--- /dev/null
+volk (3.0.0-2) unstable; urgency=medium
+
+ * upload to unstable
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 15 Jul 2023 21:58:53 -0400
+
+volk (3.0.0-1) experimental; urgency=medium
+
+ * New upstream release
+ - License switch to LGPLv3+
+ - Fix build for 32 bit arm with neon
+ - Add experimental support for MIPS and RISC-V
+ * Upload to experimental for package renames and soversion bump
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 14 Jan 2023 14:01:06 -0500
+
+volk (2.5.2-3) unstable; urgency=medium
+
+ * orc 1:0.4.33-1 dropped building static library,
+ so now volk will drop building its static library too. (Closes: #1026593)
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 20 Dec 2022 20:03:23 -0500
+
+volk (2.5.2-2) unstable; urgency=medium
+
+ * revert changes to kernels/volk/volk_8u_x2_encodeframepolar_8u.h
+ made by make-acc-happy patch since version 1.3-1 (Closes: #1021856)
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 15 Oct 2022 23:41:48 -0400
+
+volk (2.5.2-1) unstable; urgency=medium
+
+ * New upstream release.
+ * volk_8u_x4_conv_k7_r2_8u
+ - Add NEON implementation `neonspiral` via `sse2neon.h`
+ * Fixes
+ - Fix out-of-bounds reads
+ - Fix broken neon kernels
+ - Fix float to int conversion
+ * CMake
+ - Suppress superfluous warning
+ - Fix Python install path calculation and documentation
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 04 Sep 2022 12:00:56 -0400
+
+volk (2.5.1-2) unstable; urgency=medium
+
+ * VolkPython use posix prefix scheme (Closes: #1009394)
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 12 Apr 2022 18:39:33 -0400
+
+volk (2.5.1-1) unstable; urgency=medium
+
+ * New upstream release.
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 13 Feb 2022 00:18:58 -0500
+
+volk (2.5.0-2) unstable; urgency=medium
+
+ * upload to unstable
+ * with some upstream bugfixes
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 21 Oct 2021 23:30:05 -0400
+
+volk (2.5.0-1) experimental; urgency=medium
+
+ * New upstream release
+ * Use libcpu-features-dev on powerpc and x32 (Closes: #978602)
+ * Mention volk-config-info and volk_modtool in description (Closes: #989263)
+ * Upload to experimental for soversion bump
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 10 Jun 2021 18:29:47 -0400
+
+volk (2.4.1-2) unstable; urgency=medium
+
+ [ Shengjing Zhu ]
+ * Use system cpu_features package
+
+ [ A. Maitland Bottoms ]
+ * Adopt Use system cpu_features package patch (Closes: #978096)
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 27 Dec 2020 15:16:07 -0500
+
+volk (2.4.1-1) unstable; urgency=medium
+
+ * New upstream release
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 17 Dec 2020 23:53:21 -0500
+
+volk (2.4.0-4) unstable; urgency=medium
+
+ * skip cpu_features on "Unsupported OS" kFreeBSD
+ * bump Standards-Version - no other changes.
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 15 Dec 2020 19:53:16 -0500
+
+volk (2.4.0-3) unstable; urgency=medium
+
+ * Fix binary-indep build (Closes: #976300)
+ * Upload to unstable
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 03 Dec 2020 20:43:29 -0500
+
+volk (2.4.0-2) experimental; urgency=medium
+
+ * Make use of cpu_features a CMake option with sensible defaults per arch
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Mon, 30 Nov 2020 16:19:19 -0500
+
+volk (2.4.0-1) experimental; urgency=medium
+
+ * New upstream release
+ * cpu_features git submodule packaged as cpu-features source component.
+ * Upload to experimental for soversion bump
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 22 Nov 2020 12:35:43 -0500
+
+volk (2.3.0-3) unstable; urgency=medium
+
+ * update to v2.3.0-14-g91e5d07
+ emit an emms instruction after using the mmx extension
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 30 Jun 2020 19:48:20 -0400
+
+volk (2.3.0-2) unstable; urgency=medium
+
+ * Upload to unstable
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Mon, 11 May 2020 07:26:03 -0400
+
+volk (2.3.0-1) experimental; urgency=medium
+
+ * New upstream release, to experimental for soversion bump
+ * Kernels
+ - volk: accurate exp kernel
+ - exp: Rename SSE4.1 to SSE2 kernel
+ - Add 32f_s32f_add_32f kernel
+ - This kernel adds in vector + scalar functionality
+ - Fix the broken index max kernels
+ - Treat the mod_range puppet as such
+ - Add puppet for power spectral density kernel
+ - Updated log10 calcs to use faster log2 approach
+ - fix: Use unaligned load
+ - divide: Optimize complexmultiplyconjugate
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 09 May 2020 15:42:23 -0400
+
+volk (2.2.1-3) unstable; urgency=medium
+
+ * update to v2.2.1-34-gd4756c5
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 05 Apr 2020 10:37:46 -0400
+
+volk (2.2.1-2) unstable; urgency=medium
+
+ * update to v2.2.1-11-gfaf230e
+ * cmake: Remove the ORC from the VOLK public link interface
+ * Fix the broken index max kernels
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Fri, 27 Mar 2020 21:48:10 -0400
+
+volk (2.2.1-1) unstable; urgency=high
+
+ * New upstream bugfix release
+ reason for high urgency:
+ - Fix loop bound in AVX rotator (only one fixed in 2.2.0-3)
+ - Fix out-of-bounds read in AVX2 square dist kernel
+ - Fix length checks in AVX2 index max kernels
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Mon, 24 Feb 2020 18:08:05 -0500
+
+volk (2.2.0-3) unstable; urgency=high
+
+ * Update to v2.2.0-6-g5701f8f
+ reason for high urgency:
+ - Fix loop bound in AVX rotator
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 23 Feb 2020 23:49:18 -0500
+
+volk (2.2.0-2) unstable; urgency=medium
+
+ * Upload to unstable
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 18 Feb 2020 17:56:58 -0500
+
+volk (2.2.0-1) experimental; urgency=medium
+
+ * New upstream release
+ - Remove build dependency on python six
+ - Fixup VolkConfigVersion
+ - add volk_version.h
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 16 Feb 2020 18:25:20 -0500
+
+volk (2.1.0-2) unstable; urgency=medium
+
+ * Upload to unstable
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 05 Jan 2020 23:17:57 -0500
+
+volk (2.1.0-1) experimental; urgency=medium
+
+ * New upstream release
+ - The AVX FMA rotator bug is fixed
+ - VOLK offers `volk::vector<>` for C++ to follow RAII
+ - Use C++17 `std::filesystem`
+ - This enables VOLK to be built without Boost if available!
+ - lots of bugfixes
+ - more optimized kernels, especially more NEON versions
+ * Upload to experimental for new ABI library package libvolk2.1
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sun, 22 Dec 2019 10:27:36 -0500
+
+volk (2.0.0-3) unstable; urgency=medium
+
+ * update to v2.0.0-4-gf04a46f
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Thu, 14 Nov 2019 22:47:23 -0500
+
+volk (2.0.0-2) unstable; urgency=medium
+
+ * Upload to unstable
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Mon, 12 Aug 2019 22:49:11 -0400
+
+volk (2.0.0-1) experimental; urgency=medium
+
+ * New upstream release
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Wed, 07 Aug 2019 23:31:20 -0400
+
+volk (1.4-4) unstable; urgency=medium
+
+ * working volk_modtool with Python 3
+ * build and install libvolk.a
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Mon, 29 Oct 2018 01:32:05 -0400
+
+volk (1.4-3) unstable; urgency=medium
+
+ * update to v1.4-9-g297fefd
+ Added an AVX protokernel for volk_32fc_x2_32f_square_dist_scalar_mult_32f
+ fixed a buffer over-read and over-write in
+ volk_32fc_x2_s32f_square_dist_scalar_mult_32f_a_avx
+ Fix 32u_reverse_32u for ARM
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Sat, 12 May 2018 15:25:04 -0400
+
+volk (1.4-2) unstable; urgency=medium
+
+ * Upload to unstable, needed by gnuradio (>= 3.7.12.0)
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 03 Apr 2018 01:03:19 -0400
+
+volk (1.4-1) experimental; urgency=medium
+
+ * New upstream release
+ upstream changelog http://libvolk.org/release-v14.html
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 27 Mar 2018 22:57:42 -0400
+
+volk (1.3.1-1) unstable; urgency=medium
+
+ * New upstream bugfix release
+ * Refresh all debian patches for use with git am
+
+ -- A. Maitland Bottoms <bottoms@debian.org> Tue, 27 Mar 2018 21:54:29 -0400
+
+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
+Source: volk
+Section: libdevel
+Priority: optional
+Maintainer: A. Maitland Bottoms <bottoms@debian.org>
+Build-Depends: cmake,
+ debhelper-compat (= 13),
+ dh-python,
+ liborc-0.4-dev,
+ libcpu-features-dev [amd64 arm64 armel armhf i386 mips64el mipsel powerpc ppc64 ppc64el x32],
+ python3-dev,
+ python3-mako
+Build-Depends-Indep: doxygen, graphviz
+Standards-Version: 4.6.2
+Rules-Requires-Root: no
+Homepage: https://libvolk.org
+Vcs-Browser: https://salsa.debian.org/bottoms/pkg-volk
+Vcs-Git: https://salsa.debian.org/bottoms/pkg-volk.git
+
+Package: libvolk3.0
+Section: libs
+Architecture: any
+Pre-Depends: ${misc:Pre-Depends}
+Depends: ${misc:Depends}, ${shlibs:Depends}
+Multi-Arch: same
+Recommends: libvolk-bin
+Suggests: libvolk-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: libvolk-dev
+Architecture: any
+Pre-Depends: ${misc:Pre-Depends}
+Depends: libvolk3.0 (=${binary:Version}), ${misc:Depends}
+Breaks: libvolk2-dev, libvolk1.0-dev, libvolk1-dev
+Replaces: libvolk2-dev, libvolk1.0-dev, libvolk1-dev
+Suggests: libvolk-doc
+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: libvolk-bin
+Section: libs
+Architecture: any
+Pre-Depends: ${misc:Pre-Depends}
+Depends: libvolk3.0 (=${binary:Version}),
+ ${misc:Depends},
+ ${python3:Depends},
+ ${shlibs:Depends}
+Breaks: libvolk2-bin, libvolk1-bin, libvolk1.0-bin
+Replaces: libvolk2-bin, libvolk1-bin, libvolk1.0-bin
+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 to customize settings for
+ the system; volk_modtool to create new optimized modules; and
+ volk-config-info to show settings.
+
+Package: libvolk-doc
+Section: doc
+Architecture: all
+Multi-Arch: foreign
+Depends: ${misc:Depends}
+Recommends: www-browser
+Description: vector optimized library documentation
+ 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 Doxygen generated documentation in
+ /usr/share/doc/libvolk-dev/html/index.html
--- /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 Maintainers:
+ Johannes Demel <demel@uni-bremen.de>
+ Michael Dickens <michael.dickens@ettus.com>
+Copyright: 2014-2023 Free Software Foundation, Inc.
