package/fftw: add a bunch of optimisations (branch yem/fftw)
v6 cover package/fftw: add a bunch of optimisations (branch yem/fftw)
Hello All!
This series aggregates misc optimisation patches for out fftw package, based on work by Gwenhael and Guillaume.
Thanks to both for their previous work! :-)
Changes v5 -> v6:
- architecture optimisations are not available for all precisions (Thomas, Gwenhael)
Changes v4 -> v5:
- use --disable-XXX in first patch
- add architecture optimisations for x86 and ARM
- add other optimisations (threads, openMP, fast maths)
Regards,
Yann E. MORIN.
The following changes since commit c511ca937d4627df7d601a999945faa20fe4cc8d:
network-manager: add hash file (2015-03-16 14:19:30 +0100)
are available in the git repository at:
git://git.busybox.net/~ymorin/git/buildroot yem/fftw
for you to fetch changes up to 0842316b8be377f09daf4c2f6a8d05fd303237d8:
package/fftw: add option for fast maths (2015-03-16 14:37:43 +0100)
----------------------------------------------------------------
Gwenhael Goavec-Merou (1):
fftw: add compile precision option
Yann E. MORIN (6):
package/fftw: add x86 optimisations package/fftw: add ARM optimisations package/fftw: disable fortran
package/fftw: enable threads when the tooclhain has them package/fftw: enable OpenMP if the compiler has that package/fftw: add option for fast maths
package/fftw/Config.in | 63 ++++++++++++++++++++++++++++++++++++++++++++++++++ package/fftw/fftw.mk | 21 +++++++++++++++++ 2 files changed, 84 insertions(+)
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v6 1/7 fftw: add compile precision option
From: Gwenhael Goavec-Merou <[email protected]>
fftw has options to select compile precision between single, long-double and quad. These options are exclusives. This patch adds choice to select precision option.
--- package/fftw/Config.in | 38 ++++++++++++++++++++++++++++++++++++++ package/fftw/fftw.mk | 4 ++++ 2 files changed, 42 insertions(+)
package/fftw/Config.in +38 -0
@@ -9,3 +9,41 @@ config BR2_PACKAGE_FFTW double precision. http://www.fftw.org + +if BR2_PACKAGE_FFTW + +choice + prompt "fftw precision" + default BR2_PACKAGE_FFTW_PRECISION_NONE + help + Selects fftw precision + +config BR2_PACKAGE_FFTW_PRECISION_NONE + bool "none" + help + no compile precision selected + +config BR2_PACKAGE_FFTW_PRECISION_SINGLE + bool "single precision" + help + compile fftw in single precision + +config BR2_PACKAGE_FFTW_PRECISION_LONG_DOUBLE + bool "long double precision" + # long-double precision require long-double trigonometric routines + depends on !(BR2_TOOLCHAIN_BUILDROOT_UCLIBC && \ + (BR2_arm || BR2_mips || BR2_mipsel)) + + help + compile fftw in long-double precision + +config BR2_PACKAGE_FFTW_PRECISION_QUAD + bool "quad precision" + # quad-precision needs to have a gcc with libquadmath + depends on (BR2_i386 || BR2_x86_64) && BR2_USE_WCHAR + help + compile fftw in quadruple precision if available + +endchoice + +endif
package/fftw/fftw.mk +4 -0
@@ -10,4 +10,8 @@ FFTW_INSTALL_STAGING = YES FFTW_LICENSE = GPLv2+ FFTW_LICENSE_FILES = COPYING +FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_SINGLE),--enable,--disable)-single +FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_LONG_DOUBLE),--enable,--disable)-long-double +FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_QUAD),--enable,--disable)-quad-precision + $(eval $(autotools-package))
Signature · 1 line
1.9.1
PATCH v6 2/7 package/fftw: add x86 optimisations
In reply to Yann E. MORIN---
Changes v5 -> v6:
- SSE/SSE2 can't be enabled for all precisions (Gwenhael)
--- package/fftw/Config.in | 10 +++++++++- package/fftw/fftw.mk | 4 ++++ 2 files changed, 13 insertions(+), 1 deletion(-)
package/fftw/Config.in +9 -1
@@ -12,6 +12,12 @@ config BR2_PACKAGE_FFTW if BR2_PACKAGE_FFTW +config BR2_PACKAGE_FFTW_USE_SSE + bool + +config BR2_PACKAGE_FFTW_USE_SSE2 + bool + choice prompt "fftw precision" default BR2_PACKAGE_FFTW_PRECISION_NONE
