpuid.h has been removed since it was no longer being
++ used. Previously it provided cpuid functionality for ancient compilers
++ that we do not support.
++
++ All build types now use -Wall.
++
++ * QA and Testing
++
++ The documentation around the --update option to volk_profile now makes it
++ clear that the option will only profile kernels without entries in
++ volk_profile. The signature of run_volk_tests with expanded args changed
++ signed types to unsigned types to reflect the actual input.
++
++ The remaining changes are all non-functional changes to address issues
++ from Coverity.
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Fri, 10 Jul 2015 17:57:42 -0400
++
++volk (1.0-5) unstable; urgency=medium
++
++ * native-armv7-build-support skips neon on Debian armel (Closes: #789972)
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 04 Jul 2015 12:36:36 -0400
++
++volk (1.0-4) unstable; urgency=low
++
++ * update native-armv7-build-support patch from gnuradio volk package
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 25 Jun 2015 16:38:49 -0400
++
++volk (1.0-3) unstable; urgency=medium
++
++ * Add Breaks/Replaces (Closes: #789893, #789894)
++ * Allow failing tests
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 25 Jun 2015 12:46:06 -0400
++
++volk (1.0-2) unstable; urgency=medium
++
++ * kernels-add-missing-math.h-include-to-rotator
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Wed, 24 Jun 2015 21:09:32 -0400
++
++volk (1.0-1) unstable; urgency=low
++
++ * Initial package (Closes: #782417)
++ Initial Release 11 Apr 2015 by Nathan West
++
++ VOLK 1.0 is available. This is the first release of VOLK as an independently
++ tracked sub-project of GNU Radio.
++
++ * Contributors
++
++ VOLK has been tracked separately from GNU Radio since 2014 Dec 23.
++ Contributors between the split and the initial release are
++
++ Albert Holguin aholguin_77@yahoo.com
++ Doug Geiger doug.geiger@bioradiation.net
++ Elliot Briggs elliot.briggs@gmail.com
++ Julien Olivain julien.olivain@lsv.ens-cachan.fr
++ Michael Dickens michael.dickens@ettus.com
++ Nathan West nathan.west@okstate.edu
++ Tom Rondeau tom@trondeau.com
++
++ * QA
++
++ The test and profiler have significantly changed. The profiler supports
++ run-time changes to vlen and iters to help kernel development and provide
++ more flexibility on embedded systems. Additionally there is a new option
++ to update an existing volk_profile results file with only new kernels which
++ will save time when updating to newer versions of VOLK
++
++ The QA system creates a static list of kernels and test cases. The QA
++ testing and profiler iterate over this static list rather than each source
++ file keeping its own list. The QA also emits XML results to
++ lib/.unittest/kernels.xml which is formatted similarly to JUnit results.
++
++ * Modtool
++
++ Modtool was updated to support the QA and profiler changes.
++
++ * Kernels
++
++ New proto-kernels:
++
++ 16ic_deinterleave_real_8i_neon
++ 16ic_s32f_deinterleave_32f_neon
++ fix preprocessor errors for some compilers on byteswap and popcount puppets
++
++ ORC was moved to the asm kernels directory.
++ volk_malloc
++
++ The posix_memalign implementation of Volk_malloc now falls back to a standard
++ malloc if alignment is 1.
++
++ * Miscellaneous
++
++ Several build system and cmake changes have made it possible to build VOLK
++ both independently with proper soname versions and in-tree for projects
++ such as GNU Radio.
++
++ The static builds take advantage of cmake object libraries to speed up builds.
++
++ Finally, there are a number of changes to satisfy compiler warnings and make
++ QA work on multiple machines.
++
++ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 12 Apr 2015 23:20:41 -0400
--- /dev/null
--- /dev/null
++9
--- /dev/null
--- /dev/null
++Source: volk
++Section: libdevel
++Priority: extra
++Maintainer: A. Maitland Bottoms <bottoms@debian.org>
++Build-Depends: cmake,
++ debhelper (>= 9.0.0~),
++ dh-python,
++ doxygen,
++ libboost-filesystem-dev,
++ libboost-program-options-dev,
++ libboost-system-dev,
++ libboost-test-dev,
++ liborc-0.4-dev,
++ pkg-config,
++ python,
++ python-cheetah
++Standards-Version: 4.1.3
++Homepage: http://libvolk.org
++Vcs-Browser: https://salsa.debian.org/bottoms/pkg-volk
++Vcs-Git: https://salsa.debian.org/bottoms/pkg-volk.git
++
++Package: libvolk1.3
++Section: libs
++Architecture: any
++Pre-Depends: ${misc:Pre-Depends}
++Depends: ${misc:Depends}, ${shlibs:Depends}
++Multi-Arch: same
++Recommends: libvolk1-bin
++Suggests: libvolk1-dev
++Description: vector optimized functions
++ Vector-Optimized Library of Kernels is designed to help
++ applications work with the processor's SIMD instruction sets. These are
++ very powerful vector operations that can give signal processing a
++ huge boost in performance.
++
++Package: libvolk1-dev
++Architecture: any
++Pre-Depends: ${misc:Pre-Depends}
++Depends: libvolk1.3 (=${binary:Version}), ${misc:Depends}
++Breaks: gnuradio-dev (<<3.7.8), libvolk-dev, libvolk1.0-dev
++Replaces: gnuradio-dev (<<3.7.8), libvolk-dev, libvolk1.0-dev
++Multi-Arch: same
++Description: vector optimized function headers
++ Vector-Optimized Library of Kernels is designed to help
++ applications work with the processor's SIMD instruction sets. These are
++ very powerful vector operations that can give signal processing a
++ huge boost in performance.
++ .
++ This package contains the header files.
++ For documentation, see libvolk-doc.
++
++Package: libvolk1-bin
++Section: libs
++Architecture: any
++Pre-Depends: ${misc:Pre-Depends}
++Depends: libvolk1.3 (=${binary:Version}),
++ ${misc:Depends},
++ ${python:Depends},
++ ${shlibs:Depends}
++Breaks: libvolk-bin, libvolk1.0-bin, gnuradio (<=3.7.2.1)
++Replaces: libvolk-bin, libvolk1.0-bin, gnuradio (<=3.7.2.1)
++Description: vector optimized runtime tools
++ Vector-Optimized Library of Kernels is designed to help
++ applications work with the processor's SIMD instruction sets. These are
++ very powerful vector operations that can give signal processing a
++ huge boost in performance.
++ .
++ This package includes the volk_profile tool.
--- /dev/null
--- /dev/null
++Format: https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/
++Upstream-Name: volk
++Upstream-Contact: http://libvolk.org/
++Source:
++ https://github.com/gnuradio/volk
++Comment:
++ Debian packages by A. Maitland Bottoms <bottoms@debian.org>
++ .
++ Upstream Authors:
++ Albert Holguin <aholguin_77@yahoo.com>
++ Doug Geiger <doug.geiger@bioradiation.net>
++ Elliot Briggs <elliot.briggs@gmail.com>
++ Julien Olivain <julien.olivain@lsv.ens-cachan.fr>
++ Michael Dickens <michael.dickens@ettus.com>
++ Nathan West <nathan.west@okstate.edu>
++ Tom Rondeau <tom@trondeau.com>
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: *
++Copyright: 2006, 2009-2016, Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: Doxyfile.in
++ DoxygenLayout.xml
++ volk.pc.in
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: apps/volk_profile.h
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: appveyor.yml
++Copyright: 2016 Paul Cercueil <paul.cercueil@analog.com>
++License: GPL-3+
++
++Files: cmake/*
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: cmake/Modules/*
++Copyright: 2006, 2009-2016, Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: cmake/Modules/CMakeParseArgumentsCopy.cmake
++Copyright: 2010 Alexander Neundorf <neundorf@kde.org>
++License: Kitware-BSD
++ All rights reserved.
++ .
++ Redistribution and use in source and binary forms, with or without
++ modification, are permitted provided that the following conditions
++ are met:
++ .
++ * Redistributions of source code must retain the above copyright
++ notice, this list of conditions and the following disclaimer.
++ .
++ * 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.
++ .
++ * Neither the names of Kitware, Inc., the Insight Software Consortium,
++ nor the names of their 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.
++
++Files: cmake/Modules/FindORC.cmake
++ cmake/Modules/VolkConfig.cmake.in
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: cmake/msvc/*
++Copyright: 2006-2008, Alexander Chemeris
++License: BSD-2-clause
++ 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. The name of the author may be used to endorse or promote products
++ derived from this software without specific prior written permission.
++ .
++ THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
++
++Files: cmake/msvc/config.h
++Copyright: 2005, 2006 Apple Computer, Inc.
++License: LGPL-2+
++
++Files: cmake/msvc/stdbool.h
++Copyright: 2005, 2006, Apple Computer, Inc.
++License: LGPL-2+
++
++Files: debian/*
++Copyright: 2015 Free Software Foundation, Inc
++License: GPL-3+
++Comment: assigned by A. Maitland Bottoms <bottoms@debian.org>
++
++Files: debian/libvolk1-dev.abi.tar.gz.amd64
++Copyright: 2016 Free Software Foundation, Inc
++License: GPL-3+
++
++Files: docs/*
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: gen/archs.xml
++ gen/machines.xml
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: include/volk/volk_common.h
++ include/volk/volk_complex.h
++ include/volk/volk_prefs.h
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: kernels/volk/asm/*
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: kernels/volk/volk_16u_byteswappuppet_16u.h
++ kernels/volk/volk_32u_byteswappuppet_32u.h
++ kernels/volk/volk_64u_byteswappuppet_64u.h
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++Files: lib/kernel_tests.h
++ lib/qa_utils.cc
++ lib/qa_utils.h
++ lib/volk_prefs.c
++Copyright: 2014-2015 Free Software Foundation, Inc.
++License: GPL-3+
++
++License: LGPL-2+
++ This library is free software; you can redistribute it and/or
++ modify it under the terms of the GNU Library General Public
++ License as published by the Free Software Foundation; either
++ version 2 of the License, or (at your option) any later version.
++ .
++ This library is distributed in the hope that it will be useful,
++ but WITHOUT ANY WARRANTY; without even the implied warranty of
++ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
++ Library General Public License for more details.
++ .
++ You should have received a copy of the GNU Library General Public License
++ along with this library; see the file COPYING.LIB. If not, write to
++ the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
++ Boston, MA 02110-1301, USA.
++
++License: GPL-3+
++ This program is free software: you can redistribute it and/or modify
++ it under the terms of the GNU General Public License as published by
++ the Free Software Foundation; either version 3 of the License, or
++ (at your option) any later version.
++ .
++ This program is distributed in the hope that it will be useful,
++ but WITHOUT ANY WARRANTY; without even the implied warranty of
++ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
++ GNU General Public License for more details.
++ .
++ You should have received a copy of the GNU General Public License
++ along with this program. If not, see <http://www.gnu.org/licenses/>.
++ .
++ On Debian systems, the complete text of the GNU General
++ Public License version 3 can be found in "/usr/share/common-licenses/GPL-3".
--- /dev/null
--- /dev/null
++usr/bin/volk*
--- /dev/null
--- /dev/null
++debian/volk-config-info.1
++debian/volk_modtool.1
++debian/volk_profile.1
--- /dev/null
--- /dev/null
++<?xml version="1.0" encoding="utf-8"?>
++<descriptor>
++
++<headers>
++debian/libvolk1-dev/usr/include/
++</headers>
++
++<libs>
++debian/libvolk1.3/usr/lib/
++</libs>
++
++</descriptor>
--- /dev/null
--- /dev/null
++usr/include/*
++usr/lib/*/*volk*so
++usr/lib/*/cmake/volk
++usr/lib/*/pkgconfig/*volk*
--- /dev/null
--- /dev/null
++usr/lib/*/libvolk.so.1.3
--- /dev/null
--- /dev/null
++From 4461f27f6533cf29baaac0ff9cdd9b6241c0840b Mon Sep 17 00:00:00 2001
++From: Paul Cercueil <paul.cercueil@analog.com>
++Date: Wed, 17 Feb 2016 14:51:00 +0100
++Subject: [PATCH 01/18] Add a AppVeyor compatible YAML file for building on the
++ AppVeyor CI
++
++Signed-off-by: Paul Cercueil <paul.cercueil@analog.com>
++---
++ appveyor.yml | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
++ 1 file changed, 55 insertions(+)
++ create mode 100644 appveyor.yml
++
++diff --git a/appveyor.yml b/appveyor.yml
++new file mode 100644
++index 0000000..052ea51
++--- /dev/null
+++++ b/appveyor.yml
++@@ -0,0 +1,55 @@
+++clone_depth: 1
+++
+++os: Visual Studio 2013
+++
+++install:
+++ - echo "Installing Boost libraries..."
+++ - nuget install boost_system-vc120
+++ - nuget install boost_filesystem-vc120
+++ - nuget install boost_chrono-vc120
+++ - nuget install boost_program_options-vc120
+++ - nuget install boost_unit_test_framework-vc120
+++
+++ - echo "Installing Cheetah templates..."
+++ - appveyor DownloadFile https://pypi.python.org/packages/source/C/Cheetah/Cheetah-2.4.4.tar.gz
+++ - 7z x Cheetah-2.4.4.tar.gz
+++ - 7z x -y Cheetah-2.4.4.tar
+++ - cd Cheetah-2.4.4
+++ - c:\Python27\python.exe setup.py build
+++ - c:\Python27\python.exe setup.py install
+++
+++build_script:
+++ - cd c:\projects\volk
+++
+++ # Without this directory in the %PATH%, compiler tests fail because of missing DLLs
+++ - set PATH=%PATH%;C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\bin
+++
+++ - cmake -G "Visual Studio 12 Win64" \
+++ -DBoost_CHRONO_LIBRARY_RELEASE:FILEPATH=c:/projects/volk/boost_chrono-vc120.1.59.0.0/lib/native/address-model-64/lib/boost_chrono-vc120-mt-1_59.lib \
+++ -DBoost_FILESYSTEM_LIBRARY_RELEASE:FILEPATH=c:/projects/volk/boost_filesystem-vc120.1.59.0.0/lib/native/address-model-64/lib/boost_filesystem-vc120-mt-1_59.lib \
+++ -DBoost_PROGRAM_OPTIONS_LIBRARY_RELEASE:FILEPATH=c:/projects/volk/boost_program_options-vc120.1.59.0.0/lib/native/address-model-64/lib/boost_program_options-vc120-mt-1_59.lib \
+++ -DBoost_SYSTEM_LIBRARY_RELEASE:FILEPATH=c:/projects/volk/boost_system-vc120.1.59.0.0/lib/native/address-model-64/lib/boost_system-vc120-mt-1_59.lib \
+++ -DBoost_UNIT_TEST_FRAMEWORK_LIBRARY_RELEASE:FILEPATH=c:/projects/volk/boost_unit_test_framework-vc120.1.59.0.0/lib/native/address-model-64/lib/boost_unit_test_framework-vc120-mt-1_59.lib \
+++ -DBoost_INCLUDE_DIR:PATH=c:/projects/volk/boost.1.59.0.0/lib/native/include \
+++ -DCMAKE_BUILD_TYPE:STRING=Release -DENABLE_ORC:BOOL=OFF -DENABLE_TESTING:BOOL=OFF \
+++ .