+License: LGPL-3+
+
+Files: *
+Copyright: 2006, 2009-2023, Free Software Foundation, Inc.
+License: LGPL-3+
+
+Files: apps/volk_profile.h
+Copyright: 2014-2020 Free Software Foundation, Inc.
+License: LGPL-3+
+
+Files: appveyor.yml
+Copyright: 2016 Paul Cercueil <paul.cercueil@analog.com>
+License: LGPL-3+
+
+Files: cmake/*
+Copyright: 2014-2020 Free Software Foundation, Inc.
+License: LGPL-3+
+
+Files: cmake/Modules/*
+Copyright: 2006, 2009-2020, Free Software Foundation, Inc.
+License: LGPL-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: LGPL-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: debian/*
+Copyright: 2015-2020 Free Software Foundation, Inc
+License: LGPL-3+
+Comment: assigned by A. Maitland Bottoms <bottoms@debian.org>
+
+Files: docs/*
+Copyright: 2014-2015 Free Software Foundation, Inc.
+License: LGPL-3+
+
+Files: gen/archs.xml
+ gen/machines.xml
+Copyright: 2014-2015 Free Software Foundation, Inc.
+License: LGPL-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: LGPL-3+
+
+Files: kernels/volk/asm/*
+Copyright: 2014-2015 Free Software Foundation, Inc.
+License: LGPL-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: LGPL-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: LGPL-3+
+
+License: LGPL-3+
+ This program is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser 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 Lesser General Public License
+ along with this program. If not, see <http://www.gnu.org/licenses/>.
+ .
+ On Debian systems, the complete text of the GNU Lesser General
+ Public License version 3 can be found in "/usr/share/common-licenses/LGPL-3".
--- /dev/null
+usr/bin/volk*
+usr/lib/python3*/site-packages/* usr/lib/python3/dist-packages/
--- /dev/null
+debian/volk-config-info.1
+debian/volk_modtool.1
+debian/volk_profile.1
--- /dev/null
+usr/include/*
+usr/lib/*/*volk*so
+usr/lib/*/cmake/volk
+usr/lib/*/pkgconfig/*volk*
--- /dev/null
+Document: libvolk-doc
+Title: Vector-Optimized Library of Kernels Reference Manual
+Author: GNU Radio Developers
+Abstract: VOLK is the Vector-Optimized Library of Kernels.
+ It is a library that contains kernels of hand-written SIMD code for
+ different mathematical operations. Since each SIMD architecture can
+ be very different and no compiler has yet come along to handle
+ vectorization properly or highly efficiently, VOLK approaches the
+ problem differently. For each architecture or platform that a
+ developer wishes to vectorize for, a new proto-kernel is added to
+ VOLK. At runtime, VOLK will select the correct proto-kernel. In this
+ way, the users of VOLK call a kernel for performing the operation
+ that is platform/architecture agnostic. This allows us to write
+ portable SIMD code.
+Section: Programming/C++
+
+Format: HTML
+Index: /usr/share/doc/libvolk-dev/html/index.html
+Files: /usr/share/doc/libvolk-dev/html/*.html
--- /dev/null
+obj-*/html
--- /dev/null
+usr/lib/*/libvolk.so.*
--- /dev/null
+usr/bin/list_cpu_features
+usr/lib/*/cmake/CpuFeatures/CpuFeaturesConfig.cmake
+usr/lib/*/cmake/CpuFeatures/CpuFeaturesConfigVersion.cmake
+usr/lib/*/cmake/CpuFeatures/CpuFeaturesTargets-relwithdebinfo.cmake
+usr/lib/*/cmake/CpuFeatures/CpuFeaturesTargets.cmake
+usr/lib/*/libcpu_features.a
--- /dev/null
+From fe2e4a73480bf2ac2e566052ea682817dddaf61f Mon Sep 17 00:00:00 2001
+From: Johannes Demel <demel@uni-bremen.de>
+Date: Sat, 14 Jan 2023 18:35:59 +0100
+Subject: [PATCH 1/5] ci: Remove license check
+
+This check fails because new contributors don't need to re-license. All
+new contributions must be made under LGPL. Thus, this check is
+superfluous. Also, it would add an administrative burden that we should
+avoid.
+
+Signed-off-by: Johannes Demel <demel@uni-bremen.de>
+---
+ CMakeLists.txt | 15 ---------------
+ 1 file changed, 15 deletions(-)
+
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 3816757..a6e4f87 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -383,21 +383,6 @@ if(ENABLE_MODTOOL)
+ add_subdirectory(python/volk_modtool)
+ endif()
+
+-########################################################################
+-# And the LGPL license check
+-########################################################################
+-# detect default for LGPL
+-if(VERSION_INFO_MAJOR_VERSION GREATER_EQUAL 3)
+- OPTION(ENABLE_LGPL "Enable testing for LGPL" ON)
+-else()
+- OPTION(ENABLE_LGPL "Enable testing for LGPL" OFF)
+-endif()
+-
+-if(ENABLE_TESTING AND ENABLE_LGPL)
+- message(STATUS "Checking for LGPL resubmission enabled")
+- add_test("check_lgpl" ${CMAKE_SOURCE_DIR}/scripts/licensing/count_contrib.sh ${CMAKE_SOURCE_DIR}/AUTHORS_RESUBMITTING_UNDER_LGPL_LICENSE.md)
+-endif()
+-
+ ########################################################################
+ # Print summary
+ ########################################################################
+--
+2.39.2
+
--- /dev/null
+From 43a5ecb7858395fc5f2eab48d06d617672081df1 Mon Sep 17 00:00:00 2001
+From: Johannes Demel <demel@uni-bremen.de>
+Date: Sun, 15 Jan 2023 11:57:12 +0100
+Subject: [PATCH 2/5] license: Fix SPDX identifiers
+
+Some files still carried GPL identifiers. These were identified and
+updated. Thus, all files should carry the LGPL identifier now.
+
+Files with incorrect identifiers were found with:
+`grep -irnI --exclude-dir={build,.git,cpu_features} GPL . | grep -v LGPL`
+
+Fixes #613
+
+Signed-off-by: Johannes Demel <demel@uni-bremen.de>
+---
+ CMakeLists.txt | 2 +-
+ apps/CMakeLists.txt | 2 +-
+ cmake/Modules/VolkConfig.cmake.in | 2 +-
+ cmake/Modules/VolkConfigVersion.cmake.in | 2 +-
+ cmake/cmake_uninstall.cmake.in | 2 +-
+ docs/CMakeLists.txt | 2 +-
+ include/volk/volk_version.h.in | 2 +-
+ lib/CMakeLists.txt | 2 +-
+ lib/constants.c.in | 2 +-
+ python/volk_modtool/CMakeLists.txt | 2 +-
+ 10 files changed, 10 insertions(+), 10 deletions(-)
+
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index a6e4f87..2ee2c72 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -3,7 +3,7 @@
+ #
+ # This file is part of VOLK
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ ########################################################################
+diff --git a/apps/CMakeLists.txt b/apps/CMakeLists.txt
+index 8ac9187..db6ab50 100644
+--- a/apps/CMakeLists.txt
++++ b/apps/CMakeLists.txt
+@@ -3,7 +3,7 @@
+ #
+ # This file is part of VOLK
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ ########################################################################
+diff --git a/cmake/Modules/VolkConfig.cmake.in b/cmake/Modules/VolkConfig.cmake.in
+index 3ffdf36..f18e07f 100644
+--- a/cmake/Modules/VolkConfig.cmake.in
++++ b/cmake/Modules/VolkConfig.cmake.in
+@@ -2,7 +2,7 @@
+ #
+ # This file is part of VOLK.
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ get_filename_component(VOLK_CMAKE_DIR "${CMAKE_CURRENT_LIST_FILE}" PATH)
+diff --git a/cmake/Modules/VolkConfigVersion.cmake.in b/cmake/Modules/VolkConfigVersion.cmake.in
+index b6e9b51..b97d25d 100644
+--- a/cmake/Modules/VolkConfigVersion.cmake.in
++++ b/cmake/Modules/VolkConfigVersion.cmake.in
+@@ -2,7 +2,7 @@
+ #
+ # This file is part of VOLK.
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ set(MAJOR_VERSION @VERSION_INFO_MAJOR_VERSION@)
+diff --git a/cmake/cmake_uninstall.cmake.in b/cmake/cmake_uninstall.cmake.in
+index d9d13ea..7ffbc90 100644
+--- a/cmake/cmake_uninstall.cmake.in
++++ b/cmake/cmake_uninstall.cmake.in
+@@ -2,7 +2,7 @@
+ #
+ # This file is part of VOLK.
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ # http://www.vtk.org/Wiki/CMake_FAQ#Can_I_do_.22make_uninstall.22_with_CMake.3F
+diff --git a/docs/CMakeLists.txt b/docs/CMakeLists.txt
+index 90eb1d7..7c4c5e0 100644
+--- a/docs/CMakeLists.txt
++++ b/docs/CMakeLists.txt
+@@ -2,7 +2,7 @@
+ #
+ # Copyright 2022 Johannes Demel.
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ find_package(Doxygen)
+diff --git a/include/volk/volk_version.h.in b/include/volk/volk_version.h.in
+index 46013da..e565048 100644
+--- a/include/volk/volk_version.h.in
++++ b/include/volk/volk_version.h.in
+@@ -4,7 +4,7 @@
+ *
+ * This file is part of VOLK
+ *
+- * SPDX-License-Identifier: GPL-3.0-or-later
++ * SPDX-License-Identifier: LGPL-3.0-or-later
+ */
+
+ #ifndef INCLUDED_VOLK_VERSION_H
+diff --git a/lib/CMakeLists.txt b/lib/CMakeLists.txt
+index 75055ee..3cfffe5 100644
+--- a/lib/CMakeLists.txt
++++ b/lib/CMakeLists.txt
+@@ -3,7 +3,7 @@
+ #
+ # This file is part of VOLK.