@@ -25,6 +31,8 @@ config BR2_PACKAGE_FFTW_PRECISION_NONE config BR2_PACKAGE_FFTW_PRECISION_SINGLE bool "single precision" + select BR2_PACKAGE_FFTW_USE_SSE if BR2_X86_CPU_HAS_SSE + select BR2_PACKAGE_FFTW_USE_SSE2 if BR2_X86_CPU_HAS_SSE2 help compile fftw in single precision
@@ -33,7 +41,7 @@ config BR2_PACKAGE_FFTW_PRECISION_LONG_DOUBLE # long-double precision require long-double trigonometric routines depends on !(BR2_TOOLCHAIN_BUILDROOT_UCLIBC && \ (BR2_arm || BR2_mips || BR2_mipsel)) - + select BR2_PACKAGE_FFTW_USE_SSE2 if BR2_X86_CPU_HAS_SSE2 help compile fftw in long-double precision
package/fftw/fftw.mk +4 -0
@@ -14,4 +14,8 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_SINGLE),--enable,--disable)- FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_LONG_DOUBLE),--enable,--disable)-long-double FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_QUAD),--enable,--disable)-quad-precision +# x86 optimisations +FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_SSE),--enable,--disable)-sse +FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_SSE2),--enable,--disable)-sse2 + $(eval $(autotools-package))
Signature · 1 line
1.9.1
PATCH v6 3/7 package/fftw: add ARM optimisations
In reply to Yann E. MORIN---
Changes v5 -> v6:
- NEON can't be used by all precisions (Gwenhael)
--- package/fftw/Config.in | 4 ++++ package/fftw/fftw.mk | 3 +++ 2 files changed, 7 insertions(+)
package/fftw/Config.in +4 -0
@@ -18,6 +18,9 @@ config BR2_PACKAGE_FFTW_USE_SSE config BR2_PACKAGE_FFTW_USE_SSE2 bool +config BR2_PACKAGE_FFTW_USE_NEON + bool + choice prompt "fftw precision" default BR2_PACKAGE_FFTW_PRECISION_NONE
@@ -33,6 +36,7 @@ config BR2_PACKAGE_FFTW_PRECISION_SINGLE bool "single precision" select BR2_PACKAGE_FFTW_USE_SSE if BR2_X86_CPU_HAS_SSE select BR2_PACKAGE_FFTW_USE_SSE2 if BR2_X86_CPU_HAS_SSE2 + select BR2_PACKAGE_FFTW_USE_NEON if BR2_ARM_FPU_NEON help compile fftw in single precision
package/fftw/fftw.mk +3 -0
@@ -18,4 +18,7 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_QUAD),--enable,--disable)-qu FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_SSE),--enable,--disable)-sse FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_SSE2),--enable,--disable)-sse2 +# ARM optimisations +FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_NEON),--enable,--disable)-neon + $(eval $(autotools-package))
Signature · 1 line
1.9.1
PATCH v6 3/7 package/fftw: add ARM optimisations
Dear Yann E. MORIN
On Mon, 16 Mar 2015 14:54:01 +0100
"Yann E. MORIN" <[email protected]> wrote:
[SNIP]
package/fftw/Config.in config BR2_PACKAGE_FFTW_PRECISION_SINGLE · line 40
Quoted 9 lines
choice
package/fftw/Config.in +1
-0prompt "fftw precision" default BR2_PACKAGE_FFTW_PRECISION_NONE@@ -33,6 +36,7 @@ config BR2_PACKAGE_FFTW_PRECISION_SINGLE bool "single precision" select BR2_PACKAGE_FFTW_USE_SSE if BR2_X86_CPU_HAS_SSE select BR2_PACKAGE_FFTW_USE_SSE2 if BR2_X86_CPU_HAS_SSE2 + select BR2_PACKAGE_FFTW_USE_NEON if BR2_ARM_FPU_NEON help
[SNIP]
s/BR2_ARM_FPU_NEON/BR2_ARM_CPU_HAS_NEON/
Regards
Gwen
PATCH v6 3/7 package/fftw: add ARM optimisations
Gwenhael, All,
On 2015-03-16 15:30 +0100, gwenhael.goavec spake thusly:
Quoted 4 lines
Dear Yann E. MORIN
On Mon, 16 Mar 2015 14:54:01 +0100
"Yann E. MORIN" <[email protected]> wrote:
[SNIP]
Quoted 11 lines
package/fftw/Config.in config BR2_PACKAGE_FFTW_PRECISION_SINGLE · line 40
Quoted 9 lines
choice
package/fftw/Config.in +1
-0prompt "fftw precision" default BR2_PACKAGE_FFTW_PRECISION_NONE@@ -33,6 +36,7 @@ config BR2_PACKAGE_FFTW_PRECISION_SINGLE bool "single precision" select BR2_PACKAGE_FFTW_USE_SSE if BR2_X86_CPU_HAS_SSE select BR2_PACKAGE_FFTW_USE_SSE2 if BR2_X86_CPU_HAS_SSE2 + select BR2_PACKAGE_FFTW_USE_NEON if BR2_ARM_FPU_NEON help[SNIP]
s/BR2_ARM_FPU_NEON/BR2_ARM_CPU_HAS_NEON/
No, because even if the CPU has NEON, it is not enough for NEOn to be enabled.