+++
+++ - cmake --build . --config Release --target INSTALL
+++
+++ # Create an archive
+++ - cd "c:\Program Files"
+++ - 7z a "c:\libvolk-x64.zip" volk
+++
+++ # Create the deps archive
+++ - mkdir dlls
+++ - copy c:\projects\volk\boost_chrono-vc120.1.59.0.0\lib\native\address-model-64\lib\boost_chrono-vc120-mt-1_59.dll dlls\boost_chrono-vc120-mt-1_59.dll
+++ - copy c:\projects\volk\boost_filesystem-vc120.1.59.0.0\lib\native\address-model-64\lib\boost_filesystem-vc120-mt-1_59.dll dlls\boost_filesystem-vc120-mt-1_59.dll
+++ - copy c:\projects\volk\boost_program_options-vc120.1.59.0.0\lib\native\address-model-64\lib\boost_program_options-vc120-mt-1_59.dll dlls\boost_program_options-vc120-mt-1_59.dll
+++ - copy c:\projects\volk\boost_system-vc120.1.59.0.0\lib\native\address-model-64\lib\boost_system-vc120-mt-1_59.dll dlls\boost_system-vc120-mt-1_59.dll
+++ - copy c:\projects\volk\boost_unit_test_framework-vc120.1.59.0.0\lib\native\address-model-64\lib\boost_unit_test_framework-vc120-mt-1_59.dll dlls\boost_unit_test_framework-vc120-mt-1_59.dll
+++ - cd dlls
+++ - 7z a "c:\libvolk-x64-deps.zip" *
+++
+++ # Push it!
+++ - appveyor PushArtifact c:\libvolk-x64.zip
+++ - appveyor PushArtifact c:\libvolk-x64-deps.zip
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 18428fb9f718f5f7fa34707dd47ab6db07d88683 Mon Sep 17 00:00:00 2001
++From: Nathan West <nathan.west@gnuradio.org>
++Date: Sat, 2 Jul 2016 12:01:28 -0400
++Subject: [PATCH 02/18] Update CMakeLists for 1.3 development
++
++---
++ CMakeLists.txt | 2 +-
++ 1 file changed, 1 insertion(+), 1 deletion(-)
++
++diff --git a/CMakeLists.txt b/CMakeLists.txt
++index 5ecc9c2..0d0b647 100644
++--- a/CMakeLists.txt
+++++ b/CMakeLists.txt
++@@ -45,7 +45,7 @@ message(STATUS "Build type set to ${CMAKE_BUILD_TYPE}.")
++
++ set(VERSION_INFO_MAJOR_VERSION 1)
++ set(VERSION_INFO_MINOR_VERSION 3)
++-set(VERSION_INFO_MAINT_VERSION 0)
+++set(VERSION_INFO_MAINT_VERSION 0git)
++ include(VolkVersion) #setup version info
++
++
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From e296749b8fe936f72e85cbaca57215cb528ed2e5 Mon Sep 17 00:00:00 2001
++From: Nathan West <nathan.west@gnuradio.org>
++Date: Mon, 1 Aug 2016 17:12:24 -0400
++Subject: [PATCH 03/18] apps: fix profile update reading end of lines
++
++---
++ apps/volk_profile.cc | 2 +-
++ 1 file changed, 1 insertion(+), 1 deletion(-)
++
++diff --git a/apps/volk_profile.cc b/apps/volk_profile.cc
++index 2086e3f..51591cc 100644
++--- a/apps/volk_profile.cc
+++++ b/apps/volk_profile.cc
++@@ -261,7 +261,7 @@ void read_results(std::vector<volk_test_results_t> *results, std::string path)
++ found = 127;
++ }
++ str_size = config_str.size();
++- char buffer[128];
+++ char buffer[128] = {'\0'};
++ config_str.copy(buffer, found + 1, 0);
++ buffer[found] = '\0';
++ single_kernel_result.push_back(std::string(buffer));
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 0d672945dfca506d4e49e857ce886d2b3dc80e96 Mon Sep 17 00:00:00 2001
++From: Nathan West <nathan.west@gnuradio.org>
++Date: Mon, 1 Aug 2016 17:12:24 -0400
++Subject: [PATCH 04/18] apps: fix profile update reading end of lines
++
++---
++ apps/volk_profile.cc | 2 +-
++ 1 file changed, 1 insertion(+), 1 deletion(-)
++
++diff --git a/apps/volk_profile.cc b/apps/volk_profile.cc
++index 2086e3f..51591cc 100644
++--- a/apps/volk_profile.cc
+++++ b/apps/volk_profile.cc
++@@ -261,7 +261,7 @@ void read_results(std::vector<volk_test_results_t> *results, std::string path)
++ found = 127;
++ }
++ str_size = config_str.size();
++- char buffer[128];
+++ char buffer[128] = {'\0'};
++ config_str.copy(buffer, found + 1, 0);
++ buffer[found] = '\0';
++ single_kernel_result.push_back(std::string(buffer));
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 0f6d889b891bc2ac78c56ad18e43c4ec5a372574 Mon Sep 17 00:00:00 2001
++From: Nathan West <nathan.west@gnuradio.org>
++Date: Thu, 4 Aug 2016 11:30:55 -0400
++Subject: [PATCH 05/18] qa: lower tolerance for 32fc_mag to fix issue #96
++
++---
++ lib/kernel_tests.h | 4 +++-
++ 1 file changed, 3 insertions(+), 1 deletion(-)
++
++diff --git a/lib/kernel_tests.h b/lib/kernel_tests.h
++index 2bf1f0c..7c82733 100644
++--- a/lib/kernel_tests.h
+++++ b/lib/kernel_tests.h
++@@ -24,6 +24,8 @@ std::vector<volk_test_case_t> init_test_list(volk_test_params_t test_params)
++ // Some kernels need a lower tolerance
++ volk_test_params_t test_params_inacc = volk_test_params_t(1e-2, test_params.scalar(),
++ test_params.vlen(), test_params.iter(), test_params.benchmark_mode(), test_params.kernel_regex());
+++ volk_test_params_t test_params_inacc_tenth = volk_test_params_t(1e-1, test_params.scalar(),
+++ test_params.vlen(), test_params.iter(), test_params.benchmark_mode(), test_params.kernel_regex());
++ volk_test_params_t test_params_int1 = volk_test_params_t(1, test_params.scalar(),
++ test_params.vlen(), test_params.iter(), test_params.benchmark_mode(), test_params.kernel_regex());
++
++@@ -79,7 +81,7 @@ std::vector<volk_test_case_t> init_test_list(volk_test_params_t test_params)
++ (VOLK_INIT_TEST(volk_32fc_index_max_16u, volk_test_params_t(3, test_params.scalar(), test_params.vlen(), test_params.iter(), test_params.benchmark_mode(), test_params.kernel_regex())))
++ (VOLK_INIT_TEST(volk_32fc_index_max_32u, volk_test_params_t(3, test_params.scalar(), test_params.vlen(), test_params.iter(), test_params.benchmark_mode(), test_params.kernel_regex())))
++ (VOLK_INIT_TEST(volk_32fc_s32f_magnitude_16i, test_params_int1))
++- (VOLK_INIT_TEST(volk_32fc_magnitude_32f, test_params_inacc))
+++ (VOLK_INIT_TEST(volk_32fc_magnitude_32f, test_params_inacc_tenth))
++ (VOLK_INIT_TEST(volk_32fc_magnitude_squared_32f, test_params))
++ (VOLK_INIT_TEST(volk_32fc_x2_multiply_32fc, test_params))
++ (VOLK_INIT_TEST(volk_32fc_x2_multiply_conjugate_32fc, test_params))
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From aeaf56828ba0a08728a0cf2d2370b5d0153332b1 Mon Sep 17 00:00:00 2001
++From: Carles Fernandez <carlesfernandez@gmail.com>
++Date: Fri, 23 Sep 2016 19:16:27 +0200
++Subject: [PATCH 06/18] Add NEON, AVX and unaligned versions of SSE4.1 and SSE
++
++---
++ kernels/volk/volk_32f_index_max_32u.h | 316 ++++++++++++++++++++++++++++++++++
++ 1 file changed, 316 insertions(+)
++
++diff --git a/kernels/volk/volk_32f_index_max_32u.h b/kernels/volk/volk_32f_index_max_32u.h
++index 17b8f70..1888405 100644
++--- a/kernels/volk/volk_32f_index_max_32u.h
+++++ b/kernels/volk/volk_32f_index_max_32u.h
++@@ -130,6 +130,69 @@ volk_32f_index_max_32u_a_sse4_1(uint32_t* target, const float* src0, uint32_t nu
++ #endif /*LV_HAVE_SSE4_1*/
++
++
+++#ifdef LV_HAVE_SSE4_1
+++#include<smmintrin.h>
+++
+++static inline void volk_32f_index_max_32u_u_sse4_1(uint32_t* target, const float* src0, uint32_t num_points)
+++{
+++ if(num_points > 0)
+++ {
+++ uint32_t number = 0;
+++ const uint32_t quarterPoints = num_points / 4;
+++
+++ float* inputPtr = (float*)src0;
+++
+++ __m128 indexIncrementValues = _mm_set1_ps(4);
+++ __m128 currentIndexes = _mm_set_ps(-1,-2,-3,-4);
+++
+++ float max = src0[0];
+++ float index = 0;
+++ __m128 maxValues = _mm_set1_ps(max);
+++ __m128 maxValuesIndex = _mm_setzero_ps();
+++ __m128 compareResults;
+++ __m128 currentValues;
+++
+++ __VOLK_ATTR_ALIGNED(16) float maxValuesBuffer[4];
+++ __VOLK_ATTR_ALIGNED(16) float maxIndexesBuffer[4];
+++
+++ for(;number < quarterPoints; number++)
+++ {
+++ currentValues = _mm_loadu_ps(inputPtr); inputPtr += 4;
+++ currentIndexes = _mm_add_ps(currentIndexes, indexIncrementValues);
+++ compareResults = _mm_cmpgt_ps(maxValues, currentValues);
+++ maxValuesIndex = _mm_blendv_ps(currentIndexes, maxValuesIndex, compareResults);
+++ maxValues = _mm_blendv_ps(currentValues, maxValues, compareResults);
+++ }
+++
+++ // Calculate the largest value from the remaining 4 points
+++ _mm_store_ps(maxValuesBuffer, maxValues);
+++ _mm_store_ps(maxIndexesBuffer, maxValuesIndex);
+++
+++ for(number = 0; number < 4; number++)
+++ {
+++ if(maxValuesBuffer[number] > max)
+++ {
+++ index = maxIndexesBuffer[number];
+++ max = maxValuesBuffer[number];
+++ }
+++ }
+++
+++ number = quarterPoints * 4;
+++ for(;number < num_points; number++)
+++ {
+++ if(src0[number] > max)
+++ {
+++ index = number;
+++ max = src0[number];
+++ }
+++ }
+++ target[0] = (uint32_t)index;
+++ }
+++}
+++
+++#endif /*LV_HAVE_SSE4_1*/
+++
+++
++ #ifdef LV_HAVE_SSE
++
++ #include<xmmintrin.h>
++@@ -193,6 +256,259 @@ volk_32f_index_max_32u_a_sse(uint32_t* target, const float* src0, uint32_t num_p
++ #endif /*LV_HAVE_SSE*/
++
++
+++#ifdef LV_HAVE_SSE
+++#include<xmmintrin.h>
+++
+++static inline void volk_32f_index_max_32u_u_sse(uint32_t* target, const float* src0, uint32_t num_points)
+++{
+++ if(num_points > 0)
+++ {
+++ uint32_t number = 0;
+++ const uint32_t quarterPoints = num_points / 4;
+++
+++ float* inputPtr = (float*)src0;
+++
+++ __m128 indexIncrementValues = _mm_set1_ps(4);
+++ __m128 currentIndexes = _mm_set_ps(-1,-2,-3,-4);
+++
+++ float max = src0[0];
+++ float index = 0;
+++ __m128 maxValues = _mm_set1_ps(max);
+++ __m128 maxValuesIndex = _mm_setzero_ps();
+++ __m128 compareResults;
+++ __m128 currentValues;
+++
+++ __VOLK_ATTR_ALIGNED(16) float maxValuesBuffer[4];
+++ __VOLK_ATTR_ALIGNED(16) float maxIndexesBuffer[4];
+++
+++ for(;number < quarterPoints; number++)
+++ {
+++ currentValues = _mm_loadu_ps(inputPtr); inputPtr += 4;
+++ currentIndexes = _mm_add_ps(currentIndexes, indexIncrementValues);
+++ compareResults = _mm_cmpgt_ps(maxValues, currentValues);
+++ maxValuesIndex = _mm_or_ps(_mm_and_ps(compareResults, maxValuesIndex) , _mm_andnot_ps(compareResults, currentIndexes));
+++ maxValues = _mm_or_ps(_mm_and_ps(compareResults, maxValues) , _mm_andnot_ps(compareResults, currentValues));
+++ }
+++
+++ // Calculate the largest value from the remaining 4 points
+++ _mm_store_ps(maxValuesBuffer, maxValues);
+++ _mm_store_ps(maxIndexesBuffer, maxValuesIndex);
+++
+++ for(number = 0; number < 4; number++)
+++ {
+++ if(maxValuesBuffer[number] > max)
+++ {
+++ index = maxIndexesBuffer[number];
+++ max = maxValuesBuffer[number];
+++ }
+++ }
+++
+++ number = quarterPoints * 4;
+++ for(;number < num_points; number++)
+++ {
+++ if(src0[number] > max)
+++ {
+++ index = number;
+++ max = src0[number];
+++ }
+++ }
+++ target[0] = (uint32_t)index;
+++ }
+++}
+++
+++#endif /*LV_HAVE_SSE*/
+++
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void volk_32f_index_max_32u_a_avx(uint32_t* target, const float* src0, uint32_t num_points)
+++{
+++ if(num_points > 0)
+++ {
+++ uint32_t number = 0;
+++ const uint32_t quarterPoints = num_points / 8;
+++
+++ float* inputPtr = (float*)src0;
+++
+++ __m256 indexIncrementValues = _mm256_set1_ps(8);
+++ __m256 currentIndexes = _mm256_set_ps(-1,-2,-3,-4,-5,-6,-7,-8);
+++
+++ float max = src0[0];
+++ float index = 0;
+++ __m256 maxValues = _mm256_set1_ps(max);
+++ __m256 maxValuesIndex = _mm256_setzero_ps();
+++ __m256 compareResults;
+++ __m256 currentValues;
+++
+++ __VOLK_ATTR_ALIGNED(32) float maxValuesBuffer[8];
+++ __VOLK_ATTR_ALIGNED(32) float maxIndexesBuffer[8];
+++
+++ for(;number < quarterPoints; number++)
+++ {
+++ currentValues = _mm256_load_ps(inputPtr); inputPtr += 8;