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ ########################################################################
+diff --git a/lib/constants.c.in b/lib/constants.c.in
+index fba8c39..049bc04 100644
+--- a/lib/constants.c.in
++++ b/lib/constants.c.in
+@@ -4,7 +4,7 @@
+ *
+ * This file is part of VOLK
+ *
+- * SPDX-License-Identifier: GPL-3.0-or-later
++ * SPDX-License-Identifier: LGPL-3.0-or-later
+ */
+
+ #if HAVE_CONFIG_H
+diff --git a/python/volk_modtool/CMakeLists.txt b/python/volk_modtool/CMakeLists.txt
+index aff7a62..dcc79a5 100644
+--- a/python/volk_modtool/CMakeLists.txt
++++ b/python/volk_modtool/CMakeLists.txt
+@@ -3,7 +3,7 @@
+ #
+ # This file is part of VOLK
+ #
+-# SPDX-License-Identifier: GPL-3.0-or-later
++# SPDX-License-Identifier: LGPL-3.0-or-later
+ #
+
+ ########################################################################
+--
+2.39.2
+
--- /dev/null
+From da2cbdcd70509093171f4807c51d92b7636add9f Mon Sep 17 00:00:00 2001
+From: Michael Roe <michael-roe@users.noreply.github.com>
+Date: Tue, 17 Jan 2023 20:15:38 +0000
+Subject: [PATCH 3/5] Use cpu_features on RISC-V platforms
+
+Signed-off-by: Michael Roe <michael-roe@users.noreply.github.com>
+---
+ .github/workflows/run-tests.yml | 2 +-
+ CMakeLists.txt | 2 +-
+ cpu_features | 2 +-
+ tmpl/volk_cpu.tmpl.c | 6 ++++++
+ 4 files changed, 9 insertions(+), 3 deletions(-)
+
+diff --git a/.github/workflows/run-tests.yml b/.github/workflows/run-tests.yml
+index 215ded6..1b7024d 100644
+--- a/.github/workflows/run-tests.yml
++++ b/.github/workflows/run-tests.yml
+@@ -114,7 +114,7 @@ jobs:
+ submodules: 'recursive'
+ - uses: uraimo/run-on-arch-action@v2.5.0
+ name: Build in non-x86 container
+- continue-on-error: ${{ contains(fromJson('["ppc64le", "s390x", "riscv64"]'), matrix.arch) || contains(fromJson('["clang-14"]'), matrix.compiler.name) }}
++ continue-on-error: ${{ contains(fromJson('["ppc64le", "s390x"]'), matrix.arch) || contains(fromJson('["clang-14"]'), matrix.compiler.name) }}
+ id: build
+ with:
+ arch: ${{ matrix.arch }}
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 2ee2c72..92d5097 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -121,7 +121,7 @@ endif(MSVC)
+
+ # cpu_features - sensible defaults, user settable option
+ if(CMAKE_SYSTEM_PROCESSOR MATCHES
+- "(^mips)|(^arm)|(^aarch64)|(x86_64)|(AMD64|amd64)|(^i.86$)|(^powerpc)|(^ppc)")
++ "(^mips)|(^arm)|(^aarch64)|(x86_64)|(AMD64|amd64)|(^i.86$)|(^powerpc)|(^ppc)|(^riscv)")
+ option(VOLK_CPU_FEATURES "Volk uses cpu_features" ON)
+ else()
+ option(VOLK_CPU_FEATURES "Volk uses cpu_features" OFF)
+diff --git a/cpu_features b/cpu_features
+index 188d0d3..4590768 160000
+--- a/cpu_features
++++ b/cpu_features
+@@ -1 +1 @@
+-Subproject commit 188d0d3c383689cdb6bb70dc6da2469faec84f61
++Subproject commit 4590768e530a470a8fb8917dafee69932831c854
+diff --git a/tmpl/volk_cpu.tmpl.c b/tmpl/volk_cpu.tmpl.c
+index a11a626..a4a06b0 100644
+--- a/tmpl/volk_cpu.tmpl.c
++++ b/tmpl/volk_cpu.tmpl.c
+@@ -25,6 +25,8 @@
+ #include "cpuinfo_mips.h"
+ #elif defined(CPU_FEATURES_ARCH_PPC)
+ #include "cpuinfo_ppc.h"
++#elif defined(CPU_FEATURES_ARCH_RISCV)
++#include "cpuinfo_riscv.h"
+ #endif
+
+ // This is required for MSVC
+@@ -46,6 +48,10 @@ static int i_can_has_${arch.name} (void) {
+ %elif "mips" in arch.name:
+ #if defined(CPU_FEATURES_ARCH_MIPS)
+ if (GetMipsInfo().features.${check} == 0){ return 0; }
++#endif
++ %elif "riscv" in arch.name:
++#if defined(CPU_FEATURES_ARCH_RISCV)
++ if (GetRiscvInfo().features.${check} == 0){ return 0; }
+ #endif
+ %else:
+ #if defined(CPU_FEATURES_ARCH_X86)
+--
+2.39.2
+
--- /dev/null
+From d630d45e6f273cdfe1da7082afb381b3832b4c8f Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Daniel=20Est=C3=A9vez?= <daniel@destevez.net>
+Date: Thu, 19 Jan 2023 18:01:25 +0100
+Subject: [PATCH 4/5] add volk_32f_s32f_x2_convert_8u kernel
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+This adds a kernel that performs conversion from float to uint8_t
+using a scale and a bias, according to out = in * scale + bias. The
+output is clamped to the interval [0, 255] and rounded to the nearest
+integer.
+
+The kernels are implemented mirroring those for
+volk_32f_s32f_convert_8i, but there is an additional avx2_fma kernel,
+since it makes sense to perform the scale and bias conversion using FMA
+if it is available.
+
+Signed-off-by: Daniel Estévez <daniel@destevez.net>
+---
+ docs/kernels.dox | 1 +
+ kernels/volk/volk_32f_s32f_convertpuppet_8u.h | 105 +++
+ kernels/volk/volk_32f_s32f_x2_convert_8u.h | 616 ++++++++++++++++++
+ lib/kernel_tests.h | 2 +
+ 4 files changed, 724 insertions(+)
+ create mode 100644 kernels/volk/volk_32f_s32f_convertpuppet_8u.h
+ create mode 100644 kernels/volk/volk_32f_s32f_x2_convert_8u.h
+
+diff --git a/docs/kernels.dox b/docs/kernels.dox
+index 35fca1e..e6930ad 100644
+--- a/docs/kernels.dox
++++ b/docs/kernels.dox
+@@ -84,6 +84,7 @@
+ \li \subpage volk_32f_s32f_power_32f
+ \li \subpage volk_32f_s32f_s32f_mod_range_32f
+ \li \subpage volk_32f_s32f_stddev_32f
++\li \subpage volk_32f_s32f_x2_convert_8u
+ \li \subpage volk_32f_sin_32f
+ \li \subpage volk_32f_sqrt_32f
+ \li \subpage volk_32f_stddev_and_mean_32f_x2
+diff --git a/kernels/volk/volk_32f_s32f_convertpuppet_8u.h b/kernels/volk/volk_32f_s32f_convertpuppet_8u.h
+new file mode 100644
+index 0000000..7f530c4
+--- /dev/null
++++ b/kernels/volk/volk_32f_s32f_convertpuppet_8u.h
+@@ -0,0 +1,105 @@
++/* -*- c++ -*- */
++/*
++ * Copyright 2023 Daniel Estevez <daniel@destevez.net>
++ *
++ * This file is part of VOLK
++ *
++ * SPDX-License-Identifier: LGPL-3.0-or-later
++ */
++
++#ifndef INCLUDED_VOLK_32F_S32F_MOD_CONVERTPUPPET_8U_H
++#define INCLUDED_VOLK_32F_S32F_MOD_CONVERTPUPPET_8U_H
++
++#include <inttypes.h>
++#include <volk/volk_32f_s32f_x2_convert_8u.h>
++
++#ifdef LV_HAVE_GENERIC
++static inline void volk_32f_s32f_convertpuppet_8u_generic(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_generic(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#if LV_HAVE_AVX2 && LV_HAVE_FMA
++static inline void volk_32f_s32f_convertpuppet_8u_u_avx2_fma(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_u_avx2_fma(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#if LV_HAVE_AVX2 && LV_HAVE_FMA
++static inline void volk_32f_s32f_convertpuppet_8u_a_avx2_fma(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_a_avx2_fma(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#ifdef LV_HAVE_AVX2
++static inline void volk_32f_s32f_convertpuppet_8u_u_avx2(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_u_avx2(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#ifdef LV_HAVE_AVX2
++static inline void volk_32f_s32f_convertpuppet_8u_a_avx2(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_a_avx2(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#ifdef LV_HAVE_SSE2
++static inline void volk_32f_s32f_convertpuppet_8u_u_sse2(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_u_sse2(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#ifdef LV_HAVE_SSE2
++static inline void volk_32f_s32f_convertpuppet_8u_a_sse2(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_a_sse2(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#ifdef LV_HAVE_SSE
++static inline void volk_32f_s32f_convertpuppet_8u_u_sse(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_u_sse(output, input, scale, 128.0, num_points);
++}
++#endif
++
++#ifdef LV_HAVE_SSE
++static inline void volk_32f_s32f_convertpuppet_8u_a_sse(uint8_t* output,
++ const float* input,
++ float scale,
++ unsigned int num_points)
++{
++ volk_32f_s32f_x2_convert_8u_a_sse(output, input, scale, 128.0, num_points);
++}
++#endif
++#endif
+diff --git a/kernels/volk/volk_32f_s32f_x2_convert_8u.h b/kernels/volk/volk_32f_s32f_x2_convert_8u.h
+new file mode 100644
+index 0000000..a52cdf2
+--- /dev/null
++++ b/kernels/volk/volk_32f_s32f_x2_convert_8u.h
+@@ -0,0 +1,616 @@
++/* -*- c++ -*- */
++/*
++ * Copyright 2023 Daniel Estevez <daniel@destevez.net>
++ * Copyright 2012, 2014 Free Software Foundation, Inc.
++ *
++ * This file is part of VOLK
++ *
++ * SPDX-License-Identifier: LGPL-3.0-or-later
++ */
++
++/*!
++ * \page volk_32f_s32f_x2_convert_8u
++ *
++ * \b Overview
++ *
++ * Converts a floating point number to an 8-bit unsigned int after applying a
++ * multiplicative scaling factor and an additive bias.
++ *
++ * <b>Dispatcher Prototype</b>
++ * \code
++ * void volk_32f_s32f_x2_convert_8u(uint8_t* outputVector, const float* inputVector,
++ const float scale, const float bias, unsigned int num_points)
++ * \endcode
++ *
++ * \b Inputs
++ * \li inputVector: the input vector of floats.
++ * \li scale: The value multiplied against each point in the input buffer.
++ * \li bias: The value added to each multiplication by the scale.
++ * \li num_points: The number of data points.
++ *
++ * \b Outputs
++ * \li outputVector: The output vector.
