[04/10] i386: Replace assembly versions of e_exp2f with generic e_exp2f.c

Message ID 20171019173159.21402-5-hjl.tools@gmail.com
State New, archived
Headers

Commit Message

H.J. Lu Oct. 19, 2017, 5:31 p.m. UTC
  This patch replaces i386 assembly versions of e_exp2f with generic
e_exp2f.c.  For workload-spec2017.wrf, on Nehalem, it improves
performance by:

                           Before            After     Improvement
reciprocal-throughput      112.996          40.0454       182%
latency                    126.581          54.4479       132%

On Skylake, it improves performance by:

                           Before            After     Improvement
reciprocal-throughput      113.14           39.447        186%
latency                    136.068          55.684        144%

	* sysdeps/i386/fpu/e_exp2f.S: Removed.
	* sysdeps/i386/fpu/w_exp2f.c: Likewise.
	* sysdeps/i386/fpu/libm-test-ulps: Updated for generic e_exp2f.c.
	* sysdeps/i386/i686/fpu/multiarch/libm-test-ulps: Likewise.
	* sysdeps/i386/i686/fpu/multiarch/Makefile (libm-sysdep_routines):
	Add e_exp2f-sse2.
	(CFLAGS-e_exp2f-sse2.c): New.
	* sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c: New file.
	* sysdeps/i386/i686/fpu/multiarch/e_exp2f.c: Likewise.
---
 sysdeps/i386/fpu/e_exp2f.S                     | 52 --------------------------
 sysdeps/i386/fpu/libm-test-ulps                |  2 +
 sysdeps/i386/fpu/w_exp2f.c                     |  1 -
 sysdeps/i386/i686/fpu/multiarch/Makefile       |  3 +-
 sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c |  3 ++
 sysdeps/i386/i686/fpu/multiarch/e_exp2f.c      | 37 ++++++++++++++++++
 sysdeps/i386/i686/fpu/multiarch/libm-test-ulps |  2 +
 7 files changed, 46 insertions(+), 54 deletions(-)
 delete mode 100644 sysdeps/i386/fpu/e_exp2f.S
 delete mode 100644 sysdeps/i386/fpu/w_exp2f.c
 create mode 100644 sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c
 create mode 100644 sysdeps/i386/i686/fpu/multiarch/e_exp2f.c
  

Comments

H.J. Lu Oct. 19, 2017, 7:49 p.m. UTC | #1
On Thu, Oct 19, 2017 at 10:31 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> This patch replaces i386 assembly versions of e_exp2f with generic
> e_exp2f.c.  For workload-spec2017.wrf, on Nehalem, it improves
> performance by:
>
>                            Before            After     Improvement
> reciprocal-throughput      112.996          40.0454       182%
> latency                    126.581          54.4479       132%
>
> On Skylake, it improves performance by:
>
>                            Before            After     Improvement
> reciprocal-throughput      113.14           39.447        186%
> latency                    136.068          55.684        144%

On IvyBridge with --disable-multi-arch, it improves performance by:

                           Before            After     Improvement
reciprocal-throughput      132.521          40.3759       228%
latency                    145.791          58.4587       149%