+++ currentIndexes = _mm256_add_ps(currentIndexes, indexIncrementValues);
+++ compareResults = _mm256_cmp_ps(maxValues, currentValues, 0x1e);
+++ maxValuesIndex = _mm256_blendv_ps(currentIndexes, maxValuesIndex, compareResults);
+++ maxValues = _mm256_blendv_ps(currentValues, maxValues, compareResults);
+++ }
+++
+++ // Calculate the largest value from the remaining 8 points
+++ _mm256_store_ps(maxValuesBuffer, maxValues);
+++ _mm256_store_ps(maxIndexesBuffer, maxValuesIndex);
+++
+++ for(number = 0; number < 8; number++)
+++ {
+++ if(maxValuesBuffer[number] > max)
+++ {
+++ index = maxIndexesBuffer[number];
+++ max = maxValuesBuffer[number];
+++ }
+++ }
+++
+++ number = quarterPoints * 8;
+++ for(;number < num_points; number++)
+++ {
+++ if(src0[number] > max)
+++ {
+++ index = number;
+++ max = src0[number];
+++ }
+++ }
+++ target[0] = (uint32_t)index;
+++ }
+++}
+++
+++#endif /*LV_HAVE_AVX*/
+++
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void volk_32f_index_max_32u_u_avx(uint32_t* target, const float* src0, uint32_t num_points)
+++{
+++ if(num_points > 0)
+++ {
+++ uint32_t number = 0;
+++ const uint32_t quarterPoints = num_points / 8;
+++
+++ float* inputPtr = (float*)src0;
+++
+++ __m256 indexIncrementValues = _mm256_set1_ps(8);
+++ __m256 currentIndexes = _mm256_set_ps(-1,-2,-3,-4,-5,-6,-7,-8);
+++
+++ float max = src0[0];
+++ float index = 0;
+++ __m256 maxValues = _mm256_set1_ps(max);
+++ __m256 maxValuesIndex = _mm256_setzero_ps();
+++ __m256 compareResults;
+++ __m256 currentValues;
+++
+++ __VOLK_ATTR_ALIGNED(32) float maxValuesBuffer[8];
+++ __VOLK_ATTR_ALIGNED(32) float maxIndexesBuffer[8];
+++
+++ for(;number < quarterPoints; number++)
+++ {
+++ currentValues = _mm256_loadu_ps(inputPtr); inputPtr += 8;
+++ currentIndexes = _mm256_add_ps(currentIndexes, indexIncrementValues);
+++ compareResults = _mm256_cmp_ps(maxValues, currentValues, 0x1e);
+++ maxValuesIndex = _mm256_blendv_ps(currentIndexes, maxValuesIndex, compareResults);
+++ maxValues = _mm256_blendv_ps(currentValues, maxValues, compareResults);
+++ }
+++
+++ // Calculate the largest value from the remaining 8 points
+++ _mm256_store_ps(maxValuesBuffer, maxValues);
+++ _mm256_store_ps(maxIndexesBuffer, maxValuesIndex);
+++
+++ for(number = 0; number < 8; number++)
+++ {
+++ if(maxValuesBuffer[number] > max)
+++ {
+++ index = maxIndexesBuffer[number];
+++ max = maxValuesBuffer[number];
+++ }
+++ }
+++
+++ number = quarterPoints * 8;
+++ for(;number < num_points; number++)
+++ {
+++ if(src0[number] > max)
+++ {
+++ index = number;
+++ max = src0[number];
+++ }
+++ }
+++ target[0] = (uint32_t)index;
+++ }
+++}
+++
+++#endif /*LV_HAVE_AVX*/
+++
+++
+++#ifdef LV_HAVE_NEON
+++#include <arm_neon.h>
+++
+++static inline void volk_32f_index_max_32u_neon(uint32_t* target, const float* src0, uint32_t num_points)
+++{
+++ if(num_points > 0)
+++ {
+++ uint32_t number = 0;
+++ const uint32_t quarterPoints = num_points / 4;
+++
+++ float* inputPtr = (float*)src0;
+++ float32x4_t indexIncrementValues = vdupq_n_f32(4);
+++ __VOLK_ATTR_ALIGNED(16) float currentIndexes_float[4] = { -4.0f, -3.0f, -2.0f, -1.0f };
+++ float32x4_t currentIndexes = vld1q_f32(currentIndexes_float);
+++
+++ float max = src0[0];
+++ float index = 0;
+++ float32x4_t maxValues = vdupq_n_f32(max);
+++ uint32x4_t maxValuesIndex = vmovq_n_u32(0);
+++ uint32x4_t compareResults;
+++ uint32x4_t currentIndexes_u;
+++ float32x4_t currentValues;
+++
+++ __VOLK_ATTR_ALIGNED(16) float maxValuesBuffer[4];
+++ __VOLK_ATTR_ALIGNED(16) float maxIndexesBuffer[4];
+++
+++ for(;number < quarterPoints; number++)
+++ {
+++ currentValues = vld1q_f32(inputPtr); inputPtr += 4;
+++ currentIndexes = vaddq_f32(currentIndexes, indexIncrementValues);
+++ currentIndexes_u = vcvtq_u32_f32(currentIndexes);
+++ compareResults = vcgtq_f32( maxValues, currentValues);
+++ maxValuesIndex = vorrq_u32( vandq_u32( compareResults, maxValuesIndex ), vbicq_u32(currentIndexes_u, compareResults) );
+++ maxValues = vmaxq_f32(currentValues, maxValues);
+++ }
+++
+++ // Calculate the largest value from the remaining 4 points
+++ vst1q_f32(maxValuesBuffer, maxValues);
+++ vst1q_f32(maxIndexesBuffer, vcvtq_f32_u32(maxValuesIndex));
+++ for(number = 0; number < 4; number++)
+++ {
+++ if(maxValuesBuffer[number] > max)
+++ {
+++ index = maxIndexesBuffer[number];
+++ max = maxValuesBuffer[number];
+++ }
+++ }
+++
+++ number = quarterPoints * 4;
+++ for(;number < num_points; number++)
+++ {
+++ if(src0[number] > max)
+++ {
+++ index = number;
+++ max = src0[number];
+++ }
+++ }
+++ target[0] = (uint32_t)index;
+++ }
+++}
+++
+++#endif /*LV_HAVE_NEON*/
+++
+++
++ #ifdef LV_HAVE_GENERIC
++
++ static inline void
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From d065c1cdd34c0f5c78911331381e10687faa14a0 Mon Sep 17 00:00:00 2001
++From: Josh Blum <josh@joshknows.com>
++Date: Fri, 20 Jan 2017 10:03:49 -0800
++Subject: [PATCH 07/18] added __VOLK_PREFETCH() compatibility macro
++
++__VOLK_PREFETCH() performs __builtin_prefetch() on GCC compilers
++and is otherwise a NOP for other systems. The use of __builtin_prefetch
++was replaced with __VOLK_PREFETCH() to make the kernels portable.
++---
++ include/volk/volk_common.h | 3 +++
++ kernels/volk/volk_16i_max_star_16i.h | 2 +-
++ kernels/volk/volk_16i_max_star_horizontal_16i.h | 2 +-
++ kernels/volk/volk_16ic_convert_32fc.h | 2 +-
++ kernels/volk/volk_16ic_x2_dot_prod_16ic.h | 28 +++++++++++-----------
++ kernels/volk/volk_16ic_x2_multiply_16ic.h | 4 ++--
++ kernels/volk/volk_32f_x2_add_32f.h | 4 ++--
++ kernels/volk/volk_32fc_conjugate_32fc.h | 2 +-
++ kernels/volk/volk_32fc_convert_16ic.h | 6 ++---
++ .../volk/volk_32fc_x2_conjugate_dot_prod_32fc.h | 4 ++--
++ kernels/volk/volk_32fc_x2_dot_prod_32fc.h | 16 ++++++-------
++ kernels/volk/volk_32fc_x2_multiply_32fc.h | 8 +++----
++ .../volk/volk_32fc_x2_multiply_conjugate_32fc.h | 4 ++--
++ 13 files changed, 44 insertions(+), 41 deletions(-)
++
++diff --git a/include/volk/volk_common.h b/include/volk/volk_common.h
++index 4d35f5c..a53b139 100644
++--- a/include/volk/volk_common.h
+++++ b/include/volk/volk_common.h
++@@ -16,6 +16,7 @@
++ # define __VOLK_ATTR_EXPORT
++ # define __VOLK_ATTR_IMPORT
++ # endif
+++# define __VOLK_PREFETCH(addr) __builtin_prefetch(addr)
++ #elif _MSC_VER
++ # define __VOLK_ATTR_ALIGNED(x) __declspec(align(x))
++ # define __VOLK_ATTR_UNUSED
++@@ -23,6 +24,7 @@
++ # define __VOLK_ATTR_DEPRECATED __declspec(deprecated)
++ # define __VOLK_ATTR_EXPORT __declspec(dllexport)
++ # define __VOLK_ATTR_IMPORT __declspec(dllimport)
+++# define __VOLK_PREFETCH(addr)
++ #else
++ # define __VOLK_ATTR_ALIGNED(x)
++ # define __VOLK_ATTR_UNUSED
++@@ -30,6 +32,7 @@
++ # define __VOLK_ATTR_DEPRECATED
++ # define __VOLK_ATTR_EXPORT
++ # define __VOLK_ATTR_IMPORT
+++# define __VOLK_PREFETCH(addr)
++ #endif
++
++ ////////////////////////////////////////////////////////////////////////
++diff --git a/kernels/volk/volk_16i_max_star_16i.h b/kernels/volk/volk_16i_max_star_16i.h
++index e470642..531a8b5 100644
++--- a/kernels/volk/volk_16i_max_star_16i.h
+++++ b/kernels/volk/volk_16i_max_star_16i.h
++@@ -139,7 +139,7 @@ volk_16i_max_star_16i_neon(short* target, short* src0, unsigned int num_points)
++
++ for(number=0; number < eighth_points; ++number) {
++ input_vec = vld1q_s16(src0);
++- __builtin_prefetch(src0+16);
+++ __VOLK_PREFETCH(src0+16);
++ diff = vsubq_s16(candidate_vec, input_vec);
++ comp1 = vcgeq_s16(diff, zeros);
++ comp2 = vcltq_s16(diff, zeros);
++diff --git a/kernels/volk/volk_16i_max_star_horizontal_16i.h b/kernels/volk/volk_16i_max_star_horizontal_16i.h
++index 1da8356..964587c 100644
++--- a/kernels/volk/volk_16i_max_star_horizontal_16i.h
+++++ b/kernels/volk/volk_16i_max_star_horizontal_16i.h
++@@ -169,7 +169,7 @@ volk_16i_max_star_horizontal_16i_neon(int16_t* target, int16_t* src0, unsigned i
++ zeros = veorq_s16(zeros, zeros);
++ for(number=0; number < eighth_points; ++number) {
++ input_vec = vld2q_s16(src0);
++- //__builtin_prefetch(src0+16);
+++ //__VOLK_PREFETCH(src0+16);
++ diff = vsubq_s16(input_vec.val[0], input_vec.val[1]);
++ comp1 = vcgeq_s16(diff, zeros);
++ comp2 = vcltq_s16(diff, zeros);
++diff --git a/kernels/volk/volk_16ic_convert_32fc.h b/kernels/volk/volk_16ic_convert_32fc.h
++index 88e079d..9779b0f 100644
++--- a/kernels/volk/volk_16ic_convert_32fc.h
+++++ b/kernels/volk/volk_16ic_convert_32fc.h
++@@ -198,7 +198,7 @@ static inline void volk_16ic_convert_32fc_neon(lv_32fc_t* outputVector, const lv
++ for(number = 0; number < sse_iters; number++)
++ {
++ a16x4 = vld1_s16((const int16_t*)_in);
++- __builtin_prefetch(_in + 4);
+++ __VOLK_PREFETCH(_in + 4);
++ a32x4 = vmovl_s16(a16x4);
++ f32x4 = vcvtq_f32_s32(a32x4);
++ vst1q_f32((float32_t*)_out, f32x4);
++diff --git a/kernels/volk/volk_16ic_x2_dot_prod_16ic.h b/kernels/volk/volk_16ic_x2_dot_prod_16ic.h
++index 9d4c882..8e6de4c 100644
++--- a/kernels/volk/volk_16ic_x2_dot_prod_16ic.h
+++++ b/kernels/volk/volk_16ic_x2_dot_prod_16ic.h
++@@ -96,9 +96,9 @@ static inline void volk_16ic_x2_dot_prod_16ic_a_sse2(lv_16sc_t* out, const lv_16
++ {
++ // a[127:0]=[a3.i,a3.r,a2.i,a2.r,a1.i,a1.r,a0.i,a0.r]
++ a = _mm_load_si128((__m128i*)_in_a); //load (2 byte imag, 2 byte real) x 4 into 128 bits reg
++- __builtin_prefetch(_in_a + 8);
+++ __VOLK_PREFETCH(_in_a + 8);
++ b = _mm_load_si128((__m128i*)_in_b);
++- __builtin_prefetch(_in_b + 8);
+++ __VOLK_PREFETCH(_in_b + 8);
++ c = _mm_mullo_epi16(a, b); // a3.i*b3.i, a3.r*b3.r, ....
++
++ c_sr = _mm_srli_si128(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst.
++@@ -173,9 +173,9 @@ static inline void volk_16ic_x2_dot_prod_16ic_u_sse2(lv_16sc_t* out, const lv_16
++ {
++ // a[127:0]=[a3.i,a3.r,a2.i,a2.r,a1.i,a1.r,a0.i,a0.r]
++ a = _mm_loadu_si128((__m128i*)_in_a); //load (2 byte imag, 2 byte real) x 4 into 128 bits reg
++- __builtin_prefetch(_in_a + 8);
+++ __VOLK_PREFETCH(_in_a + 8);
++ b = _mm_loadu_si128((__m128i*)_in_b);
++- __builtin_prefetch(_in_b + 8);
+++ __VOLK_PREFETCH(_in_b + 8);
++ c = _mm_mullo_epi16(a, b); // a3.i*b3.i, a3.r*b3.r, ....
++
++ c_sr = _mm_srli_si128(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst.
++@@ -248,9 +248,9 @@ static inline void volk_16ic_x2_dot_prod_16ic_u_axv2(lv_16sc_t* out, const lv_16
++ for(number = 0; number < avx_iters; number++)
++ {
++ a = _mm256_loadu_si256((__m256i*)_in_a);
++- __builtin_prefetch(_in_a + 16);
+++ __VOLK_PREFETCH(_in_a + 16);
++ b = _mm256_loadu_si256((__m256i*)_in_b);
++- __builtin_prefetch(_in_b + 16);
+++ __VOLK_PREFETCH(_in_b + 16);
++ c = _mm256_mullo_epi16(a, b);
++
++ c_sr = _mm256_srli_si256(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst.
++@@ -324,9 +324,9 @@ static inline void volk_16ic_x2_dot_prod_16ic_a_axv2(lv_16sc_t* out, const lv_16
++ for(number = 0; number < avx_iters; number++)
++ {
++ a = _mm256_load_si256((__m256i*)_in_a);
++- __builtin_prefetch(_in_a + 16);
+++ __VOLK_PREFETCH(_in_a + 16);
++ b = _mm256_load_si256((__m256i*)_in_b);
++- __builtin_prefetch(_in_b + 16);
+++ __VOLK_PREFETCH(_in_b + 16);
++ c = _mm256_mullo_epi16(a, b);
++
++ c_sr = _mm256_srli_si256(c, 2); // Shift a right by imm8 bytes while shifting in zeros, and store the results in dst.