++ *
++ * \b Example
++ * Convert floats from [-1,1] to 8-bit unsigend integers with a scale of 128 and a bias of
++ 128
++ * int N = 10;
++ * unsigned int alignment = volk_get_alignment();
++ * float* increasing = (float*)volk_malloc(sizeof(float)*N, alignment);
++ * uint8_t* out = (uint8_t*)volk_malloc(sizeof(uint8_t)*N, alignment);
++ *
++ * for(unsigned int ii = 0; ii < N; ++ii){
++ * increasing[ii] = 2.f * ((float)ii / (float)N) - 1.f;
++ * }
++ *
++ * float scale = 128.0f;
++ * float bias = 128.0f;
++ *
++ * volk_32f_s32f_x2_convert_8u(out, increasing, scale, bias, N);
++ *
++ * for(unsigned int ii = 0; ii < N; ++ii){
++ * printf("out[%u] = %i\n", ii, out[ii]);
++ * }
++ *
++ * volk_free(increasing);
++ * volk_free(out);
++ * \endcode
++ */
++
++#ifndef INCLUDED_volk_32f_s32f_x2_convert_8u_u_H
++#define INCLUDED_volk_32f_s32f_x2_convert_8u_u_H
++
++#include <inttypes.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_single(uint8_t* out, const float in)
++{
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ if (in > max_val) {
++ *out = (uint8_t)(max_val);
++ } else if (in < min_val) {
++ *out = (uint8_t)(min_val);
++ } else {
++ *out = (uint8_t)(rintf(in));
++ }
++}
++
++
++#ifdef LV_HAVE_GENERIC
++
++static inline void volk_32f_s32f_x2_convert_8u_generic(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const float* inputVectorPtr = inputVector;
++
++ for (unsigned int number = 0; number < num_points; number++) {
++ const float r = *inputVectorPtr++ * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_GENERIC */
++
++
++#if LV_HAVE_AVX2 && LV_HAVE_FMA
++#include <immintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_u_avx2_fma(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int thirtysecondPoints = num_points / 32;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m256 vmin_val = _mm256_set1_ps(min_val);
++ const __m256 vmax_val = _mm256_set1_ps(max_val);
++
++ const __m256 vScale = _mm256_set1_ps(scale);
++ const __m256 vBias = _mm256_set1_ps(bias);
++
++ for (unsigned int number = 0; number < thirtysecondPoints; number++) {
++ __m256 inputVal1 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal2 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal3 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal4 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++
++ inputVal1 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal1, vScale, vBias), vmax_val), vmin_val);
++ inputVal2 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal2, vScale, vBias), vmax_val), vmin_val);
++ inputVal3 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal3, vScale, vBias), vmax_val), vmin_val);
++ inputVal4 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal4, vScale, vBias), vmax_val), vmin_val);
++
++ __m256i intInputVal1 = _mm256_cvtps_epi32(inputVal1);
++ __m256i intInputVal2 = _mm256_cvtps_epi32(inputVal2);
++ __m256i intInputVal3 = _mm256_cvtps_epi32(inputVal3);
++ __m256i intInputVal4 = _mm256_cvtps_epi32(inputVal4);
++
++ intInputVal1 = _mm256_packs_epi32(intInputVal1, intInputVal2);
++ intInputVal1 = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++ intInputVal3 = _mm256_packs_epi32(intInputVal3, intInputVal4);
++ intInputVal3 = _mm256_permute4x64_epi64(intInputVal3, 0b11011000);
++
++ intInputVal1 = _mm256_packus_epi16(intInputVal1, intInputVal3);
++ const __m256i intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++
++ _mm256_storeu_si256((__m256i*)outputVectorPtr, intInputVal);
++ outputVectorPtr += 32;
++ }
++
++ for (unsigned int number = thirtysecondPoints * 32; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_AVX2 && LV_HAVE_FMA */
++
++
++#ifdef LV_HAVE_AVX2
++#include <immintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_u_avx2(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int thirtysecondPoints = num_points / 32;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m256 vmin_val = _mm256_set1_ps(min_val);
++ const __m256 vmax_val = _mm256_set1_ps(max_val);
++
++ const __m256 vScale = _mm256_set1_ps(scale);
++ const __m256 vBias = _mm256_set1_ps(bias);
++
++ for (unsigned int number = 0; number < thirtysecondPoints; number++) {
++ __m256 inputVal1 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal2 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal3 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal4 = _mm256_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++
++ inputVal1 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal1, vScale), vBias),
++ vmax_val),
++ vmin_val);
++ inputVal2 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal2, vScale), vBias),
++ vmax_val),
++ vmin_val);
++ inputVal3 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal3, vScale), vBias),
++ vmax_val),
++ vmin_val);
++ inputVal4 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal4, vScale), vBias),
++ vmax_val),
++ vmin_val);
++
++ __m256i intInputVal1 = _mm256_cvtps_epi32(inputVal1);
++ __m256i intInputVal2 = _mm256_cvtps_epi32(inputVal2);
++ __m256i intInputVal3 = _mm256_cvtps_epi32(inputVal3);
++ __m256i intInputVal4 = _mm256_cvtps_epi32(inputVal4);
++
++ intInputVal1 = _mm256_packs_epi32(intInputVal1, intInputVal2);
++ intInputVal1 = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++ intInputVal3 = _mm256_packs_epi32(intInputVal3, intInputVal4);
++ intInputVal3 = _mm256_permute4x64_epi64(intInputVal3, 0b11011000);
++
++ intInputVal1 = _mm256_packus_epi16(intInputVal1, intInputVal3);
++ const __m256i intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++
++ _mm256_storeu_si256((__m256i*)outputVectorPtr, intInputVal);
++ outputVectorPtr += 32;
++ }
++
++ for (unsigned int number = thirtysecondPoints * 32; number < num_points; number++) {
++ float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_AVX2 */
++
++
++#ifdef LV_HAVE_SSE2
++#include <emmintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_u_sse2(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int sixteenthPoints = num_points / 16;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
++
++ const __m128 vScale = _mm_set_ps1(scale);
++ const __m128 vBias = _mm_set_ps1(bias);
++
++ for (unsigned int number = 0; number < sixteenthPoints; number++) {
++ __m128 inputVal1 = _mm_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++ __m128 inputVal2 = _mm_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++ __m128 inputVal3 = _mm_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++ __m128 inputVal4 = _mm_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++
++ inputVal1 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal1, vScale), vBias), vmax_val),
++ vmin_val);
++ inputVal2 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal2, vScale), vBias), vmax_val),
++ vmin_val);
++ inputVal3 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal3, vScale), vBias), vmax_val),
++ vmin_val);
++ inputVal4 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal4, vScale), vBias), vmax_val),
++ vmin_val);
++
++ __m128i intInputVal1 = _mm_cvtps_epi32(inputVal1);
++ __m128i intInputVal2 = _mm_cvtps_epi32(inputVal2);
++ __m128i intInputVal3 = _mm_cvtps_epi32(inputVal3);
++ __m128i intInputVal4 = _mm_cvtps_epi32(inputVal4);
++
++ intInputVal1 = _mm_packs_epi32(intInputVal1, intInputVal2);
++ intInputVal3 = _mm_packs_epi32(intInputVal3, intInputVal4);
++
++ intInputVal1 = _mm_packus_epi16(intInputVal1, intInputVal3);
++
++ _mm_storeu_si128((__m128i*)outputVectorPtr, intInputVal1);
++ outputVectorPtr += 16;
++ }
++
++ for (unsigned int number = sixteenthPoints * 16; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_SSE2 */
++
++
++#ifdef LV_HAVE_SSE
++#include <xmmintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_u_sse(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int quarterPoints = num_points / 4;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
++
++ const __m128 vScale = _mm_set_ps1(scale);
++ const __m128 vBias = _mm_set_ps1(bias);
++
++ __VOLK_ATTR_ALIGNED(16) float outputFloatBuffer[4];
++
++ for (unsigned int number = 0; number < quarterPoints; number++) {
++ __m128 ret = _mm_loadu_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++
++ ret = _mm_max_ps(_mm_min_ps(_mm_add_ps(_mm_mul_ps(ret, vScale), vBias), vmax_val),
++ vmin_val);
++
++ _mm_store_ps(outputFloatBuffer, ret);
++ for (size_t inner_loop = 0; inner_loop < 4; inner_loop++) {
++ *outputVectorPtr++ = (uint8_t)(rintf(outputFloatBuffer[inner_loop]));
++ }
++ }
++
++ for (unsigned int number = quarterPoints * 4; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_SSE */
++
++
++#endif /* INCLUDED_volk_32f_s32f_x2_convert_8u_u_H */
++#ifndef INCLUDED_volk_32f_s32f_x2_convert_8u_a_H
++#define INCLUDED_volk_32f_s32f_x2_convert_8u_a_H
++
++#include <inttypes.h>
++#include <volk/volk_common.h>
++
++#if LV_HAVE_AVX2 && LV_HAVE_FMA
++#include <immintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_a_avx2_fma(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int thirtysecondPoints = num_points / 32;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m256 vmin_val = _mm256_set1_ps(min_val);
++ const __m256 vmax_val = _mm256_set1_ps(max_val);
++
++ const __m256 vScale = _mm256_set1_ps(scale);
++ const __m256 vBias = _mm256_set1_ps(bias);
++
++ for (unsigned int number = 0; number < thirtysecondPoints; number++) {
++ __m256 inputVal1 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal2 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal3 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal4 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++
++ inputVal1 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal1, vScale, vBias), vmax_val), vmin_val);
++ inputVal2 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal2, vScale, vBias), vmax_val), vmin_val);