That's what I initially tried, but it did not work, as fftw compiles parts of its code without specifying -mfpu=neon, and thus it falls to the value specified by Buildroot, and the compilation breaks.
I should probably have said so in the commit log, indeed...
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v6 3/7 package/fftw: add ARM optimisations
Dear Yann E. MORIN,
Quoted 3 lines · Yann E. MORIN
That's what I initially tried, but it did not work, as fftw compiles parts of its code without specifying -mfpu=neon, and thus it falls to the value specified by Buildroot, and the compilation breaks.
In this case, pass -mfpu=neon explicitly, like we do in mplayer and vo-aacenc. Using NEON as the general purpose FPU is quite unlikely, since using NEON for FPU is not IEEE 754. So most likely people will use VFPv3/4 as the FPU, but may be interested in using NEON for specific things.
Thomas
Signature · 3 lines
Thomas Petazzoni, CTO, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
PATCH v6 4/7 package/fftw: disable fortran
In reply to Yann E. MORINWe do not have any fortran package and we do not build a fortran compiler.
Still, an external toolchain may have a fortran compiler and that is totally useless to us, since we anyway have no fortran package...
So, entirely disable fortran support; that will at least slightly speed up the build in case an external toolchain has a fortran compiler.
--- package/fftw/fftw.mk | 2 ++ 1 file changed, 2 insertions(+)
package/fftw/fftw.mk +2 -0
@@ -10,6 +10,8 @@ FFTW_INSTALL_STAGING = YES FFTW_LICENSE = GPLv2+ FFTW_LICENSE_FILES = COPYING +FFTW_CONF_OPTS = --disable-fortran + FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_SINGLE),--enable,--disable)-single FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_LONG_DOUBLE),--enable,--disable)-long-double FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_QUAD),--enable,--disable)-quad-precision
Signature · 1 line
1.9.1
PATCH v6 4/7 package/fftw: disable fortran
Quoted 6 lines · On Mon, 16 Mar 2015 14:54:02 +0100 "Yann E. MORIN" <[email protected]>…
We do not have any fortran package and we do not build a fortran compiler.
Still, an external toolchain may have a fortran compiler and that is totally useless to us, since we anyway have no fortran package...
Quoted 7 lines
So, entirely disable fortran support; that will at least slightly speed up the build in case an external toolchain has a fortran compiler.
Signed-off-by "Yann E. MORIN" <[email protected]>[based on a patch by Guillaume]Cc guillaume william brs <[email protected]>Cc Gwenhael Goavec-Merou <[email protected]>
Quoted 20 lines
--- package/fftw/fftw.mk | 2 ++ 1 file changed, 2 insertions(+)package/fftw/fftw.mk +2
-0@@ -10,6 +10,8 @@ FFTW_INSTALL_STAGING = YES FFTW_LICENSE = GPLv2+ FFTW_LICENSE_FILES = COPYING +FFTW_CONF_OPTS = --disable-fortran + FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_SINGLE),--enable,--disable)-single FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_LONG_DOUBLE),--enable,--disable)-long-double FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_PRECISION_QUAD),--enable,--disable)-quad-precisionSignature · 1 line
1.9.1
PATCH v6 5/7 package/fftw: enable threads when the tooclhain has them
In reply to Yann E. MORIN--- package/fftw/fftw.mk | 3 +++ 1 file changed, 3 insertions(+)
package/fftw/fftw.mk +3 -0
@@ -23,4 +23,7 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_SSE2),--enable,--disable)-sse2 # ARM optimisations FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_NEON),--enable,--disable)-neon +# Generic optimisations +FFTW_CONF_OPTS += $(if $(BR2_TOOLCHAIN_HAS_THREADS),--enable,--disable)-threads + $(eval $(autotools-package))
Signature · 1 line
1.9.1
PATCH v6 5/7 package/fftw: enable threads when the tooclhain has them
Quoted 4 lines · On Mon, 16 Mar 2015 14:54:03 +0100 "Yann E. MORIN" <[email protected]>…
Signed-off-by "Yann E. MORIN" <[email protected]>[based on a patch by Guillaume]Cc guillaume william brs <[email protected]>Cc Gwenhael Goavec-Merou <[email protected]>
Quoted 19 lines
--- package/fftw/fftw.mk | 3 +++ 1 file changed, 3 insertions(+)package/fftw/fftw.mk +3
-0@@ -23,4 +23,7 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_SSE2),--enable,--disable)-sse2 # ARM optimisations FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_NEON),--enable,--disable)-neon +# Generic optimisations +FFTW_CONF_OPTS += $(if $(BR2_TOOLCHAIN_HAS_THREADS),--enable,--disable)-threads + $(eval $(autotools-package))Signature · 1 line
1.9.1
PATCH v6 6/7 package/fftw: enable OpenMP if the compiler has that
In reply to Yann E. MORINUnfortunately, this is limited to internal toolchains; we currently have no way to know if an external toolchain has support for OpenMP.