++@@ -399,8 +399,8 @@ static inline void volk_16ic_x2_dot_prod_16ic_neon(lv_16sc_t* out, const lv_16sc
++ {
++ a_val = vld2_s16((int16_t*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2_s16((int16_t*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr + 8);
++- __builtin_prefetch(b_ptr + 8);
+++ __VOLK_PREFETCH(a_ptr + 8);
+++ __VOLK_PREFETCH(b_ptr + 8);
++
++ // multiply the real*real and imag*imag to get real result
++ // a0r*b0r|a1r*b1r|a2r*b2r|a3r*b3r
++@@ -465,8 +465,8 @@ static inline void volk_16ic_x2_dot_prod_16ic_neon_vma(lv_16sc_t* out, const lv_
++ {
++ a_val = vld2_s16((int16_t*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2_s16((int16_t*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr + 8);
++- __builtin_prefetch(b_ptr + 8);
+++ __VOLK_PREFETCH(a_ptr + 8);
+++ __VOLK_PREFETCH(b_ptr + 8);
++
++ tmp.val[0] = vmul_s16(a_val.val[0], b_val.val[0]);
++ tmp.val[1] = vmul_s16(a_val.val[1], b_val.val[0]);
++@@ -519,8 +519,8 @@ static inline void volk_16ic_x2_dot_prod_16ic_neon_optvma(lv_16sc_t* out, const
++ {
++ a_val = vld2_s16((int16_t*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2_s16((int16_t*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr + 8);
++- __builtin_prefetch(b_ptr + 8);
+++ __VOLK_PREFETCH(a_ptr + 8);
+++ __VOLK_PREFETCH(b_ptr + 8);
++
++ // use 2 accumulators to remove inter-instruction data dependencies
++ accumulator1.val[0] = vmla_s16(accumulator1.val[0], a_val.val[0], b_val.val[0]);
++diff --git a/kernels/volk/volk_16ic_x2_multiply_16ic.h b/kernels/volk/volk_16ic_x2_multiply_16ic.h
++index 17033ae..9dcf06f 100644
++--- a/kernels/volk/volk_16ic_x2_multiply_16ic.h
+++++ b/kernels/volk/volk_16ic_x2_multiply_16ic.h
++@@ -291,8 +291,8 @@ static inline void volk_16ic_x2_multiply_16ic_neon(lv_16sc_t* out, const lv_16sc
++ {
++ a_val = vld2_s16((int16_t*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2_s16((int16_t*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr + 4);
++- __builtin_prefetch(b_ptr + 4);
+++ __VOLK_PREFETCH(a_ptr + 4);
+++ __VOLK_PREFETCH(b_ptr + 4);
++
++ // multiply the real*real and imag*imag to get real result
++ // a0r*b0r|a1r*b1r|a2r*b2r|a3r*b3r
++diff --git a/kernels/volk/volk_32f_x2_add_32f.h b/kernels/volk/volk_32f_x2_add_32f.h
++index fc9cf5b..28cf73d 100644
++--- a/kernels/volk/volk_32f_x2_add_32f.h
+++++ b/kernels/volk/volk_32f_x2_add_32f.h
++@@ -191,8 +191,8 @@ volk_32f_x2_add_32f_u_neon(float* cVector, const float* aVector,
++ // Load in to NEON registers
++ aVal = vld1q_f32(aPtr);
++ bVal = vld1q_f32(bPtr);
++- __builtin_prefetch(aPtr+4);
++- __builtin_prefetch(bPtr+4);
+++ __VOLK_PREFETCH(aPtr+4);
+++ __VOLK_PREFETCH(bPtr+4);
++
++ // vector add
++ cVal = vaddq_f32(aVal, bVal);
++diff --git a/kernels/volk/volk_32fc_conjugate_32fc.h b/kernels/volk/volk_32fc_conjugate_32fc.h
++index 1fdb6c2..6994d0e 100644
++--- a/kernels/volk/volk_32fc_conjugate_32fc.h
+++++ b/kernels/volk/volk_32fc_conjugate_32fc.h
++@@ -248,7 +248,7 @@ volk_32fc_conjugate_32fc_a_neon(lv_32fc_t* cVector, const lv_32fc_t* aVector, un
++ const lv_32fc_t* a = aVector;
++
++ for(number=0; number < quarterPoints; number++){
++- __builtin_prefetch(a+4);
+++ __VOLK_PREFETCH(a+4);
++ x = vld2q_f32((float*)a); // Load the complex data as ar,br,cr,dr; ai,bi,ci,di
++
++ // xor the imaginary lane
++diff --git a/kernels/volk/volk_32fc_convert_16ic.h b/kernels/volk/volk_32fc_convert_16ic.h
++index 4f6e6a5..307ab36 100644
++--- a/kernels/volk/volk_32fc_convert_16ic.h
+++++ b/kernels/volk/volk_32fc_convert_16ic.h
++@@ -75,7 +75,7 @@ static inline void volk_32fc_convert_16ic_u_sse2(lv_16sc_t* outputVector, const
++ {
++ inputVal1 = _mm_loadu_ps((float*)inputVectorPtr); inputVectorPtr += 4;
++ inputVal2 = _mm_loadu_ps((float*)inputVectorPtr); inputVectorPtr += 4;
++- __builtin_prefetch(inputVectorPtr + 8);
+++ __VOLK_PREFETCH(inputVectorPtr + 8);
++
++ // Clip
++ ret1 = _mm_max_ps(_mm_min_ps(inputVal1, vmax_val), vmin_val);
++@@ -128,7 +128,7 @@ static inline void volk_32fc_convert_16ic_a_sse2(lv_16sc_t* outputVector, const
++ {
++ inputVal1 = _mm_load_ps((float*)inputVectorPtr); inputVectorPtr += 4;
++ inputVal2 = _mm_load_ps((float*)inputVectorPtr); inputVectorPtr += 4;
++- __builtin_prefetch(inputVectorPtr + 8);
+++ __VOLK_PREFETCH(inputVectorPtr + 8);
++
++ // Clip
++ ret1 = _mm_max_ps(_mm_min_ps(inputVal1, vmax_val), vmin_val);
++@@ -184,7 +184,7 @@ static inline void volk_32fc_convert_16ic_neon(lv_16sc_t* outputVector, const lv
++ {
++ a = vld1q_f32((const float32_t*)(inputVectorPtr)); inputVectorPtr += 4;
++ b = vld1q_f32((const float32_t*)(inputVectorPtr)); inputVectorPtr += 4;
++- __builtin_prefetch(inputVectorPtr + 8);
+++ __VOLK_PREFETCH(inputVectorPtr + 8);
++
++ ret1 = vmaxq_f32(vminq_f32(a, max_val), min_val);
++ ret2 = vmaxq_f32(vminq_f32(b, max_val), min_val);
++diff --git a/kernels/volk/volk_32fc_x2_conjugate_dot_prod_32fc.h b/kernels/volk/volk_32fc_x2_conjugate_dot_prod_32fc.h
++index 981899c..4addf80 100644
++--- a/kernels/volk/volk_32fc_x2_conjugate_dot_prod_32fc.h
+++++ b/kernels/volk/volk_32fc_x2_conjugate_dot_prod_32fc.h
++@@ -219,8 +219,8 @@ static inline void volk_32fc_x2_conjugate_dot_prod_32fc_neon(lv_32fc_t* result,
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+8);
++- __builtin_prefetch(b_ptr+8);
+++ __VOLK_PREFETCH(a_ptr+8);
+++ __VOLK_PREFETCH(b_ptr+8);
++
++ // do the first multiply
++ tmp_imag.val[1] = vmulq_f32(a_val.val[1], b_val.val[0]);
++diff --git a/kernels/volk/volk_32fc_x2_dot_prod_32fc.h b/kernels/volk/volk_32fc_x2_dot_prod_32fc.h
++index 39d0c78..0c3271c 100644
++--- a/kernels/volk/volk_32fc_x2_dot_prod_32fc.h
+++++ b/kernels/volk/volk_32fc_x2_dot_prod_32fc.h
++@@ -894,8 +894,8 @@ static inline void volk_32fc_x2_dot_prod_32fc_neon(lv_32fc_t* result, const lv_3
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+8);
++- __builtin_prefetch(b_ptr+8);
+++ __VOLK_PREFETCH(a_ptr+8);
+++ __VOLK_PREFETCH(b_ptr+8);
++
++ // multiply the real*real and imag*imag to get real result
++ // a0r*b0r|a1r*b1r|a2r*b2r|a3r*b3r
++@@ -949,8 +949,8 @@ static inline void volk_32fc_x2_dot_prod_32fc_neon_opttests(lv_32fc_t* result, c
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+8);
++- __builtin_prefetch(b_ptr+8);
+++ __VOLK_PREFETCH(a_ptr+8);
+++ __VOLK_PREFETCH(b_ptr+8);
++
++ // do the first multiply
++ tmp_imag.val[1] = vmulq_f32(a_val.val[1], b_val.val[0]);
++@@ -998,8 +998,8 @@ static inline void volk_32fc_x2_dot_prod_32fc_neon_optfma(lv_32fc_t* result, con
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+8);
++- __builtin_prefetch(b_ptr+8);
+++ __VOLK_PREFETCH(a_ptr+8);
+++ __VOLK_PREFETCH(b_ptr+8);
++
++ // use 2 accumulators to remove inter-instruction data dependencies
++ accumulator1.val[0] = vmlaq_f32(accumulator1.val[0], a_val.val[0], b_val.val[0]);
++@@ -1050,8 +1050,8 @@ static inline void volk_32fc_x2_dot_prod_32fc_neon_optfmaunroll(lv_32fc_t* resul
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld4q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld4q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+8);
++- __builtin_prefetch(b_ptr+8);
+++ __VOLK_PREFETCH(a_ptr+8);
+++ __VOLK_PREFETCH(b_ptr+8);
++
++ // use 2 accumulators to remove inter-instruction data dependencies
++ accumulator1.val[0] = vmlaq_f32(accumulator1.val[0], a_val.val[0], b_val.val[0]);
++diff --git a/kernels/volk/volk_32fc_x2_multiply_32fc.h b/kernels/volk/volk_32fc_x2_multiply_32fc.h
++index 1709140..0b9d3fe 100644
++--- a/kernels/volk/volk_32fc_x2_multiply_32fc.h
+++++ b/kernels/volk/volk_32fc_x2_multiply_32fc.h
++@@ -372,8 +372,8 @@ volk_32fc_x2_multiply_32fc_neon(lv_32fc_t* cVector, const lv_32fc_t* aVector,
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+4);
++- __builtin_prefetch(b_ptr+4);
+++ __VOLK_PREFETCH(a_ptr+4);
+++ __VOLK_PREFETCH(b_ptr+4);
++
++ // multiply the real*real and imag*imag to get real result
++ // a0r*b0r|a1r*b1r|a2r*b2r|a3r*b3r
++@@ -420,8 +420,8 @@ volk_32fc_x2_multiply_32fc_neon_opttests(lv_32fc_t* cVector, const lv_32fc_t* aV
++ for(number = 0; number < quarter_points; ++number) {
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++- __builtin_prefetch(a_ptr+4);
++- __builtin_prefetch(b_ptr+4);
+++ __VOLK_PREFETCH(a_ptr+4);
+++ __VOLK_PREFETCH(b_ptr+4);
++
++ // do the first multiply
++ tmp_imag.val[1] = vmulq_f32(a_val.val[1], b_val.val[0]);
++diff --git a/kernels/volk/volk_32fc_x2_multiply_conjugate_32fc.h b/kernels/volk/volk_32fc_x2_multiply_conjugate_32fc.h
++index 703c78d..c13a32e 100644
++--- a/kernels/volk/volk_32fc_x2_multiply_conjugate_32fc.h
+++++ b/kernels/volk/volk_32fc_x2_multiply_conjugate_32fc.h
++@@ -262,8 +262,8 @@ volk_32fc_x2_multiply_conjugate_32fc_neon(lv_32fc_t* cVector, const lv_32fc_t* a
++ a_val = vld2q_f32((float*)a_ptr); // a0r|a1r|a2r|a3r || a0i|a1i|a2i|a3i
++ b_val = vld2q_f32((float*)b_ptr); // b0r|b1r|b2r|b3r || b0i|b1i|b2i|b3i
++ b_val.val[1] = vnegq_f32(b_val.val[1]);
++- __builtin_prefetch(a_ptr+4);
++- __builtin_prefetch(b_ptr+4);
+++ __VOLK_PREFETCH(a_ptr+4);
+++ __VOLK_PREFETCH(b_ptr+4);
++
++ // multiply the real*real and imag*imag to get real result
++ // a0r*b0r|a1r*b1r|a2r*b2r|a3r*b3r
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From b0b9615e4e5d38c0d8d6bcc06ccefe08682ec352 Mon Sep 17 00:00:00 2001
++From: Nick Foster <bistromath@gmail.com>
++Date: Fri, 20 Jan 2017 16:36:01 -0800
++Subject: [PATCH 08/18] Fix bug 106 (volk_64u_popcnt bug in generic
++ implementation)
++
++---
++ kernels/volk/volk_64u_popcnt.h | 2 +-
++ 1 file changed, 1 insertion(+), 1 deletion(-)
++
++diff --git a/kernels/volk/volk_64u_popcnt.h b/kernels/volk/volk_64u_popcnt.h
++index 653bfb9..cbce2ec 100644
++--- a/kernels/volk/volk_64u_popcnt.h
+++++ b/kernels/volk/volk_64u_popcnt.h
++@@ -84,7 +84,7 @@ volk_64u_popcnt_generic(uint64_t* ret, const uint64_t value)
++ uint64_t retVal64 = retVal;
++
++ //retVal = valueVector[1];
++- retVal = (uint32_t)((value & 0xFFFFFFFF00000000ull) >> 31);
+++ retVal = (uint32_t)((value & 0xFFFFFFFF00000000ull) >> 32);
++ retVal = (retVal & 0x55555555) + (retVal >> 1 & 0x55555555);
++ retVal = (retVal & 0x33333333) + (retVal >> 2 & 0x33333333);
++ retVal = (retVal + (retVal >> 4)) & 0x0F0F0F0F;
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 5af8aa45fa23f72aff8593f54e7b67e449927681 Mon Sep 17 00:00:00 2001
++From: Nathan West <nathan.west@nrl.navy.mil>
++Date: Mon, 13 Mar 2017 12:25:35 -0400
++Subject: [PATCH 09/18] modtool: deconflict module include guards from main
++ volk
++
++---
++ python/volk_modtool/volk_modtool_generate.py | 8 ++++----
++ 1 file changed, 4 insertions(+), 4 deletions(-)
++
++diff --git a/python/volk_modtool/volk_modtool_generate.py b/python/volk_modtool/volk_modtool_generate.py
++index 83e0d26..6040a7d 100644
++--- a/python/volk_modtool/volk_modtool_generate.py
+++++ b/python/volk_modtool/volk_modtool_generate.py
++@@ -98,6 +98,9 @@ class volk_modtool:
++ os.makedirs(os.path.join(self.my_dict['destination'], 'volk_' + self.my_dict['name'], 'kernels/volk_' + self.my_dict['name']))
++
++ current_kernel_names = self.get_current_kernels()
+++ need_ifdef_updates = ["constant.h", "volk_complex.h", "volk_malloc.h", "volk_prefs.h",
+++ "volk_common.h", "volk_cpu.tmpl.h", "volk_config_fixed.tmpl.h",
+++ "volk_typedefs.h", "volk.tmpl.h"]
++ for root, dirnames, filenames in os.walk(self.my_dict['base']):
++ for name in filenames:
++ t_table = map(lambda a: re.search(a, name), current_kernel_names)
++@@ -107,10 +110,7 @@ class volk_modtool:
++ instring = open(infile, 'r').read()
++ outstring = re.sub(self.volk, 'volk_' + self.my_dict['name'], instring)
++ # Update the header ifdef guards only where needed
++- if((name == "constants.h") or
++- (name == "volk_complex.h") or
++- (name == "volk_malloc.h") or
++- (name == "volk_prefs.h")):
+++ if name in need_ifdef_updates:
++ outstring = re.sub(self.volk_included, 'INCLUDED_VOLK_' + self.my_dict['name'].upper(), outstring)
++ newname = re.sub(self.volk, 'volk_' + self.my_dict['name'], name)
++ relpath = os.path.relpath(infile, self.my_dict['base'])
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 663dbd00b3e4bd3ddb0b7f8a9360df132d7f0d56 Mon Sep 17 00:00:00 2001
++From: Nathan West <nathan.west@nrl.navy.mil>
++Date: Mon, 13 Mar 2017 12:37:18 -0400
++Subject: [PATCH 10/18] modtool: update the cmake find module for volk mods
++
++---
++ python/volk_modtool/volk_modtool_generate.py | 4 ++++
++ 1 file changed, 4 insertions(+)
++
++diff --git a/python/volk_modtool/volk_modtool_generate.py b/python/volk_modtool/volk_modtool_generate.py
++index 6040a7d..75232ed 100644
++--- a/python/volk_modtool/volk_modtool_generate.py
+++++ b/python/volk_modtool/volk_modtool_generate.py
++@@ -113,6 +113,10 @@ class volk_modtool:
++ if name in need_ifdef_updates:
++ outstring = re.sub(self.volk_included, 'INCLUDED_VOLK_' + self.my_dict['name'].upper(), outstring)
++ newname = re.sub(self.volk, 'volk_' + self.my_dict['name'], name)
+++ if name == 'VolkConfig.cmake.in':
+++ outstring = re.sub("VOLK", 'VOLK_' + self.my_dict['name'].upper(), outstring)
+++ newname = "Volk%sConfig.cmake.in" % self.my_dict['name']
+++
++ relpath = os.path.relpath(infile, self.my_dict['base'])
++ newrelpath = re.sub(self.volk, 'volk_' + self.my_dict['name'], relpath)
++ dest = os.path.join(self.my_dict['destination'], 'volk_' + self.my_dict['name'], os.path.dirname(newrelpath), newname)
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 28b03a9a338dc24b002413e880222fe1d49f77f5 Mon Sep 17 00:00:00 2001
++From: Michael Dickens <michael.dickens@ettus.com>
++Date: Sat, 1 Apr 2017 15:24:46 -0400
++Subject: [PATCH 11/18] Use 'powf' to match variables and avoid implicit type
++ converstion. Makes some older compilers happy, allowing 'make test' to pass.