++ inputVal3 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal3, vScale, vBias), vmax_val), vmin_val);
++ inputVal4 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_fmadd_ps(inputVal4, vScale, vBias), vmax_val), vmin_val);
++
++ __m256i intInputVal1 = _mm256_cvtps_epi32(inputVal1);
++ __m256i intInputVal2 = _mm256_cvtps_epi32(inputVal2);
++ __m256i intInputVal3 = _mm256_cvtps_epi32(inputVal3);
++ __m256i intInputVal4 = _mm256_cvtps_epi32(inputVal4);
++
++ intInputVal1 = _mm256_packs_epi32(intInputVal1, intInputVal2);
++ intInputVal1 = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++ intInputVal3 = _mm256_packs_epi32(intInputVal3, intInputVal4);
++ intInputVal3 = _mm256_permute4x64_epi64(intInputVal3, 0b11011000);
++
++ intInputVal1 = _mm256_packus_epi16(intInputVal1, intInputVal3);
++ const __m256i intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++
++ _mm256_store_si256((__m256i*)outputVectorPtr, intInputVal);
++ outputVectorPtr += 32;
++ }
++
++ for (unsigned int number = thirtysecondPoints * 32; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_AVX2 && LV_HAVE_FMA */
++
++
++#ifdef LV_HAVE_AVX2
++#include <immintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_a_avx2(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int thirtysecondPoints = num_points / 32;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m256 vmin_val = _mm256_set1_ps(min_val);
++ const __m256 vmax_val = _mm256_set1_ps(max_val);
++
++ const __m256 vScale = _mm256_set1_ps(scale);
++ const __m256 vBias = _mm256_set1_ps(bias);
++
++ for (unsigned int number = 0; number < thirtysecondPoints; number++) {
++ __m256 inputVal1 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal2 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal3 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++ __m256 inputVal4 = _mm256_load_ps(inputVectorPtr);
++ inputVectorPtr += 8;
++
++ inputVal1 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal1, vScale), vBias),
++ vmax_val),
++ vmin_val);
++ inputVal2 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal2, vScale), vBias),
++ vmax_val),
++ vmin_val);
++ inputVal3 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal3, vScale), vBias),
++ vmax_val),
++ vmin_val);
++ inputVal4 = _mm256_max_ps(
++ _mm256_min_ps(_mm256_add_ps(_mm256_mul_ps(inputVal4, vScale), vBias),
++ vmax_val),
++ vmin_val);
++
++ __m256i intInputVal1 = _mm256_cvtps_epi32(inputVal1);
++ __m256i intInputVal2 = _mm256_cvtps_epi32(inputVal2);
++ __m256i intInputVal3 = _mm256_cvtps_epi32(inputVal3);
++ __m256i intInputVal4 = _mm256_cvtps_epi32(inputVal4);
++
++ intInputVal1 = _mm256_packs_epi32(intInputVal1, intInputVal2);
++ intInputVal1 = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++ intInputVal3 = _mm256_packs_epi32(intInputVal3, intInputVal4);
++ intInputVal3 = _mm256_permute4x64_epi64(intInputVal3, 0b11011000);
++
++ intInputVal1 = _mm256_packus_epi16(intInputVal1, intInputVal3);
++ const __m256i intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++
++ _mm256_store_si256((__m256i*)outputVectorPtr, intInputVal);
++ outputVectorPtr += 32;
++ }
++
++ for (unsigned int number = thirtysecondPoints * 32; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_AVX2 */
++
++
++#ifdef LV_HAVE_SSE2
++#include <emmintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_a_sse2(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int sixteenthPoints = num_points / 16;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
++
++ const __m128 vScale = _mm_set_ps1(scale);
++ const __m128 vBias = _mm_set_ps1(bias);
++
++ for (unsigned int number = 0; number < sixteenthPoints; number++) {
++ __m128 inputVal1 = _mm_load_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++ __m128 inputVal2 = _mm_load_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++ __m128 inputVal3 = _mm_load_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++ __m128 inputVal4 = _mm_load_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++
++ inputVal1 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal1, vScale), vBias), vmax_val),
++ vmin_val);
++ inputVal2 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal2, vScale), vBias), vmax_val),
++ vmin_val);
++ inputVal3 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal3, vScale), vBias), vmax_val),
++ vmin_val);
++ inputVal4 = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(inputVal4, vScale), vBias), vmax_val),
++ vmin_val);
++
++ __m128i intInputVal1 = _mm_cvtps_epi32(inputVal1);
++ __m128i intInputVal2 = _mm_cvtps_epi32(inputVal2);
++ __m128i intInputVal3 = _mm_cvtps_epi32(inputVal3);
++ __m128i intInputVal4 = _mm_cvtps_epi32(inputVal4);
++
++ intInputVal1 = _mm_packs_epi32(intInputVal1, intInputVal2);
++ intInputVal3 = _mm_packs_epi32(intInputVal3, intInputVal4);
++
++ intInputVal1 = _mm_packus_epi16(intInputVal1, intInputVal3);
++
++ _mm_store_si128((__m128i*)outputVectorPtr, intInputVal1);
++ outputVectorPtr += 16;
++ }
++
++ for (unsigned int number = sixteenthPoints * 16; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++#endif /* LV_HAVE_SSE2 */
++
++
++#ifdef LV_HAVE_SSE
++#include <xmmintrin.h>
++
++static inline void volk_32f_s32f_x2_convert_8u_a_sse(uint8_t* outputVector,
++ const float* inputVector,
++ const float scale,
++ const float bias,
++ unsigned int num_points)
++{
++ const unsigned int quarterPoints = num_points / 4;
++
++ const float* inputVectorPtr = (const float*)inputVector;
++ uint8_t* outputVectorPtr = outputVector;
++
++ const float min_val = 0.0f;
++ const float max_val = UINT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
++
++ const __m128 vScalar = _mm_set_ps1(scale);
++ const __m128 vBias = _mm_set_ps1(bias);
++
++ __VOLK_ATTR_ALIGNED(16) float outputFloatBuffer[4];
++
++ for (unsigned int number = 0; number < quarterPoints; number++) {
++ __m128 ret = _mm_load_ps(inputVectorPtr);
++ inputVectorPtr += 4;
++
++ ret = _mm_max_ps(
++ _mm_min_ps(_mm_add_ps(_mm_mul_ps(ret, vScalar), vBias), vmax_val), vmin_val);
++
++ _mm_store_ps(outputFloatBuffer, ret);
++ for (size_t inner_loop = 0; inner_loop < 4; inner_loop++) {
++ *outputVectorPtr++ = (uint8_t)(rintf(outputFloatBuffer[inner_loop]));
++ }
++ }
++
++ for (unsigned int number = quarterPoints * 4; number < num_points; number++) {
++ const float r = inputVector[number] * scale + bias;
++ volk_32f_s32f_x2_convert_8u_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_SSE */
++
++
++#endif /* INCLUDED_volk_32f_s32f_x2_convert_8u_a_H */
+diff --git a/lib/kernel_tests.h b/lib/kernel_tests.h
+index 72330e4..d7e4ad1 100644
+--- a/lib/kernel_tests.h
++++ b/lib/kernel_tests.h
+@@ -173,6 +173,8 @@ std::vector<volk_test_case_t> init_test_list(volk_test_params_t test_params)
+ QA(VOLK_INIT_PUPP(volk_32fc_s32f_power_spectral_densitypuppet_32f,
+ volk_32fc_s32f_x2_power_spectral_density_32f,
+ test_params))
++ QA(VOLK_INIT_PUPP(
++ volk_32f_s32f_convertpuppet_8u, volk_32f_s32f_x2_convert_8u, test_params))
+ // no one uses these, so don't test them
+ // VOLK_PROFILE(volk_16i_x5_add_quad_16i_x4, 1e-4, 2046, 10000, &results,
+ // benchmark_mode, kernel_regex); VOLK_PROFILE(volk_16i_branch_4_state_8, 1e-4, 2046,
+--
+2.39.2
+
--- /dev/null
+From 59c9539bfd1dff6d3e7b6c9f6ddd9588d4fcb661 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Daniel=20Est=C3=A9vez?= <daniel@destevez.net>
+Date: Sun, 22 Jan 2023 16:10:27 +0100
+Subject: [PATCH 5/5] volk_32f_s32f_convert_8i: code style
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Apply code style suggestions from #617.
+
+Signed-off-by: Daniel Estévez <daniel@destevez.net>
+---
+ kernels/volk/volk_32f_s32f_convert_8i.h | 283 +++++++++---------------
+ 1 file changed, 110 insertions(+), 173 deletions(-)
+
+diff --git a/kernels/volk/volk_32f_s32f_convert_8i.h b/kernels/volk/volk_32f_s32f_convert_8i.h
+index 4d7c5ca..d47f95a 100644
+--- a/kernels/volk/volk_32f_s32f_convert_8i.h
++++ b/kernels/volk/volk_32f_s32f_convert_8i.h
+@@ -30,12 +30,12 @@
+ * \li outputVector: The output vector.
+ *
+ * \b Example
+- * Convert floats from [-1,1] to 16-bit integers with a scale of 5 to maintain smallest
++ * Convert floats from [-1,1] to 8-bit integers with a scale of 5 to maintain smallest
+ delta
+ * int N = 10;
+ * unsigned int alignment = volk_get_alignment();
+ * float* increasing = (float*)volk_malloc(sizeof(float)*N, alignment);
+- * int16_t* out = (int16_t*)volk_malloc(sizeof(int16_t)*N, alignment);
++ * int8_t* out = (int8_t*)volk_malloc(sizeof(int8_t)*N, alignment);
+ *
+ * for(unsigned int ii = 0; ii < N; ++ii){
+ * increasing[ii] = 2.f * ((float)ii / (float)N) - 1.f;
+@@ -46,7 +46,7 @@
+
+ * float scale = 5.1f;
+ *
+- * volk_32f_s32f_convert_32i(out, increasing, scale, N);
++ * volk_32f_s32f_convert_8i(out, increasing, scale, N);
+ *
+ * for(unsigned int ii = 0; ii < N; ++ii){
+ * printf("out[%u] = %i\n", ii, out[ii]);
+@@ -61,12 +61,11 @@
+ #define INCLUDED_volk_32f_s32f_convert_8i_u_H
+
+ #include <inttypes.h>
+-#include <stdio.h>
+
+ static inline void volk_32f_s32f_convert_8i_single(int8_t* out, const float in)
+ {
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
+ if (in > max_val) {
+ *out = (int8_t)(max_val);
+ } else if (in < min_val) {
+@@ -76,6 +75,24 @@ static inline void volk_32f_s32f_convert_8i_single(int8_t* out, const float in)
+ }
+ }
+
++#ifdef LV_HAVE_GENERIC
++
++static inline void volk_32f_s32f_convert_8i_generic(int8_t* outputVector,
++ const float* inputVector,
++ const float scalar,
++ unsigned int num_points)
++{
++ const float* inputVectorPtr = inputVector;
++
++ for (unsigned int number = 0; number < num_points; number++) {
++ const float r = *inputVectorPtr++ * scalar;
++ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
++ }
++}
++
++#endif /* LV_HAVE_GENERIC */
++
++
+ #ifdef LV_HAVE_AVX2
+ #include <immintrin.h>
+