--- package/fftw/fftw.mk | 1 + 1 file changed, 1 insertion(+)
package/fftw/fftw.mk +1 -0
@@ -25,5 +25,6 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_NEON),--enable,--disable)-neon # Generic optimisations FFTW_CONF_OPTS += $(if $(BR2_TOOLCHAIN_HAS_THREADS),--enable,--disable)-threads +FFTW_CONF_OPTS += $(if $(BR2_GCC_ENABLE_OPENMP),--enable-openmp,--disable-openmp) $(eval $(autotools-package))
Signature · 1 line
1.9.1
PATCH v6 6/7 package/fftw: enable OpenMP if the compiler has that
Quoted 6 lines · On Mon, 16 Mar 2015 14:54:04 +0100 "Yann E. MORIN" <[email protected]>…
Unfortunately, this is limited to internal toolchains; we currently have no way to know if an external toolchain has support for OpenMP.
Signed-off-by "Yann E. MORIN" <[email protected]>Cc guillaume william brs <[email protected]>Cc Gwenhael Goavec-Merou <[email protected]>
Quoted 18 lines
--- package/fftw/fftw.mk | 1 + 1 file changed, 1 insertion(+)package/fftw/fftw.mk +1
-0@@ -25,5 +25,6 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_NEON),--enable,--disable)-neon # Generic optimisations FFTW_CONF_OPTS += $(if $(BR2_TOOLCHAIN_HAS_THREADS),--enable,--disable)-threads +FFTW_CONF_OPTS += $(if $(BR2_GCC_ENABLE_OPENMP),--enable-openmp,--disable-openmp) $(eval $(autotools-package))Signature · 1 line
1.9.1
PATCH v6 7/7 package/fftw: add option for fast maths
In reply to Yann E. MORIN--- package/fftw/Config.in | 13 +++++++++++++ package/fftw/fftw.mk | 4 ++++ 2 files changed, 17 insertions(+)
package/fftw/Config.in +13 -0
@@ -58,4 +58,17 @@ config BR2_PACKAGE_FFTW_PRECISION_QUAD endchoice +config BR2_PACKAGE_FFTW_FAST + bool "optimise for speed over accuracy" + help + Optimise for fast math functions, at the expense of accuracy. + + Say 'y' if you need speed and can live with inaccuracies in + the results. Say 'n' (the default) if accuracy is of utmost + importance. + + This basically uses gcc's -Ofast optimisation level, which in + turn is basically using gcc's -ffast-math. See the gcc manual + for what this means. + endif
package/fftw/fftw.mk +4 -0
@@ -27,4 +27,8 @@ FFTW_CONF_OPTS += $(if $(BR2_PACKAGE_FFTW_USE_NEON),--enable,--disable)-neon FFTW_CONF_OPTS += $(if $(BR2_TOOLCHAIN_HAS_THREADS),--enable,--disable)-threads FFTW_CONF_OPTS += $(if $(BR2_GCC_ENABLE_OPENMP),--enable-openmp,--disable-openmp) +ifeq ($(BR2_PACKAGE_FFTW_FAST),y) +FFTW_CONF_OPTS += CFLAGS="$(TARGET_CFLAGS) -Ofast" +endif + $(eval $(autotools-package))
Signature · 1 line
1.9.1