++
++---
++ kernels/volk/volk_32f_x2_pow_32f.h | 6 +++---
++ 1 file changed, 3 insertions(+), 3 deletions(-)
++
++diff --git a/kernels/volk/volk_32f_x2_pow_32f.h b/kernels/volk/volk_32f_x2_pow_32f.h
++index 58fecb6..a8cb2e1 100644
++--- a/kernels/volk/volk_32f_x2_pow_32f.h
+++++ b/kernels/volk/volk_32f_x2_pow_32f.h
++@@ -190,7 +190,7 @@ volk_32f_x2_pow_32f_a_sse4_1(float* cVector, const float* bVector,
++
++ number = quarterPoints * 4;
++ for(;number < num_points; number++){
++- *cPtr++ = pow(*aPtr++, *bPtr++);
+++ *cPtr++ = powf(*aPtr++, *bPtr++);
++ }
++ }
++
++@@ -215,7 +215,7 @@ volk_32f_x2_pow_32f_generic(float* cVector, const float* bVector,
++ unsigned int number = 0;
++
++ for(number = 0; number < num_points; number++){
++- *cPtr++ = pow(*aPtr++, *bPtr++);
+++ *cPtr++ = powf(*aPtr++, *bPtr++);
++ }
++ }
++ #endif /* LV_HAVE_GENERIC */
++@@ -326,7 +326,7 @@ volk_32f_x2_pow_32f_u_sse4_1(float* cVector, const float* bVector,
++
++ number = quarterPoints * 4;
++ for(;number < num_points; number++){
++- *cPtr++ = pow(*aPtr++, *bPtr++);
+++ *cPtr++ = powf(*aPtr++, *bPtr++);
++ }
++ }
++
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 67202d7b46f9ce55625d0ce5c3a2d98dff56b09a Mon Sep 17 00:00:00 2001
++From: Josh Blum <josh@joshknows.com>
++Date: Wed, 5 Oct 2016 14:09:05 -0700
++Subject: [PATCH 12/18] cmake: support empty CMAKE_INSTALL_PREFIX
++
++Needed quotes for the string escape command
++---
++ lib/CMakeLists.txt | 2 +-
++ 1 file changed, 1 insertion(+), 1 deletion(-)
++
++diff --git a/lib/CMakeLists.txt b/lib/CMakeLists.txt
++index ad5653c..45b6c51 100644
++--- a/lib/CMakeLists.txt
+++++ b/lib/CMakeLists.txt
++@@ -489,7 +489,7 @@ endif()
++ message(STATUS "Loading version ${VERSION} into constants...")
++
++ #double escape for windows backslash path separators
++-string(REPLACE "\\" "\\\\" prefix ${prefix})
+++string(REPLACE "\\" "\\\\" prefix "${prefix}")
++
++ configure_file(
++ ${CMAKE_CURRENT_SOURCE_DIR}/constants.c.in
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 0d1065854848494f211c990ed26267565cc44647 Mon Sep 17 00:00:00 2001
++From: Josh Blum <josh@joshknows.com>
++Date: Thu, 6 Oct 2016 15:06:09 -0700
++Subject: [PATCH 13/18] Support relocated install with VOLK_PREFIX env var
++
++Some packaging systems such as snaps will install
++the volk library to a dynamically chosen location.
++The install script can set an evironment variable
++so that the library reports the correct prefix.
++---
++ lib/constants.c.in | 3 +++
++ 1 file changed, 3 insertions(+)
++
++diff --git a/lib/constants.c.in b/lib/constants.c.in
++index 3839f53..a81c7cb 100644
++--- a/lib/constants.c.in
+++++ b/lib/constants.c.in
++@@ -24,11 +24,14 @@
++ #include <config.h>
++ #endif
++
+++#include <stdlib.h>
++ #include <volk/constants.h>
++
++ char*
++ volk_prefix()
++ {
+++ const char *prefix = getenv("VOLK_PREFIX");
+++ if (prefix != NULL) return (char *)prefix;
++ return "@prefix@";
++ }
++
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From ee70be38a66beb5eb236a3ffb3fc147a5d053979 Mon Sep 17 00:00:00 2001
++From: =?UTF-8?q?Marcus=20M=C3=BCller?= <marcus@hostalia.de>
++Date: Mon, 20 Nov 2017 15:12:06 +0100
++Subject: [PATCH 14/18] Fixing a minimal bug in the log2 docstring
++
++---
++ kernels/volk/volk_32f_log2_32f.h | 2 +-
++ 1 file changed, 1 insertion(+), 1 deletion(-)
++
++diff --git a/kernels/volk/volk_32f_log2_32f.h b/kernels/volk/volk_32f_log2_32f.h
++index 6704129..c3bfeaa 100644
++--- a/kernels/volk/volk_32f_log2_32f.h
+++++ b/kernels/volk/volk_32f_log2_32f.h
++@@ -62,7 +62,7 @@
++ * \li num_points: The number of data points.
++ *
++ * \b Outputs
++- * \li cVector: The output vector.
+++ * \li bVector: The output vector.
++ *
++ * \b Example
++ * \code
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 82a88672d80ef7652548182e819e726874e0adc0 Mon Sep 17 00:00:00 2001
++From: Damian Miralles <dmiralles2009@gmail.com>
++Date: Wed, 13 Dec 2017 13:27:17 -0700
++Subject: [PATCH 15/18] kernel: Adds unaligned protokernles to
++ `32f_x2_s32f_interleave_16ic` and `32f_x2_subtract_32f`
++
++Adds unaligned versions to the afore mentioned kernels, relative speeds
++improvements shown in both cases.
++---
++ kernels/volk/volk_32f_x2_s32f_interleave_16ic.h | 63 +++++++++++++++++++++++++
++ kernels/volk/volk_32f_x2_subtract_32f.h | 45 ++++++++++++++++++
++ 2 files changed, 108 insertions(+)
++
++diff --git a/kernels/volk/volk_32f_x2_s32f_interleave_16ic.h b/kernels/volk/volk_32f_x2_s32f_interleave_16ic.h
++index 99f1b5e..20f66ff 100644
++--- a/kernels/volk/volk_32f_x2_s32f_interleave_16ic.h
+++++ b/kernels/volk/volk_32f_x2_s32f_interleave_16ic.h
++@@ -214,3 +214,66 @@ volk_32f_x2_s32f_interleave_16ic_generic(lv_16sc_t* complexVector, const float*
++
++
++ #endif /* INCLUDED_volk_32f_x2_s32f_interleave_16ic_a_H */
+++
+++#ifndef INCLUDED_volk_32f_x2_s32f_interleave_16ic_u_H
+++#define INCLUDED_volk_32f_x2_s32f_interleave_16ic_u_H
+++
+++#include <volk/volk_common.h>
+++#include <inttypes.h>
+++#include <stdio.h>
+++
+++#ifdef LV_HAVE_AVX2
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_s32f_interleave_16ic_u_avx2(lv_16sc_t* complexVector, const float* iBuffer,
+++ const float* qBuffer, const float scalar, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const float* iBufferPtr = iBuffer;
+++ const float* qBufferPtr = qBuffer;
+++
+++ __m256 vScalar = _mm256_set1_ps(scalar);
+++
+++ const unsigned int eighthPoints = num_points / 8;
+++
+++ __m256 iValue, qValue, cplxValue1, cplxValue2;
+++ __m256i intValue1, intValue2;
+++
+++ int16_t* complexVectorPtr = (int16_t*)complexVector;
+++
+++ for(;number < eighthPoints; number++){
+++ iValue = _mm256_loadu_ps(iBufferPtr);
+++ qValue = _mm256_loadu_ps(qBufferPtr);
+++
+++ // Interleaves the lower two values in the i and q variables into one buffer
+++ cplxValue1 = _mm256_unpacklo_ps(iValue, qValue);
+++ cplxValue1 = _mm256_mul_ps(cplxValue1, vScalar);
+++
+++ // Interleaves the upper two values in the i and q variables into one buffer
+++ cplxValue2 = _mm256_unpackhi_ps(iValue, qValue);
+++ cplxValue2 = _mm256_mul_ps(cplxValue2, vScalar);
+++
+++ intValue1 = _mm256_cvtps_epi32(cplxValue1);
+++ intValue2 = _mm256_cvtps_epi32(cplxValue2);
+++
+++ intValue1 = _mm256_packs_epi32(intValue1, intValue2);
+++
+++ _mm256_storeu_si256((__m256i*)complexVectorPtr, intValue1);
+++ complexVectorPtr += 16;
+++
+++ iBufferPtr += 8;
+++ qBufferPtr += 8;
+++ }
+++
+++ number = eighthPoints * 8;
+++ complexVectorPtr = (int16_t*)(&complexVector[number]);
+++ for(; number < num_points; number++){
+++ *complexVectorPtr++ = (int16_t)(*iBufferPtr++ * scalar);
+++ *complexVectorPtr++ = (int16_t)(*qBufferPtr++ * scalar);
+++ }
+++}
+++#endif /* LV_HAVE_AVX2 */
+++
+++
+++#endif /* INCLUDED_volk_32f_x2_s32f_interleave_16ic_u_H */
++diff --git a/kernels/volk/volk_32f_x2_subtract_32f.h b/kernels/volk/volk_32f_x2_subtract_32f.h
++index 4a452fd..b7f36cf 100644
++--- a/kernels/volk/volk_32f_x2_subtract_32f.h
+++++ b/kernels/volk/volk_32f_x2_subtract_32f.h
++@@ -176,3 +176,48 @@ volk_32f_x2_subtract_32f_u_orc(float* cVector, const float* aVector,
++
++
++ #endif /* INCLUDED_volk_32f_x2_subtract_32f_a_H */
+++
+++
+++#ifndef INCLUDED_volk_32f_x2_subtract_32f_u_H
+++#define INCLUDED_volk_32f_x2_subtract_32f_u_H
+++
+++#include <inttypes.h>
+++#include <stdio.h>
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_subtract_32f_u_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eighthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr = bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eighthPoints; number++){
+++
+++ aVal = _mm256_loadu_ps(aPtr);
+++ bVal = _mm256_loadu_ps(bPtr);
+++
+++ cVal = _mm256_sub_ps(aVal, bVal);
+++
+++ _mm256_storeu_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eighthPoints * 8;
+++ for(;number < num_points; number++){
+++ *cPtr++ = (*aPtr++) - (*bPtr++);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++#endif /* INCLUDED_volk_32f_x2_subtract_32f_u_H */
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 940489f72b2c80f6b5dc514401773bf67a992f23 Mon Sep 17 00:00:00 2001
++From: Damian Miralles <damian.miralles@colorado.edu>
++Date: Fri, 15 Dec 2017 23:05:58 -0700
++Subject: [PATCH 16/18] kernels: Adds AVX support to `volk_32f_*` kernels
++
++Adds AVX support to `volk_32f_s32f_normalize`,`volk_32f_s32f_stddev_32f`,
++`volk_32f_sqrt_32f`, `volk_32f_x2_max_32f` and `volk_32f_x2_min_32f`.
++Some speed improvements can be seen with the new protokernel addition.