+@@ -84,32 +101,26 @@ static inline void volk_32f_s32f_convert_8i_u_avx2(int8_t* outputVector,
+ const float scalar,
+ unsigned int num_points)
+ {
+- unsigned int number = 0;
+-
+ const unsigned int thirtysecondPoints = num_points / 32;
+
+ const float* inputVectorPtr = (const float*)inputVector;
+ int8_t* outputVectorPtr = outputVector;
+
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
+- float r;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
++ const __m256 vmin_val = _mm256_set1_ps(min_val);
++ const __m256 vmax_val = _mm256_set1_ps(max_val);
+
+- __m256 vScalar = _mm256_set1_ps(scalar);
+- __m256 inputVal1, inputVal2, inputVal3, inputVal4;
+- __m256i intInputVal1, intInputVal2, intInputVal3, intInputVal4;
+- __m256 vmin_val = _mm256_set1_ps(min_val);
+- __m256 vmax_val = _mm256_set1_ps(max_val);
+- __m256i intInputVal;
++ const __m256 vScalar = _mm256_set1_ps(scalar);
+
+- for (; number < thirtysecondPoints; number++) {
+- inputVal1 = _mm256_loadu_ps(inputVectorPtr);
++ for (unsigned int number = 0; number < thirtysecondPoints; number++) {
++ __m256 inputVal1 = _mm256_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+- inputVal2 = _mm256_loadu_ps(inputVectorPtr);
++ __m256 inputVal2 = _mm256_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+- inputVal3 = _mm256_loadu_ps(inputVectorPtr);
++ __m256 inputVal3 = _mm256_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+- inputVal4 = _mm256_loadu_ps(inputVectorPtr);
++ __m256 inputVal4 = _mm256_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+
+ inputVal1 = _mm256_max_ps(
+@@ -121,10 +132,10 @@ static inline void volk_32f_s32f_convert_8i_u_avx2(int8_t* outputVector,
+ inputVal4 = _mm256_max_ps(
+ _mm256_min_ps(_mm256_mul_ps(inputVal4, vScalar), vmax_val), vmin_val);
+
+- intInputVal1 = _mm256_cvtps_epi32(inputVal1);
+- intInputVal2 = _mm256_cvtps_epi32(inputVal2);
+- intInputVal3 = _mm256_cvtps_epi32(inputVal3);
+- intInputVal4 = _mm256_cvtps_epi32(inputVal4);
++ __m256i intInputVal1 = _mm256_cvtps_epi32(inputVal1);
++ __m256i intInputVal2 = _mm256_cvtps_epi32(inputVal2);
++ __m256i intInputVal3 = _mm256_cvtps_epi32(inputVal3);
++ __m256i intInputVal4 = _mm256_cvtps_epi32(inputVal4);
+
+ intInputVal1 = _mm256_packs_epi32(intInputVal1, intInputVal2);
+ intInputVal1 = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
+@@ -132,15 +143,14 @@ static inline void volk_32f_s32f_convert_8i_u_avx2(int8_t* outputVector,
+ intInputVal3 = _mm256_permute4x64_epi64(intInputVal3, 0b11011000);
+
+ intInputVal1 = _mm256_packs_epi16(intInputVal1, intInputVal3);
+- intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++ const __m256i intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
+
+ _mm256_storeu_si256((__m256i*)outputVectorPtr, intInputVal);
+ outputVectorPtr += 32;
+ }
+
+- number = thirtysecondPoints * 32;
+- for (; number < num_points; number++) {
+- r = inputVector[number] * scalar;
++ for (unsigned int number = thirtysecondPoints * 32; number < num_points; number++) {
++ float r = inputVector[number] * scalar;
+ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+ }
+ }
+@@ -156,31 +166,26 @@ static inline void volk_32f_s32f_convert_8i_u_sse2(int8_t* outputVector,
+ const float scalar,
+ unsigned int num_points)
+ {
+- unsigned int number = 0;
+-
+ const unsigned int sixteenthPoints = num_points / 16;
+
+ const float* inputVectorPtr = (const float*)inputVector;
+ int8_t* outputVectorPtr = outputVector;
+
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
+- float r;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
+
+- __m128 vScalar = _mm_set_ps1(scalar);
+- __m128 inputVal1, inputVal2, inputVal3, inputVal4;
+- __m128i intInputVal1, intInputVal2, intInputVal3, intInputVal4;
+- __m128 vmin_val = _mm_set_ps1(min_val);
+- __m128 vmax_val = _mm_set_ps1(max_val);
++ const __m128 vScalar = _mm_set_ps1(scalar);
+
+- for (; number < sixteenthPoints; number++) {
+- inputVal1 = _mm_loadu_ps(inputVectorPtr);
++ for (unsigned int number = 0; number < sixteenthPoints; number++) {
++ __m128 inputVal1 = _mm_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+- inputVal2 = _mm_loadu_ps(inputVectorPtr);
++ __m128 inputVal2 = _mm_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+- inputVal3 = _mm_loadu_ps(inputVectorPtr);
++ __m128 inputVal3 = _mm_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+- inputVal4 = _mm_loadu_ps(inputVectorPtr);
++ __m128 inputVal4 = _mm_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+
+ inputVal1 =
+@@ -192,10 +197,10 @@ static inline void volk_32f_s32f_convert_8i_u_sse2(int8_t* outputVector,
+ inputVal4 =
+ _mm_max_ps(_mm_min_ps(_mm_mul_ps(inputVal4, vScalar), vmax_val), vmin_val);
+
+- intInputVal1 = _mm_cvtps_epi32(inputVal1);
+- intInputVal2 = _mm_cvtps_epi32(inputVal2);
+- intInputVal3 = _mm_cvtps_epi32(inputVal3);
+- intInputVal4 = _mm_cvtps_epi32(inputVal4);
++ __m128i intInputVal1 = _mm_cvtps_epi32(inputVal1);
++ __m128i intInputVal2 = _mm_cvtps_epi32(inputVal2);
++ __m128i intInputVal3 = _mm_cvtps_epi32(inputVal3);
++ __m128i intInputVal4 = _mm_cvtps_epi32(inputVal4);
+
+ intInputVal1 = _mm_packs_epi32(intInputVal1, intInputVal2);
+ intInputVal3 = _mm_packs_epi32(intInputVal3, intInputVal4);
+@@ -206,9 +211,8 @@ static inline void volk_32f_s32f_convert_8i_u_sse2(int8_t* outputVector,
+ outputVectorPtr += 16;
+ }
+
+- number = sixteenthPoints * 16;
+- for (; number < num_points; number++) {
+- r = inputVector[number] * scalar;
++ for (unsigned int number = sixteenthPoints * 16; number < num_points; number++) {
++ const float r = inputVector[number] * scalar;
+ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+ }
+ }
+@@ -224,40 +228,34 @@ static inline void volk_32f_s32f_convert_8i_u_sse(int8_t* outputVector,
+ const float scalar,
+ unsigned int num_points)
+ {
+- unsigned int number = 0;
+- size_t inner_loop;
+-
+ const unsigned int quarterPoints = num_points / 4;
+
+ const float* inputVectorPtr = (const float*)inputVector;
+ int8_t* outputVectorPtr = outputVector;
+
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
+- float r;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
+
+- __m128 vScalar = _mm_set_ps1(scalar);
+- __m128 ret;
+- __m128 vmin_val = _mm_set_ps1(min_val);
+- __m128 vmax_val = _mm_set_ps1(max_val);
++ const __m128 vScalar = _mm_set_ps1(scalar);
+
+ __VOLK_ATTR_ALIGNED(16) float outputFloatBuffer[4];
+
+- for (; number < quarterPoints; number++) {
+- ret = _mm_loadu_ps(inputVectorPtr);
++ for (unsigned int number = 0; number < quarterPoints; number++) {
++ __m128 ret = _mm_loadu_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+
+ ret = _mm_max_ps(_mm_min_ps(_mm_mul_ps(ret, vScalar), vmax_val), vmin_val);
+
+ _mm_store_ps(outputFloatBuffer, ret);
+- for (inner_loop = 0; inner_loop < 4; inner_loop++) {
++ for (size_t inner_loop = 0; inner_loop < 4; inner_loop++) {
+ *outputVectorPtr++ = (int8_t)(rintf(outputFloatBuffer[inner_loop]));
+ }
+ }
+
+- number = quarterPoints * 4;
+- for (; number < num_points; number++) {
+- r = inputVector[number] * scalar;
++ for (unsigned int number = quarterPoints * 4; number < num_points; number++) {
++ const float r = inputVector[number] * scalar;
+ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+ }
+ }
+@@ -265,33 +263,11 @@ static inline void volk_32f_s32f_convert_8i_u_sse(int8_t* outputVector,
+ #endif /* LV_HAVE_SSE */
+
+
+-#ifdef LV_HAVE_GENERIC
+-
+-static inline void volk_32f_s32f_convert_8i_generic(int8_t* outputVector,
+- const float* inputVector,
+- const float scalar,
+- unsigned int num_points)
+-{
+- const float* inputVectorPtr = inputVector;
+- unsigned int number = 0;
+- float r;
+-
+- for (number = 0; number < num_points; number++) {
+- r = *inputVectorPtr++ * scalar;
+- volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+- }
+-}
+-
+-#endif /* LV_HAVE_GENERIC */
+-
+-
+ #endif /* INCLUDED_volk_32f_s32f_convert_8i_u_H */
+ #ifndef INCLUDED_volk_32f_s32f_convert_8i_a_H
+ #define INCLUDED_volk_32f_s32f_convert_8i_a_H
+
+ #include <inttypes.h>
+-#include <stdio.h>
+-#include <volk/volk_common.h>
+
+ #ifdef LV_HAVE_AVX2
+ #include <immintrin.h>
+@@ -301,32 +277,26 @@ static inline void volk_32f_s32f_convert_8i_a_avx2(int8_t* outputVector,
+ const float scalar,
+ unsigned int num_points)
+ {
+- unsigned int number = 0;
+-
+ const unsigned int thirtysecondPoints = num_points / 32;
+
+ const float* inputVectorPtr = (const float*)inputVector;
+ int8_t* outputVectorPtr = outputVector;
+
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
+- float r;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
++ const __m256 vmin_val = _mm256_set1_ps(min_val);
++ const __m256 vmax_val = _mm256_set1_ps(max_val);
+
+- __m256 vScalar = _mm256_set1_ps(scalar);
+- __m256 inputVal1, inputVal2, inputVal3, inputVal4;
+- __m256i intInputVal1, intInputVal2, intInputVal3, intInputVal4;
+- __m256 vmin_val = _mm256_set1_ps(min_val);
+- __m256 vmax_val = _mm256_set1_ps(max_val);
+- __m256i intInputVal;
++ const __m256 vScalar = _mm256_set1_ps(scalar);
+
+- for (; number < thirtysecondPoints; number++) {
+- inputVal1 = _mm256_load_ps(inputVectorPtr);
++ for (unsigned int number = 0; number < thirtysecondPoints; number++) {
++ __m256 inputVal1 = _mm256_load_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+- inputVal2 = _mm256_load_ps(inputVectorPtr);
++ __m256 inputVal2 = _mm256_load_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+- inputVal3 = _mm256_load_ps(inputVectorPtr);
++ __m256 inputVal3 = _mm256_load_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+- inputVal4 = _mm256_load_ps(inputVectorPtr);
++ __m256 inputVal4 = _mm256_load_ps(inputVectorPtr);
+ inputVectorPtr += 8;
+
+ inputVal1 = _mm256_max_ps(
+@@ -338,10 +308,10 @@ static inline void volk_32f_s32f_convert_8i_a_avx2(int8_t* outputVector,
+ inputVal4 = _mm256_max_ps(