++---
++ kernels/volk/volk_32f_s32f_normalize.h | 74 ++++++++++++++++++++++++++++-
++ kernels/volk/volk_32f_s32f_stddev_32f.h | 59 +++++++++++++++++++++++
++ kernels/volk/volk_32f_sqrt_32f.h | 33 +++++++++++++
++ kernels/volk/volk_32f_x2_max_32f.h | 84 +++++++++++++++++++++++++++++++++
++ kernels/volk/volk_32f_x2_min_32f.h | 84 +++++++++++++++++++++++++++++++++
++ 5 files changed, 333 insertions(+), 1 deletion(-)
++
++diff --git a/kernels/volk/volk_32f_s32f_normalize.h b/kernels/volk/volk_32f_s32f_normalize.h
++index 52bf006..17d9da9 100644
++--- a/kernels/volk/volk_32f_s32f_normalize.h
+++++ b/kernels/volk/volk_32f_s32f_normalize.h
++@@ -105,6 +105,39 @@ static inline void volk_32f_s32f_normalize_a_sse(float* vecBuffer, const float s
++ }
++ #endif /* LV_HAVE_SSE */
++
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void volk_32f_s32f_normalize_a_avx(float* vecBuffer, const float scalar, unsigned int num_points){
+++ unsigned int number = 0;
+++ float* inputPtr = vecBuffer;
+++
+++ const float invScalar = 1.0 / scalar;
+++ __m256 vecScalar = _mm256_set1_ps(invScalar);
+++ __m256 input1;
+++
+++ const uint64_t eigthPoints = num_points / 8;
+++ for(;number < eigthPoints; number++){
+++
+++ input1 = _mm256_load_ps(inputPtr);
+++
+++ input1 = _mm256_mul_ps(input1, vecScalar);
+++
+++ _mm256_store_ps(inputPtr, input1);
+++
+++ inputPtr += 8;
+++ }
+++
+++ number = eigthPoints*8;
+++ for(; number < num_points; number++){
+++ *inputPtr *= invScalar;
+++ inputPtr++;
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++
++ #ifdef LV_HAVE_GENERIC
++
++ static inline void volk_32f_s32f_normalize_generic(float* vecBuffer, const float scalar, unsigned int num_points){
++@@ -128,6 +161,45 @@ static inline void volk_32f_s32f_normalize_u_orc(float* vecBuffer, const float s
++ #endif /* LV_HAVE_GENERIC */
++
++
+++#endif /* INCLUDED_volk_32f_s32f_normalize_a_H */
++
++
++-#endif /* INCLUDED_volk_32f_s32f_normalize_a_H */
+++#ifndef INCLUDED_volk_32f_s32f_normalize_u_H
+++#define INCLUDED_volk_32f_s32f_normalize_u_H
+++
+++#include <inttypes.h>
+++#include <stdio.h>
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void volk_32f_s32f_normalize_u_avx(float* vecBuffer, const float scalar, unsigned int num_points){
+++ unsigned int number = 0;
+++ float* inputPtr = vecBuffer;
+++
+++ const float invScalar = 1.0 / scalar;
+++ __m256 vecScalar = _mm256_set1_ps(invScalar);
+++ __m256 input1;
+++
+++ const uint64_t eigthPoints = num_points / 8;
+++ for(;number < eigthPoints; number++){
+++
+++ input1 = _mm256_loadu_ps(inputPtr);
+++
+++ input1 = _mm256_mul_ps(input1, vecScalar);
+++
+++ _mm256_storeu_ps(inputPtr, input1);
+++
+++ inputPtr += 8;
+++ }
+++
+++ number = eigthPoints*8;
+++ for(; number < num_points; number++){
+++ *inputPtr *= invScalar;
+++ inputPtr++;
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++
+++#endif /* INCLUDED_volk_32f_s32f_normalize_u_H */
++diff --git a/kernels/volk/volk_32f_s32f_stddev_32f.h b/kernels/volk/volk_32f_s32f_stddev_32f.h
++index 30f0ed6..f97a783 100644
++--- a/kernels/volk/volk_32f_s32f_stddev_32f.h
+++++ b/kernels/volk/volk_32f_s32f_stddev_32f.h
++@@ -132,6 +132,65 @@ volk_32f_s32f_stddev_32f_a_sse4_1(float* stddev, const float* inputBuffer,
++ #endif /* LV_HAVE_SSE4_1 */
++
++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_s32f_stddev_32f_a_avx(float* stddev, const float* inputBuffer,
+++ const float mean, unsigned int num_points)
+++{
+++ float returnValue = 0;
+++ if(num_points > 0){
+++ unsigned int number = 0;
+++ const unsigned int thirtySecondPoints = num_points / 32;
+++
+++ const float* aPtr = inputBuffer;
+++
+++ __VOLK_ATTR_ALIGNED(32) float squareBuffer[8];
+++
+++ __m256 squareAccumulator = _mm256_setzero_ps();
+++ __m256 aVal1, aVal2, aVal3, aVal4;
+++ __m256 cVal1, cVal2, cVal3, cVal4;
+++ for(;number < thirtySecondPoints; number++) {
+++ aVal1 = _mm256_load_ps(aPtr); aPtr += 8;
+++ cVal1 = _mm256_dp_ps(aVal1, aVal1, 0xF1);
+++
+++ aVal2 = _mm256_load_ps(aPtr); aPtr += 8;
+++ cVal2 = _mm256_dp_ps(aVal2, aVal2, 0xF2);
+++
+++ aVal3 = _mm256_load_ps(aPtr); aPtr += 8;
+++ cVal3 = _mm256_dp_ps(aVal3, aVal3, 0xF4);
+++
+++ aVal4 = _mm256_load_ps(aPtr); aPtr += 8;
+++ cVal4 = _mm256_dp_ps(aVal4, aVal4, 0xF8);
+++
+++ cVal1 = _mm256_or_ps(cVal1, cVal2);
+++ cVal3 = _mm256_or_ps(cVal3, cVal4);
+++ cVal1 = _mm256_or_ps(cVal1, cVal3);
+++
+++ squareAccumulator = _mm256_add_ps(squareAccumulator, cVal1); // squareAccumulator += x^2
+++ }
+++ _mm256_store_ps(squareBuffer,squareAccumulator); // Store the results back into the C container
+++ returnValue = squareBuffer[0]; returnValue += squareBuffer[1];
+++ returnValue += squareBuffer[2]; returnValue += squareBuffer[3];
+++ returnValue += squareBuffer[4]; returnValue += squareBuffer[5];
+++ returnValue += squareBuffer[6]; returnValue += squareBuffer[7];
+++
+++ number = thirtySecondPoints * 32;
+++ for(;number < num_points; number++){
+++ returnValue += (*aPtr) * (*aPtr);
+++ aPtr++;
+++ }
+++ returnValue /= num_points;
+++ returnValue -= (mean * mean);
+++ returnValue = sqrtf(returnValue);
+++ }
+++ *stddev = returnValue;
+++}
+++
+++#endif /* LV_HAVE_AVX */
+++
+++
++ #ifdef LV_HAVE_SSE
++ #include <xmmintrin.h>
++
++diff --git a/kernels/volk/volk_32f_sqrt_32f.h b/kernels/volk/volk_32f_sqrt_32f.h
++index a5851a0..174f8e3 100644
++--- a/kernels/volk/volk_32f_sqrt_32f.h
+++++ b/kernels/volk/volk_32f_sqrt_32f.h
++@@ -102,6 +102,39 @@ volk_32f_sqrt_32f_a_sse(float* cVector, const float* aVector, unsigned int num_p
++ #endif /* LV_HAVE_SSE */
++
++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_sqrt_32f_a_avx(float* cVector, const float* aVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++
+++ __m256 aVal, cVal;
+++ for(;number < eigthPoints; number++) {
+++ aVal = _mm256_load_ps(aPtr);
+++
+++ cVal = _mm256_sqrt_ps(aVal);
+++
+++ _mm256_store_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++) {
+++ *cPtr++ = sqrtf(*aPtr++);
+++ }
+++}
+++
+++#endif /* LV_HAVE_AVX */
+++
+++
++ #ifdef LV_HAVE_NEON
++ #include <arm_neon.h>
++
++diff --git a/kernels/volk/volk_32f_x2_max_32f.h b/kernels/volk/volk_32f_x2_max_32f.h
++index 14747c2..1dc0f7d 100644
++--- a/kernels/volk/volk_32f_x2_max_32f.h
+++++ b/kernels/volk/volk_32f_x2_max_32f.h
++@@ -112,6 +112,44 @@ volk_32f_x2_max_32f_a_sse(float* cVector, const float* aVector,
++ #endif /* LV_HAVE_SSE */
++
++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_max_32f_a_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr= bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eigthPoints; number++){
+++ aVal = _mm256_load_ps(aPtr);
+++ bVal = _mm256_load_ps(bPtr);
+++
+++ cVal = _mm256_max_ps(aVal, bVal);
+++
+++ _mm256_store_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++){
+++ const float a = *aPtr++;
+++ const float b = *bPtr++;
+++ *cPtr++ = ( a > b ? a : b);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++
++ #ifdef LV_HAVE_NEON
++ #include <arm_neon.h>
++
++@@ -180,3 +218,49 @@ volk_32f_x2_max_32f_u_orc(float* cVector, const float* aVector,
++
++
++ #endif /* INCLUDED_volk_32f_x2_max_32f_a_H */
+++
+++
+++#ifndef INCLUDED_volk_32f_x2_max_32f_u_H
+++#define INCLUDED_volk_32f_x2_max_32f_u_H
+++
+++#include <inttypes.h>
+++#include <stdio.h>
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_max_32f_u_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr= bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eigthPoints; number++){
+++ aVal = _mm256_loadu_ps(aPtr);
+++ bVal = _mm256_loadu_ps(bPtr);
+++
+++ cVal = _mm256_max_ps(aVal, bVal);
+++
+++ _mm256_storeu_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++){
+++ const float a = *aPtr++;
+++ const float b = *bPtr++;
+++ *cPtr++ = ( a > b ? a : b);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++#endif /* INCLUDED_volk_32f_x2_max_32f_u_H */
++diff --git a/kernels/volk/volk_32f_x2_min_32f.h b/kernels/volk/volk_32f_x2_min_32f.h
++index f3cbae1..3beb5fa 100644
++--- a/kernels/volk/volk_32f_x2_min_32f.h
+++++ b/kernels/volk/volk_32f_x2_min_32f.h
++@@ -112,6 +112,44 @@ volk_32f_x2_min_32f_a_sse(float* cVector, const float* aVector,
++ #endif /* LV_HAVE_SSE */
++
++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_min_32f_a_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr= bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eigthPoints; number++){
+++ aVal = _mm256_load_ps(aPtr);
+++ bVal = _mm256_load_ps(bPtr);
+++
+++ cVal = _mm256_min_ps(aVal, bVal);
+++
+++ _mm256_store_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++){
+++ const float a = *aPtr++;
+++ const float b = *bPtr++;
+++ *cPtr++ = ( a < b ? a : b);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++
++ #ifdef LV_HAVE_NEON
++ #include <arm_neon.h>
++
++@@ -183,3 +221,49 @@ volk_32f_x2_min_32f_u_orc(float* cVector, const float* aVector,
++
++
++ #endif /* INCLUDED_volk_32f_x2_min_32f_a_H */
+++
+++
+++#ifndef INCLUDED_volk_32f_x2_min_32f_u_H
+++#define INCLUDED_volk_32f_x2_min_32f_u_H
+++
+++#include <inttypes.h>
+++#include <stdio.h>
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_min_32f_u_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr= bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eigthPoints; number++){
+++ aVal = _mm256_loadu_ps(aPtr);
+++ bVal = _mm256_loadu_ps(bPtr);
+++
+++ cVal = _mm256_min_ps(aVal, bVal);
+++
+++ _mm256_storeu_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++){
+++ const float a = *aPtr++;
+++ const float b = *bPtr++;
+++ *cPtr++ = ( a < b ? a : b);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++#endif /* INCLUDED_volk_32f_x2_min_32f_u_H */
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 0dd53d3ad8e24a833342b369743f274a15a66274 Mon Sep 17 00:00:00 2001
++From: Damian Miralles <damian.miralles@colorado.edu>
++Date: Wed, 20 Dec 2017 21:01:52 -0700
++Subject: [PATCH 17/18] kernels: Add AVX support to
++ `32f_x2_divide_32f`,`32f_x2_dot_prod_16i`
++
++Adds protokernels for AVX support. Modest speed improvements in some of
++the kernels, however, it seems to be related to the host architecture
++being used
++---
++ kernels/volk/volk_32f_x2_divide_32f.h | 80 +++++++++++++++++
++ kernels/volk/volk_32f_x2_dot_prod_16i.h | 148 ++++++++++++++++++++++++++++++++
++ 2 files changed, 228 insertions(+)
++
++diff --git a/kernels/volk/volk_32f_x2_divide_32f.h b/kernels/volk/volk_32f_x2_divide_32f.h
++index d724173..7cc34ca 100644
++--- a/kernels/volk/volk_32f_x2_divide_32f.h
+++++ b/kernels/volk/volk_32f_x2_divide_32f.h
++@@ -110,6 +110,42 @@ volk_32f_x2_divide_32f_a_sse(float* cVector, const float* aVector,
++ #endif /* LV_HAVE_SSE */
++
++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_divide_32f_a_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr= bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eigthPoints; number++){
+++ aVal = _mm256_load_ps(aPtr);
+++ bVal = _mm256_load_ps(bPtr);
+++
+++ cVal = _mm256_div_ps(aVal, bVal);
+++
+++ _mm256_store_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++){
+++ *cPtr++ = (*aPtr++) / (*bPtr++);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++
++ #ifdef LV_HAVE_GENERIC
++
++ static inline void
++@@ -145,3 +181,47 @@ volk_32f_x2_divide_32f_u_orc(float* cVector, const float* aVector,
++
++
++ #endif /* INCLUDED_volk_32f_x2_divide_32f_a_H */
+++
+++
+++#ifndef INCLUDED_volk_32f_x2_divide_32f_u_H
+++#define INCLUDED_volk_32f_x2_divide_32f_u_H
+++
+++#include <inttypes.h>
+++#include <stdio.h>
+++
+++#ifdef LV_HAVE_AVX
+++#include <immintrin.h>
+++
+++static inline void
+++volk_32f_x2_divide_32f_u_avx(float* cVector, const float* aVector,
+++ const float* bVector, unsigned int num_points)
+++{
+++ unsigned int number = 0;
+++ const unsigned int eigthPoints = num_points / 8;
+++
+++ float* cPtr = cVector;
+++ const float* aPtr = aVector;
+++ const float* bPtr= bVector;
+++
+++ __m256 aVal, bVal, cVal;
+++ for(;number < eigthPoints; number++){
+++ aVal = _mm256_loadu_ps(aPtr);
+++ bVal = _mm256_loadu_ps(bPtr);
+++
+++ cVal = _mm256_div_ps(aVal, bVal);
+++
+++ _mm256_storeu_ps(cPtr,cVal); // Store the results back into the C container
+++
+++ aPtr += 8;
+++ bPtr += 8;
+++ cPtr += 8;
+++ }
+++
+++ number = eigthPoints * 8;
+++ for(;number < num_points; number++){
+++ *cPtr++ = (*aPtr++) / (*bPtr++);
+++ }
+++}
+++#endif /* LV_HAVE_AVX */
+++
+++#endif /* INCLUDED_volk_32f_x2_divide_32f_u_H */
++diff --git a/kernels/volk/volk_32f_x2_dot_prod_16i.h b/kernels/volk/volk_32f_x2_dot_prod_16i.h
++index 15f01b7..a1279cf 100644
++--- a/kernels/volk/volk_32f_x2_dot_prod_16i.h
+++++ b/kernels/volk/volk_32f_x2_dot_prod_16i.h
++@@ -82,6 +82,154 @@ static inline void volk_32f_x2_dot_prod_16i_generic(int16_t* result, const float
++ #endif /*LV_HAVE_GENERIC*/
++
++
+++#ifdef LV_HAVE_AVX
+++
+++static inline void volk_32f_x2_dot_prod_16i_a_avx(int16_t* result, const float* input, const float* taps, unsigned int num_points) {
+++
+++ unsigned int number = 0;
+++ const unsigned int thirtySecondPoints = num_points / 32;
+++
+++ float dotProduct = 0;
+++ const float* aPtr = input;
+++ const float* bPtr = taps;
+++
+++ __m256 a0Val, a1Val, a2Val, a3Val;
+++ __m256 b0Val, b1Val, b2Val, b3Val;
+++ __m256 c0Val, c1Val, c2Val, c3Val;
+++
+++ __m256 dotProdVal0 = _mm256_setzero_ps();
+++ __m256 dotProdVal1 = _mm256_setzero_ps();
+++ __m256 dotProdVal2 = _mm256_setzero_ps();
+++ __m256 dotProdVal3 = _mm256_setzero_ps();
+++
+++ for(;number < thirtySecondPoints; number++){
+++
+++ a0Val = _mm256_load_ps(aPtr);
+++ a1Val = _mm256_load_ps(aPtr+8);
+++ a2Val = _mm256_load_ps(aPtr+16);
+++ a3Val = _mm256_load_ps(aPtr+24);
+++
+++ b0Val = _mm256_load_ps(bPtr);
+++ b1Val = _mm256_load_ps(bPtr+8);
+++ b2Val = _mm256_load_ps(bPtr+16);
+++ b3Val = _mm256_load_ps(bPtr+24);
+++
+++ c0Val = _mm256_mul_ps(a0Val, b0Val);
+++ c1Val = _mm256_mul_ps(a1Val, b1Val);
+++ c2Val = _mm256_mul_ps(a2Val, b2Val);
+++ c3Val = _mm256_mul_ps(a3Val, b3Val);
+++
+++ dotProdVal0 = _mm256_add_ps(c0Val, dotProdVal0);
+++ dotProdVal1 = _mm256_add_ps(c1Val, dotProdVal1);
+++ dotProdVal2 = _mm256_add_ps(c2Val, dotProdVal2);
+++ dotProdVal3 = _mm256_add_ps(c3Val, dotProdVal3);
+++
+++ aPtr += 32;
+++ bPtr += 32;
+++ }
+++
+++ dotProdVal0 = _mm256_add_ps(dotProdVal0, dotProdVal1);
+++ dotProdVal0 = _mm256_add_ps(dotProdVal0, dotProdVal2);
+++ dotProdVal0 = _mm256_add_ps(dotProdVal0, dotProdVal3);
+++
+++ __VOLK_ATTR_ALIGNED(32) float dotProductVector[8];
+++
+++ _mm256_store_ps(dotProductVector,dotProdVal0); // Store the results back into the dot product vector
+++
+++ dotProduct = dotProductVector[0];
+++ dotProduct += dotProductVector[1];
+++ dotProduct += dotProductVector[2];
+++ dotProduct += dotProductVector[3];
+++ dotProduct += dotProductVector[4];
+++ dotProduct += dotProductVector[5];
+++ dotProduct += dotProductVector[6];
+++ dotProduct += dotProductVector[7];
+++
+++ number = thirtySecondPoints*32;
+++ for(;number < num_points; number++){
+++ dotProduct += ((*aPtr++) * (*bPtr++));
+++ }
+++
+++ *result = (short)dotProduct;
+++}
+++
+++#endif /*LV_HAVE_AVX*/
+++
+++
+++#ifdef LV_HAVE_AVX
+++
+++static inline void volk_32f_x2_dot_prod_16i_u_avx(int16_t* result, const float* input, const float* taps, unsigned int num_points) {
+++
+++ unsigned int number = 0;
+++ const unsigned int thirtySecondPoints = num_points / 32;
+++
+++ float dotProduct = 0;
+++ const float* aPtr = input;
+++ const float* bPtr = taps;
+++
+++ __m256 a0Val, a1Val, a2Val, a3Val;
+++ __m256 b0Val, b1Val, b2Val, b3Val;
+++ __m256 c0Val, c1Val, c2Val, c3Val;
+++
+++ __m256 dotProdVal0 = _mm256_setzero_ps();
+++ __m256 dotProdVal1 = _mm256_setzero_ps();
+++ __m256 dotProdVal2 = _mm256_setzero_ps();
+++ __m256 dotProdVal3 = _mm256_setzero_ps();
+++
+++ for(;number < thirtySecondPoints; number++){
+++
+++ a0Val = _mm256_loadu_ps(aPtr);
+++ a1Val = _mm256_loadu_ps(aPtr+8);
+++ a2Val = _mm256_loadu_ps(aPtr+16);
+++ a3Val = _mm256_loadu_ps(aPtr+24);
+++
+++ b0Val = _mm256_loadu_ps(bPtr);
+++ b1Val = _mm256_loadu_ps(bPtr+8);
+++ b2Val = _mm256_loadu_ps(bPtr+16);
+++ b3Val = _mm256_loadu_ps(bPtr+24);
+++
+++ c0Val = _mm256_mul_ps(a0Val, b0Val);
+++ c1Val = _mm256_mul_ps(a1Val, b1Val);
+++ c2Val = _mm256_mul_ps(a2Val, b2Val);
+++ c3Val = _mm256_mul_ps(a3Val, b3Val);
+++
+++ dotProdVal0 = _mm256_add_ps(c0Val, dotProdVal0);
+++ dotProdVal1 = _mm256_add_ps(c1Val, dotProdVal1);
+++ dotProdVal2 = _mm256_add_ps(c2Val, dotProdVal2);
+++ dotProdVal3 = _mm256_add_ps(c3Val, dotProdVal3);
+++
+++ aPtr += 32;
+++ bPtr += 32;
+++ }
+++
+++ dotProdVal0 = _mm256_add_ps(dotProdVal0, dotProdVal1);
+++ dotProdVal0 = _mm256_add_ps(dotProdVal0, dotProdVal2);
+++ dotProdVal0 = _mm256_add_ps(dotProdVal0, dotProdVal3);
+++
+++ __VOLK_ATTR_ALIGNED(32) float dotProductVector[8];
+++
+++ _mm256_storeu_ps(dotProductVector,dotProdVal0); // Store the results back into the dot product vector
+++
+++ dotProduct = dotProductVector[0];
+++ dotProduct += dotProductVector[1];
+++ dotProduct += dotProductVector[2];
+++ dotProduct += dotProductVector[3];
+++ dotProduct += dotProductVector[4];
+++ dotProduct += dotProductVector[5];
+++ dotProduct += dotProductVector[6];
+++ dotProduct += dotProductVector[7];
+++
+++ number = thirtySecondPoints*32;
+++ for(;number < num_points; number++){
+++ dotProduct += ((*aPtr++) * (*bPtr++));
+++ }
+++
+++ *result = (short)dotProduct;
+++}
+++
+++#endif /*LV_HAVE_AVX*/
+++
+++
++ #ifdef LV_HAVE_SSE
++
++ static inline void volk_32f_x2_dot_prod_16i_a_sse(int16_t* result, const float* input, const float* taps, unsigned int num_points) {
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From 0109b2ed06f907363d3ea5a05d24db4992e2d1a5 Mon Sep 17 00:00:00 2001
++From: Nathan West <nwest@deepsig.io>
++Date: Tue, 23 Jan 2018 12:02:03 -0500
++Subject: [PATCH 18/18] fix GH issue #139 for 32fc_index_max_* kernels
++
++---
++ kernels/volk/volk_32fc_index_max_16u.h | 3 +--
++ kernels/volk/volk_32fc_index_max_32u.h | 2 +-
++ 2 files changed, 2 insertions(+), 3 deletions(-)
++
++diff --git a/kernels/volk/volk_32fc_index_max_16u.h b/kernels/volk/volk_32fc_index_max_16u.h
++index c13196a..14b0d22 100644
++--- a/kernels/volk/volk_32fc_index_max_16u.h
+++++ b/kernels/volk/volk_32fc_index_max_16u.h
++@@ -115,10 +115,9 @@ volk_32fc_index_max_16u_a_sse3(uint16_t* target, lv_32fc_t* src0,
++ int i = 0;
++
++ xmm8 = _mm_set_epi32(3, 2, 1, 0);//remember the crazy reverse order!
++- xmm9 = xmm8 = _mm_setzero_si128();
+++ xmm9 = _mm_setzero_si128();
++ xmm10 = _mm_set_epi32(4, 4, 4, 4);
++ xmm3 = _mm_setzero_ps();
++-
++ //printf("%f, %f, %f, %f\n", ((float*)&xmm10)[0], ((float*)&xmm10)[1], ((float*)&xmm10)[2], ((float*)&xmm10)[3]);
++
++ for(; i < bound; ++i) {
++diff --git a/kernels/volk/volk_32fc_index_max_32u.h b/kernels/volk/volk_32fc_index_max_32u.h
++index ad794fb..5665582 100644
++--- a/kernels/volk/volk_32fc_index_max_32u.h
+++++ b/kernels/volk/volk_32fc_index_max_32u.h
++@@ -104,7 +104,7 @@ volk_32fc_index_max_32u_a_sse3(uint32_t* target, lv_32fc_t* src0,
++ int i = 0;
++
++ xmm8 = _mm_set_epi32(3, 2, 1, 0);//remember the crazy reverse order!
++- xmm9 = xmm8 = _mm_setzero_si128();
+++ xmm9 = _mm_setzero_si128();
++ xmm10 = _mm_set_epi32(4, 4, 4, 4);
++ xmm3 = _mm_setzero_ps();
++
++--
++2.11.0
++
--- /dev/null
--- /dev/null
++From: A. Maitland Bottoms <bottoms@debian.org>
++Subject: install all headers
++
++(Along with some sorting)
++
++--- a/CMakeLists.txt
+++++ b/CMakeLists.txt
++@@ -158,17 +158,20 @@
++ )
++
++ install(FILES
++- ${CMAKE_SOURCE_DIR}/include/volk/volk_prefs.h
++- ${CMAKE_SOURCE_DIR}/include/volk/volk_complex.h
++- ${CMAKE_SOURCE_DIR}/include/volk/volk_common.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/constants.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/saturation_arithmetic.h
++ ${CMAKE_SOURCE_DIR}/include/volk/volk_avx_intrinsics.h
++- ${CMAKE_SOURCE_DIR}/include/volk/volk_sse3_intrinsics.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/volk_common.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/volk_complex.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/volk_malloc.h
++ ${CMAKE_SOURCE_DIR}/include/volk/volk_neon_intrinsics.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/volk_prefs.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/volk_sse3_intrinsics.h
+++ ${CMAKE_SOURCE_DIR}/include/volk/volk_sse_intrinsics.h
++ ${CMAKE_BINARY_DIR}/include/volk/volk.h
++ ${CMAKE_BINARY_DIR}/include/volk/volk_cpu.h
++ ${CMAKE_BINARY_DIR}/include/volk/volk_config_fixed.h
++ ${CMAKE_BINARY_DIR}/include/volk/volk_typedefs.h
++- ${CMAKE_SOURCE_DIR}/include/volk/volk_malloc.h
++ DESTINATION include/volk
++ COMPONENT "volk_devel"
++ )
--- /dev/null
--- /dev/null
++--- a/lib/CMakeLists.txt
+++++ b/lib/CMakeLists.txt
++@@ -544,7 +544,7 @@
++
++ #Add dynamic library
++ add_library(volk SHARED $<TARGET_OBJECTS:volk_obj>)
++- target_link_libraries(volk ${volk_libraries})
+++ target_link_libraries(volk ${volk_libraries} m)
++
++ #Configure target properties
++ set_target_properties(volk_obj PROPERTIES COMPILE_FLAGS "-fPIC")
--- /dev/null
--- /dev/null
++From: A. Maitland Bottoms <bottoms@debian.org>
++Subject: make acc happy
++
++The abi-compliance-checker grabs all the .h files it finds
++and tries to compile them all. Even though some are not
++appropriate for the architecture being run on. Being careful
++with preprocessor protections avoids probplems.
++
++--- a/kernels/volk/volk_32f_8u_polarbutterflypuppet_32f.h
+++++ b/kernels/volk/volk_32f_8u_polarbutterflypuppet_32f.h
++@@ -31,6 +31,7 @@
++ #include <volk/volk_32f_8u_polarbutterfly_32f.h>
++ #include <volk/volk_8u_x3_encodepolar_8u_x2.h>
++ #include <volk/volk_8u_x3_encodepolarpuppet_8u.h>
+++#include <volk/volk_8u_x2_encodeframepolar_8u.h>
++
++
++ static inline void
++--- a/include/volk/volk_neon_intrinsics.h
+++++ b/include/volk/volk_neon_intrinsics.h
++@@ -27,6 +27,7 @@
++
++ #ifndef INCLUDE_VOLK_VOLK_NEON_INTRINSICS_H_
++ #define INCLUDE_VOLK_VOLK_NEON_INTRINSICS_H_
+++#ifdef LV_HAVE_NEON
++ #include <arm_neon.h>
++
++ static inline float32x4_t
++@@ -119,4 +120,5 @@
++ return log2_approx;
++ }
++
+++#endif /*LV_HAVE_NEON*/
++ #endif /* INCLUDE_VOLK_VOLK_NEON_INTRINSICS_H_ */
++--- a/kernels/volk/volk_8u_x2_encodeframepolar_8u.h
+++++ b/kernels/volk/volk_8u_x2_encodeframepolar_8u.h
++@@ -58,8 +58,6 @@
++ }
++ }
++
++-#ifdef LV_HAVE_GENERIC
++-
++ static inline void
++ volk_8u_x2_encodeframepolar_8u_generic(unsigned char* frame, unsigned char* temp,
++ unsigned int frame_size)
++@@ -79,7 +77,6 @@
++ --stage;
++ }
++ }
++-#endif /* LV_HAVE_GENERIC */
++
++ #ifdef LV_HAVE_SSSE3
++ #include <tmmintrin.h>
--- /dev/null
--- /dev/null
++From: A. Maitland Bottoms <bottoms@debian.org>
++Subject: native armv7 build support
++
++Debian, unlike other GNU Radio deployments, does not cross-compile
++packages, but builds natively on a set of build machines, including
++both arm and armhf.