+ _mm256_min_ps(_mm256_mul_ps(inputVal4, vScalar), vmax_val), vmin_val);
+
+- intInputVal1 = _mm256_cvtps_epi32(inputVal1);
+- intInputVal2 = _mm256_cvtps_epi32(inputVal2);
+- intInputVal3 = _mm256_cvtps_epi32(inputVal3);
+- intInputVal4 = _mm256_cvtps_epi32(inputVal4);
++ __m256i intInputVal1 = _mm256_cvtps_epi32(inputVal1);
++ __m256i intInputVal2 = _mm256_cvtps_epi32(inputVal2);
++ __m256i intInputVal3 = _mm256_cvtps_epi32(inputVal3);
++ __m256i intInputVal4 = _mm256_cvtps_epi32(inputVal4);
+
+ intInputVal1 = _mm256_packs_epi32(intInputVal1, intInputVal2);
+ intInputVal1 = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
+@@ -349,15 +319,14 @@ static inline void volk_32f_s32f_convert_8i_a_avx2(int8_t* outputVector,
+ intInputVal3 = _mm256_permute4x64_epi64(intInputVal3, 0b11011000);
+
+ intInputVal1 = _mm256_packs_epi16(intInputVal1, intInputVal3);
+- intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
++ __m256i intInputVal = _mm256_permute4x64_epi64(intInputVal1, 0b11011000);
+
+ _mm256_store_si256((__m256i*)outputVectorPtr, intInputVal);
+ outputVectorPtr += 32;
+ }
+
+- number = thirtysecondPoints * 32;
+- for (; number < num_points; number++) {
+- r = inputVector[number] * scalar;
++ for (unsigned int number = thirtysecondPoints * 32; number < num_points; number++) {
++ const float r = inputVector[number] * scalar;
+ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+ }
+ }
+@@ -373,31 +342,26 @@ static inline void volk_32f_s32f_convert_8i_a_sse2(int8_t* outputVector,
+ const float scalar,
+ unsigned int num_points)
+ {
+- unsigned int number = 0;
+-
+ const unsigned int sixteenthPoints = num_points / 16;
+
+ const float* inputVectorPtr = (const float*)inputVector;
+ int8_t* outputVectorPtr = outputVector;
+
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
+- float r;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
+
+- __m128 vScalar = _mm_set_ps1(scalar);
+- __m128 inputVal1, inputVal2, inputVal3, inputVal4;
+- __m128i intInputVal1, intInputVal2, intInputVal3, intInputVal4;
+- __m128 vmin_val = _mm_set_ps1(min_val);
+- __m128 vmax_val = _mm_set_ps1(max_val);
++ const __m128 vScalar = _mm_set_ps1(scalar);
+
+- for (; number < sixteenthPoints; number++) {
+- inputVal1 = _mm_load_ps(inputVectorPtr);
++ for (unsigned int number = 0; number < sixteenthPoints; number++) {
++ __m128 inputVal1 = _mm_load_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+- inputVal2 = _mm_load_ps(inputVectorPtr);
++ __m128 inputVal2 = _mm_load_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+- inputVal3 = _mm_load_ps(inputVectorPtr);
++ __m128 inputVal3 = _mm_load_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+- inputVal4 = _mm_load_ps(inputVectorPtr);
++ __m128 inputVal4 = _mm_load_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+
+ inputVal1 =
+@@ -409,10 +373,10 @@ static inline void volk_32f_s32f_convert_8i_a_sse2(int8_t* outputVector,
+ inputVal4 =
+ _mm_max_ps(_mm_min_ps(_mm_mul_ps(inputVal4, vScalar), vmax_val), vmin_val);
+
+- intInputVal1 = _mm_cvtps_epi32(inputVal1);
+- intInputVal2 = _mm_cvtps_epi32(inputVal2);
+- intInputVal3 = _mm_cvtps_epi32(inputVal3);
+- intInputVal4 = _mm_cvtps_epi32(inputVal4);
++ __m128i intInputVal1 = _mm_cvtps_epi32(inputVal1);
++ __m128i intInputVal2 = _mm_cvtps_epi32(inputVal2);
++ __m128i intInputVal3 = _mm_cvtps_epi32(inputVal3);
++ __m128i intInputVal4 = _mm_cvtps_epi32(inputVal4);
+
+ intInputVal1 = _mm_packs_epi32(intInputVal1, intInputVal2);
+ intInputVal3 = _mm_packs_epi32(intInputVal3, intInputVal4);
+@@ -423,9 +387,8 @@ static inline void volk_32f_s32f_convert_8i_a_sse2(int8_t* outputVector,
+ outputVectorPtr += 16;
+ }
+
+- number = sixteenthPoints * 16;
+- for (; number < num_points; number++) {
+- r = inputVector[number] * scalar;
++ for (unsigned int number = sixteenthPoints * 16; number < num_points; number++) {
++ const float r = inputVector[number] * scalar;
+ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+ }
+ }
+@@ -440,40 +403,34 @@ static inline void volk_32f_s32f_convert_8i_a_sse(int8_t* outputVector,
+ const float scalar,
+ unsigned int num_points)
+ {
+- unsigned int number = 0;
+- size_t inner_loop;
+-
+ const unsigned int quarterPoints = num_points / 4;
+
+ const float* inputVectorPtr = (const float*)inputVector;
++ int8_t* outputVectorPtr = outputVector;
+
+- float min_val = INT8_MIN;
+- float max_val = INT8_MAX;
+- float r;
++ const float min_val = INT8_MIN;
++ const float max_val = INT8_MAX;
++ const __m128 vmin_val = _mm_set_ps1(min_val);
++ const __m128 vmax_val = _mm_set_ps1(max_val);
+
+- int8_t* outputVectorPtr = outputVector;
+- __m128 vScalar = _mm_set_ps1(scalar);
+- __m128 ret;
+- __m128 vmin_val = _mm_set_ps1(min_val);
+- __m128 vmax_val = _mm_set_ps1(max_val);
++ const __m128 vScalar = _mm_set_ps1(scalar);
+
+ __VOLK_ATTR_ALIGNED(16) float outputFloatBuffer[4];
+
+- for (; number < quarterPoints; number++) {
+- ret = _mm_load_ps(inputVectorPtr);
++ for (unsigned int number = 0; number < quarterPoints; number++) {
++ __m128 ret = _mm_load_ps(inputVectorPtr);
+ inputVectorPtr += 4;
+
+ ret = _mm_max_ps(_mm_min_ps(_mm_mul_ps(ret, vScalar), vmax_val), vmin_val);
+
+ _mm_store_ps(outputFloatBuffer, ret);
+- for (inner_loop = 0; inner_loop < 4; inner_loop++) {
++ for (size_t inner_loop = 0; inner_loop < 4; inner_loop++) {
+ *outputVectorPtr++ = (int8_t)(rintf(outputFloatBuffer[inner_loop]));
+ }
+ }
+
+- number = quarterPoints * 4;
+- for (; number < num_points; number++) {
+- r = inputVector[number] * scalar;
++ for (unsigned int number = quarterPoints * 4; number < num_points; number++) {
++ const float r = inputVector[number] * scalar;
+ volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+ }
+ }
+@@ -481,24 +438,4 @@ static inline void volk_32f_s32f_convert_8i_a_sse(int8_t* outputVector,
+ #endif /* LV_HAVE_SSE */
+
+
+-#ifdef LV_HAVE_GENERIC
+-
+-static inline void volk_32f_s32f_convert_8i_a_generic(int8_t* outputVector,
+- const float* inputVector,
+- const float scalar,
+- unsigned int num_points)
+-{
+- const float* inputVectorPtr = inputVector;
+- unsigned int number = 0;
+- float r;
+-
+- for (number = 0; number < num_points; number++) {
+- r = *inputVectorPtr++ * scalar;
+- volk_32f_s32f_convert_8i_single(&outputVector[number], r);
+- }
+-}
+-
+-#endif /* LV_HAVE_GENERIC */
+-
+-
+ #endif /* INCLUDED_volk_32f_s32f_convert_8i_a_H */
+--
+2.39.2
+
--- /dev/null
+From f3ee79c2c366f79e3e59b5deec438d4145e12cd3 Mon Sep 17 00:00:00 2001
+From: "A. Maitland Bottoms" <bottoms@debian.org>
+Date: Sun, 4 Sep 2022 21:20:13 -0400
+Subject: [PATCH] omit build path
+
+Have CMake filter out build path from COMPILER_INFO
+before using the string in constants.c
+
+Signed-off-by: A. Maitland Bottoms <bottoms@debian.org>
+---
+ lib/CMakeLists.txt | 1 +
+ 1 file changed, 1 insertion(+)
+
+diff --git a/lib/CMakeLists.txt b/lib/CMakeLists.txt
+index 75055ee..412cb11 100644
+--- a/lib/CMakeLists.txt
++++ b/lib/CMakeLists.txt
+@@ -454,6 +454,7 @@ message(STATUS "Loading version ${VERSION} into constants...")
+
+ #double escape for windows backslash path separators
+ string(REPLACE "\\" "\\\\" prefix "${prefix}")
++string(REPLACE "${CMAKE_SOURCE_DIR}" "$BUILD_DIR" COMPILER_INFO "${COMPILER_INFO}")
+
+ configure_file(
+ ${CMAKE_CURRENT_SOURCE_DIR}/constants.c.in
+--
+2.35.1
+
--- /dev/null
+From 58cc2b105211f0e5beab4dc228b478ebe105be06 Mon Sep 17 00:00:00 2001
+From: "A. Maitland Bottoms" <bottoms@debian.org>
+Date: Sun, 4 Sep 2022 21:37:45 -0400
+Subject: [PATCH] omit doxygen build paths
+
+Use reproducible-builds friendly configuration settings.
+
+Signed-off-by: A. Maitland Bottoms <bottoms@debian.org>
+---
+ docs/Doxyfile.in | 14 +++++++-------
+ 1 file changed, 7 insertions(+), 7 deletions(-)
+
+diff --git a/docs/Doxyfile.in b/docs/Doxyfile.in
+index 70913f5..a0fb23f 100644
+--- a/docs/Doxyfile.in
++++ b/docs/Doxyfile.in
+@@ -157,7 +157,7 @@ FULL_PATH_NAMES = YES
+ # will be relative from the directory where doxygen is started.
+ # This tag requires that the tag FULL_PATH_NAMES is set to YES.
+
+-STRIP_FROM_PATH =
++STRIP_FROM_PATH = @CMAKE_BINARY_DIR@ @CMAKE_SOURCE_DIR@
+
+ # The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
+ # path mentioned in the documentation of a class, which tells the reader which
+@@ -166,7 +166,7 @@ STRIP_FROM_PATH =
+ # specify the list of include paths that are normally passed to the compiler
+ # using the -I flag.
+
+-STRIP_FROM_INC_PATH =
++STRIP_FROM_INC_PATH = @CMAKE_SOURCE_DIR@/include @CMAKE_BINARY_DIR@ @CMAKE_SOURCE_DIR@/lib
+
+ # If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
+ # less readable) file names. This can be useful is your file systems doesn't
+@@ -637,7 +637,7 @@ MAX_INITIALIZER_LINES = 30
+ # will mention the files that were used to generate the documentation.
+ # The default value is: YES.
+
+-SHOW_USED_FILES = YES
++SHOW_USED_FILES = NO
+
+ # Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
+ # will remove the Files entry from the Quick Index and from the Folder Tree View
+@@ -832,7 +832,7 @@ RECURSIVE = YES
+ # Note that relative paths are relative to the directory from which doxygen is
+ # run.
+
+-EXCLUDE = @CMAKE_BINARY_DIR@ @CMAKE_SOURCE_DIR@/cpu_features @CMAKE_SOURCE_DIR@/README.md @CMAKE_SOURCE_DIR@/docs/AUTHORS_RESUBMITTING_UNDER_LGPL_LICENSE.md
++EXCLUDE = @CMAKE_BINARY_DIR@ @CMAKE_SOURCE_DIR@/cpu_features @CMAKE_SOURCE_DIR@/README.md @CMAKE_SOURCE_DIR@/cmake @CMAKE_SOURCE_DIR@/docs/AUTHORS_RESUBMITTING_UNDER_LGPL_LICENSE.md @CMAKE_SOURCE_DIR@/apps @CMAKE_SOURCE_DIR@/lib/*qa* @CMAKE_SOURCE_DIR@/tmpl
+
+ # The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
+ # directories that are symbolic links (a Unix file system feature) are excluded
+@@ -979,7 +979,7 @@ REFERENCES_RELATION = NO
+ # link to the documentation.
+ # The default value is: YES.