++
++--- a/lib/CMakeLists.txt
+++++ b/lib/CMakeLists.txt
++@@ -250,6 +250,13 @@
++ endif(NOT CPU_IS_x86)
++
++ ########################################################################
+++# if building Debian armel, eliminate neon
+++########################################################################
+++if(${CMAKE_LIBRARY_ARCHITECTURE} STREQUAL "arm-linux-gnueabi")
+++ OVERRULE_ARCH(neon "Architecture is not armhf")
+++endif(${CMAKE_LIBRARY_ARCHITECTURE} STREQUAL "arm-linux-gnueabi")
+++
+++########################################################################
++ # implement overruling in the ORC case,
++ # since ORC always passes flag detection
++ ########################################################################
++@@ -414,7 +421,7 @@
++ # Handle ASM support
++ # on by default, but let users turn it off
++ ########################################################################
++-if(${CMAKE_VERSION} VERSION_GREATER "2.8.9")
+++if((${CMAKE_VERSION} VERSION_GREATER "2.8.9") AND NOT (${CMAKE_LIBRARY_ARCHITECTURE} STREQUAL "arm-linux-gnueabi"))
++ set(ASM_ARCHS_AVAILABLE "neon")
++
++ set(FULL_C_FLAGS "${CMAKE_C_FLAGS}" "${CMAKE_CXX_COMPILER_ARG1}")
++@@ -424,7 +431,7 @@
++ # set up the assembler flags and include the source files
++ foreach(ARCH ${ASM_ARCHS_AVAILABLE})
++ string(REGEX MATCH "${ARCH}" ASM_ARCH "${available_archs}")
++- if( ASM_ARCH STREQUAL "neon" )
+++ if(( ASM_ARCH STREQUAL "neon" ) OR ( ${CMAKE_SYSTEM_PROCESSOR} MATCHES "armv7"))
++ message(STATUS "---- Adding ASM files") # we always use ATT syntax
++ message(STATUS "-- Detected neon architecture; enabling ASM")
++ # setup architecture specific assembler flags
++@@ -443,7 +450,7 @@
++ message(STATUS "asm flags: ${CMAKE_ASM_FLAGS}")
++ endforeach(ARCH)
++
++-else(${CMAKE_VERSION} VERSION_GREATER "2.8.9")
+++else((${CMAKE_VERSION} VERSION_GREATER "2.8.9") AND NOT (${CMAKE_LIBRARY_ARCHITECTURE} STREQUAL "arm-linux-gnueabi"))
++ message(STATUS "Not enabling ASM support. CMake >= 2.8.10 required.")
++ foreach(machine_name ${available_machines})
++ string(REGEX MATCH "neon" NEON_MACHINE ${machine_name})
++@@ -451,7 +458,7 @@
++ message(FATAL_ERROR "CMake >= 2.8.10 is required for ARM NEON support")
++ endif()
++ endforeach()
++-endif(${CMAKE_VERSION} VERSION_GREATER "2.8.9")
+++endif((${CMAKE_VERSION} VERSION_GREATER "2.8.9") AND NOT (${CMAKE_LIBRARY_ARCHITECTURE} STREQUAL "arm-linux-gnueabi"))
++
++ ########################################################################
++ # Handle orc support
--- /dev/null
--- /dev/null
++0001-Add-a-AppVeyor-compatible-YAML-file-for-building-on-.patch
++0003-apps-fix-profile-update-reading-end-of-lines.patch
++0005-qa-lower-tolerance-for-32fc_mag-to-fix-issue-96.patch
++0006-Add-NEON-AVX-and-unaligned-versions-of-SSE4.1-and-SS.patch
++0007-added-__VOLK_PREFETCH-compatibility-macro.patch
++0008-Fix-bug-106-volk_64u_popcnt-bug-in-generic-implement.patch
++0009-modtool-deconflict-module-include-guards-from-main-v.patch
++0010-modtool-update-the-cmake-find-module-for-volk-mods.patch
++0011-Use-powf-to-match-variables-and-avoid-implicit-type-.patch
++0012-cmake-support-empty-CMAKE_INSTALL_PREFIX.patch
++0013-Support-relocated-install-with-VOLK_PREFIX-env-var.patch
++0014-Fixing-a-minimal-bug-in-the-log2-docstring.patch
++0015-kernel-Adds-unaligned-protokernles-to-32f_x2_s32f_in.patch
++0016-kernels-Adds-AVX-support-to-volk_32f_-kernels.patch
++0017-kernels-Add-AVX-support-to-32f_x2_divide_32f-32f_x2_.patch
++0018-fix-GH-issue-139-for-32fc_index_max_-kernels.patch
++native-armv7-build-support
++make-acc-happy
++sort-cmake-glob-lists
++install-all-headers
++sort-input-files.patch
++libm-link
--- /dev/null
--- /dev/null
++From: A. Maitland Bottoms <bottoms@debian.org>
++Subject sort cmake glob lists
++
++File lists are generated in a CMakeLists.txt file with file(GLOB ...), which varies
++with the readdir() order. Sorting the lists should help make reproducinble builds.
++
++--- a/lib/CMakeLists.txt
+++++ b/lib/CMakeLists.txt
++@@ -328,8 +328,11 @@
++
++ #dependencies are all python, xml, and header implementation files
++ file(GLOB xml_files ${PROJECT_SOURCE_DIR}/gen/*.xml)
+++list(SORT xml_files)
++ file(GLOB py_files ${PROJECT_SOURCE_DIR}/gen/*.py)
+++list(SORT py_files)
++ file(GLOB h_files ${PROJECT_SOURCE_DIR}/kernels/volk/*.h)
+++list(SORT h_files)
++
++ macro(gen_template tmpl output)
++ list(APPEND volk_gen_sources ${output})
--- /dev/null
--- /dev/null
++From f6dbb5f8ba840075dde9f0aa1cc48b805ea4d1c5 Mon Sep 17 00:00:00 2001
++From: "Bernhard M. Wiedemann" <bwiedemann@suse.de>
++Date: Mon, 5 Jun 2017 21:37:38 +0200
++Subject: [PATCH 2/5] sort input files
++
++when building packages (e.g. for openSUSE Linux)
++(random) filesystem order of input files
++influences ordering of entries in the output,
++thus without the patch, builds (in disposable VMs) would usually differ.
++
++See https://reproducible-builds.org/ for why this matters.
++---
++ gen/volk_kernel_defs.py | 2 +-
++ python/volk_modtool/volk_modtool_generate.py | 6 +++---
++ 2 files changed, 4 insertions(+), 4 deletions(-)
++
++--- a/gen/volk_kernel_defs.py
+++++ b/gen/volk_kernel_defs.py
++@@ -202,7 +202,7 @@
++ ########################################################################
++ __file__ = os.path.abspath(__file__)
++ srcdir = os.path.dirname(os.path.dirname(__file__))
++-kernel_files = glob.glob(os.path.join(srcdir, "kernels", "volk", "*.h"))
+++kernel_files = sorted(glob.glob(os.path.join(srcdir, "kernels", "volk", "*.h")))
++ kernels = map(kernel_class, kernel_files)
++
++ if __name__ == '__main__':
++--- a/python/volk_modtool/volk_modtool_generate.py
+++++ b/python/volk_modtool/volk_modtool_generate.py
++@@ -58,10 +58,10 @@
++ else:
++ name = self.get_basename(base)
++ if name == '':
++- hdr_files = glob.glob(os.path.join(base, "kernels/volk/*.h"))
+++ hdr_files = sorted(glob.glob(os.path.join(base, "kernels/volk/*.h")))
++ begins = re.compile("(?<=volk_).*")
++ else:
++- hdr_files = glob.glob(os.path.join(base, "kernels/volk_" + name + "/*.h"))
+++ hdr_files = sorted(glob.glob(os.path.join(base, "kernels/volk_" + name + "/*.h")))
++ begins = re.compile("(?<=volk_" + name + "_).*")
++
++ datatypes = []
++@@ -156,7 +156,7 @@
++ open(dest, 'w+').write(outstring)
++
++ # copy orc proto-kernels if they exist
++- for orcfile in glob.glob(inpath + '/kernels/volk/asm/orc/' + top + name + '*.orc'):
+++ for orcfile in sorted(glob.glob(inpath + '/kernels/volk/asm/orc/' + top + name + '*.orc')):
++ if os.path.isfile(orcfile):
++ instring = open(orcfile, 'r').read()
++ outstring = re.sub(oldvolk, 'volk_' + self.my_dict['name'], instring)
--- /dev/null
--- /dev/null
++#!/usr/bin/make -f
++DEB_HOST_MULTIARCH ?= $(shell dpkg-architecture -qDEB_HOST_MULTIARCH)
++export DEB_HOST_MULTIARCH
++DH_VERBOSE=1
++export DH_VERBOSE
++
++%:
++ dh $@ --with python2
++
++override_dh_auto_configure:
++ dh_auto_configure -- -DLIB_SUFFIX="/$(DEB_HOST_MULTIARCH)" -DPYTHON_EXECUTABLE=/usr/bin/python
++
++override_dh_auto_test:
++ - dh_auto_test -- CTEST_TEST_TIMEOUT=60
++
++override_dh_acc:
++ - dh_acc
--- /dev/null
--- /dev/null
++3.0 (quilt)
--- /dev/null
--- /dev/null
++debian/libvolk1-dev.abi.tar.gz.amd64
--- /dev/null
--- /dev/null
++-----BEGIN PGP PUBLIC KEY BLOCK-----
++Version: GnuPG v1
++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++=MlbD
++-----END PGP PUBLIC KEY BLOCK-----
--- /dev/null
--- /dev/null
++.\" DO NOT MODIFY THIS FILE! It was generated by help2man 1.40.10.
++.TH VOLK-CONFIG-INFO "1" "July 2014" "volk-config-info 0.1" "User Commands"
++.SH NAME
++volk-config-info \- pkgconfig-like tool for Vector Optimized Library of Kernels 0.1
++.SH DESCRIPTION
++.SS "Program options: volk-config-info [options]:"
++.TP
++\fB\-h\fR [ \fB\-\-help\fR ]
++print help message
++.TP
++\fB\-\-prefix\fR
++print VOLK installation prefix
++.TP
++\fB\-\-builddate\fR
++print VOLK build date (RFC2822 format)
++.TP
++\fB\-\-cc\fR
++print VOLK C compiler version
++.TP
++\fB\-\-cflags\fR
++print VOLK CFLAGS
++.TP
++\fB\-\-all\-machines\fR
++print VOLK machines built into library
++.TP
++\fB\-\-avail\-machines\fR
++print VOLK machines the current platform can use
++.TP
++\fB\-\-machine\fR
++print the VOLK machine that will be used
++.TP
++\fB\-v\fR [ \fB\-\-version\fR ]
++print VOLK version
++.SH "SEE ALSO"
++The full documentation for
++.B volk-config-info
++is maintained as a Texinfo manual. If the
++.B info
++and
++.B volk-config-info
++programs are properly installed at your site, the command
++.IP
++.B info volk-config-info
++.PP
++should give you access to the complete manual.
--- /dev/null
--- /dev/null
++.TH GNURADIO "1" "August 2013" "volk_modtool 3.7" "User Commands"
++.SH NAME
++volk_modtool \- tailor VOLK modules
++.SH DESCRIPTION
++The volk_modtool tool is installed along with VOLK as a way of helping
++to construct, add to, and interogate the VOLK library or companion
++libraries.
++.P
++volk_modtool is installed into $prefix/bin.
++.P
++VOLK modtool enables creating standalone (out-of-tree) VOLK modules
++and provides a few tools for sharing VOLK kernels between VOLK
++modules. If you need to design or work with VOLK kernels away from
++the canonical VOLK library, this is the tool. If you need to tailor
++your own VOLK library for whatever reason, this is the tool.
++.P
++The canonical VOLK library installs a volk.h and a libvolk.so. Your
++own library will install volk_$name.h and libvolk_$name.so. Ya Gronk?
++Good.
++.P
++There isn't a substantial difference between the canonical VOLK
++module and any other VOLK module. They're all peers. Any module
++created via VOLK modtool will come complete with a default
++volk_modtool.cfg file associating the module with the base from which
++it came, its distinctive $name and its destination (or path). These
++values (created from user input if VOLK modtool runs without a
++user-supplied config file or a default config file) serve as default
++values for some VOLK modtool actions. It's more or less intended for
++the user to change directories to the top level of a created VOLK
++module and then run volk_modtool to take advantage of the values
++stored in the default volk_modtool.cfg file.
++.P
++Apart from creating new VOLK modules, VOLK modtool allows you to list
++the names of kernels in other modules, list the names of kernels in
++the current module, add kernels from another module into the current
++module, and remove kernels from the current module. When moving
++kernels between modules, VOLK modtool does its best to keep the qa
++and profiling code for those kernels intact. If the base has a test
++or a profiling call for some kernel, those calls will follow the
++kernel when VOLK modtool adds that kernel. If QA or profiling
++requires a puppet kernel, the puppet kernel will follow the original
++kernel when VOLK modtool adds that original kernel. VOLK modtool
++respects puppets.
++.P
++======================================================================
++.P
++.SH Installing a new VOLK Library:
++.P
++Run the command "volk_modtool -i". This will ask you three questions:
++.P
++ name: // the name to give your VOLK library: volk_<name>
++ destination: // directory new source tree is built under -- must exists.
++ // It will create <directory>/volk_<name>
++ base: // the directory containing the original VOLK source code
++.P
++This will build a new skeleton directory in the destination provided
++with the name volk_<name>. It will contain the necessary structure to
++build:
++.P
++ mkdir build
++ cd build
++ cmake -DCMAKE_INSTALL_PREFIX=/opt/volk ../
++ make
++ sudo make install
++.P
++Right now, the library is empty and contains no kernels. Kernels can
++be added from another VOLK library using the '-a' option. If not
++specified, the kernel will be extracted from the base VOLK
++directory. Using the '-b' allows us to specify another VOLK library to
++use for this purpose.
++.P
++ volk_modtool -a -n 32fc_x2_conjugate_dot_prod_32fc
++.P
++This will put the code for the new kernel into
++<destination>/volk_<name>/kernels/volk_<name>/
++.P
++Other kernels must be added by hand. See the following webpages for
++more information about creating VOLK kernels:
++ http://gnuradio.org/doc/doxygen/volk_guide.html
++ http://gnuradio.org/redmine/projects/gnuradio/wiki/Volk
++.P
++======================================================================
++.P
++.SH OPTIONS
++.P
++Options for Adding and Removing Kernels:
++ -a, --add_kernel
++ Add kernel from existing VOLK module. Uses the base VOLK module
++ unless -b is used. Use -n to specify the kernel name.
++ Requires: -n.
++ Optional: -b
++.P
++ -A, --add_all_kernels
++ Add all kernels from existing VOLK module. Uses the base VOLK
++ module unless -b is used.
++ Optional: -b
++.P
++ -x, --remove_kernel
++ Remove kernel from module.
++ Required: -n.
++ Optional: -b
++.P
++Options for Listing Kernels:
++ -l, --list
++ Lists all kernels available in the base VOLK module.
++.P
++ -k, --kernels
++ Lists all kernels in this VOLK module.
++.P
++ -r, --remote-list
++ Lists all kernels in another VOLK module that is specified
++ using the -b option.
--- /dev/null
--- /dev/null
++.TH UHD_FFT "1" "March 2012" "volk_profile 3.5" "User Commands"
++.SH NAME
++volk_profile \- Quality Assurance application for libvolk functions
++.SH DESCRIPTION
++Writes profile results to a file.
--- /dev/null
--- /dev/null
++version=3
++opts="pgpsigurlmangle=s/$/.asc/" \
++ http://libvolk.org/releases/volk-(\d\S*)\.tar\.gz
projects / volk.git / commitdiff