+
+-REFERENCES_LINK_SOURCE = YES
++REFERENCES_LINK_SOURCE = NO
+
+ # If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
+ # source code will show a tooltip with additional information such as prototype,
+@@ -989,7 +989,7 @@ REFERENCES_LINK_SOURCE = YES
+ # The default value is: YES.
+ # This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+-SOURCE_TOOLTIPS = YES
++SOURCE_TOOLTIPS = NO
+
+ # If the USE_HTAGS tag is set to YES then the references to source code will
+ # point to the HTML generated by the htags(1) tool instead of doxygen built-in
+@@ -1099,7 +1099,7 @@ HTML_HEADER =
+ # that doxygen normally uses.
+ # This tag requires that the tag GENERATE_HTML is set to YES.
+
+-HTML_FOOTER =
++HTML_FOOTER = ""
+
+ # The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
+ # sheet that is used by each HTML page. It can be used to fine-tune the look of
+--
+2.35.1
+
--- /dev/null
+Author: A. Maitland Bottoms <bottoms@debian.org>
+Description: optional static apps
+ For Debian, build apps with static libs if ENABLE_STATIC_APPS.
+
+--- a/apps/CMakeLists.txt
++++ b/apps/CMakeLists.txt
+@@ -44,7 +44,7 @@
+ endif()
+ target_link_libraries(volk_profile PRIVATE std::filesystem)
+
+-if(ENABLE_STATIC_LIBS)
++if(ENABLE_STATIC_LIBS AND ENABLE_STATIC_APPS)
+ target_link_libraries(volk_profile PRIVATE volk_static)
+ set_target_properties(volk_profile PROPERTIES LINK_FLAGS "-static")
+ else()
+@@ -61,7 +61,7 @@
+ add_executable(volk-config-info volk-config-info.cc ${CMAKE_CURRENT_SOURCE_DIR}/volk_option_helpers.cc
+ )
+
+-if(ENABLE_STATIC_LIBS)
++if(ENABLE_STATIC_LIBS AND ENABLE_STATIC_APPS)
+ target_link_libraries(volk-config-info volk_static)
+ set_target_properties(volk-config-info PROPERTIES LINK_FLAGS "-static")
+ else()
--- /dev/null
+Author: A. Maitland Bottoms <bottoms@debian.org>
+Description: remove external HTML resources
+ Debian packages should not generate traffic to external web services.
+
+--- a/README.md
++++ b/README.md
+@@ -1,8 +1,3 @@
+-[](https://travis-ci.com/gnuradio/volk) [](https://ci.appveyor.com/project/gnuradio/volk/branch/master)
+-
+-
+-[](https://doi.org/10.5281/zenodo.3360942)
+-
+ 
+
+ # Welcome to VOLK!
+--- a/cpu_features/README.md
++++ b/cpu_features/README.md
+@@ -1,14 +1,4 @@
+ # cpu_features
+-[![Linux Status][linux_svg]][linux_link]
+-[![Macos Status][macos_svg]][macos_link]
+-[![Windows Status][windows_svg]][windows_link]
+-
+-[linux_svg]: https://github.com/google/cpu_features/actions/workflows/amd64_linux.yml/badge.svg?branch=master
+-[linux_link]: https://github.com/google/cpu_features/actions/workflows/amd64_linux.yml
+-[macos_svg]: https://github.com/google/cpu_features/actions/workflows/amd64_macos.yml/badge.svg?branch=master
+-[macos_link]: https://github.com/google/cpu_features/actions/workflows/amd64_macos.yml
+-[windows_svg]: https://github.com/google/cpu_features/actions/workflows/amd64_windows.yml/badge.svg?branch=master
+-[windows_link]: https://github.com/google/cpu_features/actions/workflows/amd64_windows.yml
+
+ A cross-platform C library to retrieve CPU features (such as available
+ instructions) at runtime.
--- /dev/null
+0001-ci-Remove-license-check.patch
+0002-license-Fix-SPDX-identifiers.patch
+# 0003-Use-cpu_features-on-RISC-V-platforms.patch
+0004-add-volk_32f_s32f_x2_convert_8u-kernel.patch
+0005-volk_32f_s32f_convert_8i-code-style.patch
+optional-static-apps
+skip-cpu_features-on-kfreebsd
+# remove-external-HTML-resources
+omit-build-paths
+omit-doxygen-build-paths
+update-doxygen
--- /dev/null
+Subject: skip cpu_freatures on kfreebsd
+Author: A. Maitland Bottoms <bottoms@debian.org>
+
+ Avoid #error "Unsupported OS" on kFreeBSD
+
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -120,8 +120,10 @@
+ ########################################################################
+
+ # cpu_features - sensible defaults, user settable option
+-if(CMAKE_SYSTEM_PROCESSOR MATCHES
+- "(^mips)|(^arm)|(^aarch64)|(x86_64)|(AMD64|amd64)|(^i.86$)|(^powerpc)|(^ppc)")
++message(STATUS "Building Volk for ${CMAKE_SYSTEM_NAME} on ${CMAKE_SYSTEM_PROCESSOR}")
++if((CMAKE_SYSTEM_PROCESSOR MATCHES
++ "(^mips)|(^arm)|(^aarch64)|(x86_64)|(AMD64|amd64)|(^i.86$)|(^powerpc)|(^ppc)")
++ AND (NOT CMAKE_SYSTEM_NAME MATCHES "kFreeBSD"))
+ option(VOLK_CPU_FEATURES "Volk uses cpu_features" ON)
+ else()
+ option(VOLK_CPU_FEATURES "Volk uses cpu_features" OFF)
--- /dev/null
+Author: A. Maitland Bottoms <bottoms@debian.org>
+Description: update doxygen
+ Debian has recent version that complains of these settings.
+
+--- a/docs/Doxyfile.in
++++ b/docs/Doxyfile.in
+@@ -235,12 +235,6 @@
+
+ ALIASES =
+
+-# This tag can be used to specify a number of word-keyword mappings (TCL only).
+-# A mapping has the form "name=value". For example adding "class=itcl::class"
+-# will allow you to use the command class in the itcl::class meaning.
+-
+-TCL_SUBST =
+-
+ # Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
+ # only. Doxygen will then generate output that is more tailored for C. For
+ # instance, some of the names that are used will be different. The list of all
+@@ -1032,13 +1026,6 @@
+
+ ALPHABETICAL_INDEX = YES
+
+-# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
+-# which the alphabetical index list will be split.
+-# Minimum value: 1, maximum value: 20, default value: 5.
+-# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+-
+-COLS_IN_ALPHA_INDEX = 5
+-
+ # In case all classes in a project start with a common prefix, all classes will
+ # be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
+ # can be used to specify a prefix (or a list of prefixes) that should be ignored
+@@ -1714,16 +1701,6 @@
+
+ LATEX_HIDE_INDICES = NO
+
+-# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
+-# code with syntax highlighting in the LaTeX output.
+-#
+-# Note that which sources are shown also depends on other settings such as
+-# SOURCE_BROWSER.
+-# The default value is: NO.
+-# This tag requires that the tag GENERATE_LATEX is set to YES.
+-
+-LATEX_SOURCE_CODE = NO
+-
+ # The LATEX_BIB_STYLE tag can be used to specify the style to use for the
+ # bibliography, e.g. plainnat, or ieeetr. See
+ # http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
+@@ -1843,18 +1820,6 @@
+
+ XML_OUTPUT = xml
+
+-# The XML_SCHEMA tag can be used to specify a XML schema, which can be used by a
+-# validating XML parser to check the syntax of the XML files.
+-# This tag requires that the tag GENERATE_XML is set to YES.
+-
+-XML_SCHEMA =
+-
+-# The XML_DTD tag can be used to specify a XML DTD, which can be used by a
+-# validating XML parser to check the syntax of the XML files.
+-# This tag requires that the tag GENERATE_XML is set to YES.
+-
+-XML_DTD =
+-
+ # If the XML_PROGRAMLISTING tag is set to YES doxygen will dump the program
+ # listings (including syntax highlighting and cross-referencing information) to
+ # the XML output. Note that enabling this will significantly increase the size
+@@ -2062,34 +2027,10 @@
+
+ EXTERNAL_PAGES = YES
+
+-# The PERL_PATH should be the absolute path and name of the perl script
+-# interpreter (i.e. the result of 'which perl').
+-# The default file (with absolute path) is: /usr/bin/perl.
+-
+-PERL_PATH = /usr/bin/perl
+-
+ #---------------------------------------------------------------------------
+ # Configuration options related to the dot tool
+ #---------------------------------------------------------------------------
+
+-# If the CLASS_DIAGRAMS tag is set to YES doxygen will generate a class diagram
+-# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
+-# NO turns the diagrams off. Note that this option also works with HAVE_DOT
+-# disabled, but it is recommended to install and use dot, since it yields more
+-# powerful graphs.
+-# The default value is: YES.
+-
+-CLASS_DIAGRAMS = NO
+-
+-# You can define message sequence charts within doxygen comments using the \msc
+-# command. Doxygen will then run the mscgen tool (see:
+-# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
+-# documentation. The MSCGEN_PATH tag allows you to specify the directory where
+-# the mscgen tool resides. If left empty the tool is assumed to be found in the
+-# default search path.
+-
+-MSCGEN_PATH =
+-
+ # You can include diagrams made with dia in doxygen documentation. Doxygen will
+ # then run dia to produce the diagram and insert it in the documentation. The
+ # DIA_PATH tag allows you to specify the directory where the dia binary resides.
+@@ -2226,7 +2167,7 @@
+ # The default value is: NO.
+ # This tag requires that the tag HAVE_DOT is set to YES.
+
+-CALL_GRAPH = NO
++CALL_GRAPH = YES
+
+ # If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
+ # dependency graph for every global function or class method.
--- /dev/null
+#!/usr/bin/make -f
+DEB_HOST_MULTIARCH ?= $(shell dpkg-architecture -qDEB_HOST_MULTIARCH)
+export DEB_HOST_MULTIARCH
+#export DH_VERBOSE=1
+
+%:
+ dh $@ --with python3
+
+override_dh_auto_configure:
+ dh_auto_configure -- -DLIB_SUFFIX="/$(DEB_HOST_MULTIARCH)" \
+ -DPYTHON_EXECUTABLE=/usr/bin/python3 \
+ -DCMAKE_BUILD_TYPE=RelWithDebInfo
+
+override_dh_auto_build-indep:
+ cmake --build obj-* --target all
+ cmake --build obj-* --target volk_doc
+
+override_dh_auto_install:
+ dh_auto_install
+ find debian -type d -empty -delete
+
+override_dh_auto_test:
+ - dh_auto_test -- CTEST_TEST_TIMEOUT=60
--- /dev/null
+3.0 (quilt)
--- /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
+.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
+.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
+version=4
+ opts="filenamemangle=s%(?:.*?)?v?(\d[\d.]*@ARCHIVE_EXT@)%@PACKAGE@-$1%,uversionmangle=s/-rc/~rc/" \
+ https://github.com/gnuradio/volk/tags \
+ (?:.*?/)?v?@ANY_VERSION@@ARCHIVE_EXT@
projects / volk.git / commitdiff