>         * sysdeps/i386/fpu/e_exp2f.S: Removed.
>         * sysdeps/i386/fpu/w_exp2f.c: Likewise.
>         * sysdeps/i386/fpu/libm-test-ulps: Updated for generic e_exp2f.c.
>         * sysdeps/i386/i686/fpu/multiarch/libm-test-ulps: Likewise.
>         * sysdeps/i386/i686/fpu/multiarch/Makefile (libm-sysdep_routines):
>         Add e_exp2f-sse2.
>         (CFLAGS-e_exp2f-sse2.c): New.
>         * sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c: New file.
>         * sysdeps/i386/i686/fpu/multiarch/e_exp2f.c: Likewise.
> ---
>  sysdeps/i386/fpu/e_exp2f.S                     | 52 --------------------------
>  sysdeps/i386/fpu/libm-test-ulps                |  2 +
>  sysdeps/i386/fpu/w_exp2f.c                     |  1 -
>  sysdeps/i386/i686/fpu/multiarch/Makefile       |  3 +-
>  sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c |  3 ++
>  sysdeps/i386/i686/fpu/multiarch/e_exp2f.c      | 37 ++++++++++++++++++
>  sysdeps/i386/i686/fpu/multiarch/libm-test-ulps |  2 +
>  7 files changed, 46 insertions(+), 54 deletions(-)
>  delete mode 100644 sysdeps/i386/fpu/e_exp2f.S
>  delete mode 100644 sysdeps/i386/fpu/w_exp2f.c
>  create mode 100644 sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c
>  create mode 100644 sysdeps/i386/i686/fpu/multiarch/e_exp2f.c
>
> diff --git a/sysdeps/i386/fpu/e_exp2f.S b/sysdeps/i386/fpu/e_exp2f.S
> deleted file mode 100644
> index 30623cd850..0000000000
> --- a/sysdeps/i386/fpu/e_exp2f.S
> +++ /dev/null
> @@ -1,52 +0,0 @@
> -/*
> - * Written by J.T. Conklin <jtc@netbsd.org>.
> - * Adapted for exp2 by Ulrich Drepper <drepper@cygnus.com>.
> - * Public domain.
> - */
> -
> -#include <machine/asm.h>
> -#include <i386-math-asm.h>
> -
> -DEFINE_FLT_MIN
> -
> -#ifdef PIC
> -# define MO(op) op##@GOTOFF(%ecx)
> -#else
> -# define MO(op) op
> -#endif
> -
> -       .text
> -ENTRY(__ieee754_exp2f)
> -#ifdef  PIC
> -       LOAD_PIC_REG (cx)
> -#endif
> -       flds    4(%esp)
> -/* I added the following ugly construct because exp(+-Inf) resulted
> -   in NaN.  The ugliness results from the bright minds at Intel.
> -   For the i686 the code can be written better.
> -   -- drepper@cygnus.com.  */
> -       fxam                            /* Is NaN or +-Inf?  */
> -       fstsw   %ax
> -       movb    $0x45, %dh
> -       andb    %ah, %dh
> -       cmpb    $0x05, %dh
> -       je      1f                      /* Is +-Inf, jump.  */
> -       fld     %st
> -       frndint                         /* int(x) */
> -       fsubr   %st,%st(1)              /* fract(x) */
> -       fxch
> -       f2xm1                           /* 2^(fract(x)) - 1 */
> -       fld1
> -       faddp                           /* 2^(fract(x)) */
> -       fscale                          /* e^x */
> -       fstp    %st(1)
> -       FLT_NARROW_EVAL_UFLOW_NONNEG_NAN
> -       ret
> -
> -1:     testl   $0x200, %eax            /* Test sign.  */
> -       jz      2f                      /* If positive, jump.  */
> -       fstp    %st
> -       fldz                            /* Set result to 0.  */
> -2:     ret
> -END (__ieee754_exp2f)
> -strong_alias (__ieee754_exp2f, __exp2f_finite)
> diff --git a/sysdeps/i386/fpu/libm-test-ulps b/sysdeps/i386/fpu/libm-test-ulps
> index 60a721c000..5a319f147b 100644
> --- a/sysdeps/i386/fpu/libm-test-ulps
> +++ b/sysdeps/i386/fpu/libm-test-ulps
> @@ -1921,7 +1921,9 @@ ildouble: 1
>  ldouble: 1
>
>  Function: "exp2_upward":
> +float: 1
>  float128: 2
> +ifloat: 1
>  ifloat128: 2
>  ildouble: 1
>  ldouble: 1
> diff --git a/sysdeps/i386/fpu/w_exp2f.c b/sysdeps/i386/fpu/w_exp2f.c
> deleted file mode 100644
> index 583065d12a..0000000000
> --- a/sysdeps/i386/fpu/w_exp2f.c
> +++ /dev/null
> @@ -1 +0,0 @@
> -#include <sysdeps/../math/w_exp2f.c>
> diff --git a/sysdeps/i386/i686/fpu/multiarch/Makefile b/sysdeps/i386/i686/fpu/multiarch/Makefile
> index c5e8cfd689..2723f570af 100644
> --- a/sysdeps/i386/i686/fpu/multiarch/Makefile
> +++ b/sysdeps/i386/i686/fpu/multiarch/Makefile
> @@ -1,6 +1,7 @@
>  ifeq ($(subdir),math)
> -libm-sysdep_routines += e_expf-sse2 s_sinf-sse2 s_cosf-sse2 \
> +libm-sysdep_routines += e_exp2f-sse2 e_expf-sse2 s_sinf-sse2 s_cosf-sse2 \
>                          s_sincosf-sse2
>
> +CFLAGS-e_exp2f-sse2.c = -msse2 -mfpmath=sse
>  CFLAGS-e_expf-sse2.c = -msse2 -mfpmath=sse
>  endif
> diff --git a/sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c b/sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c
> new file mode 100644
> index 0000000000..dc3138dc24
> --- /dev/null
> +++ b/sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c
> @@ -0,0 +1,3 @@
> +#define __exp2f __exp2f_sse2
> +
> +#include <sysdeps/ieee754/flt-32/e_exp2f.c>
> diff --git a/sysdeps/i386/i686/fpu/multiarch/e_exp2f.c b/sysdeps/i386/i686/fpu/multiarch/e_exp2f.c
> new file mode 100644
> index 0000000000..0a5727f7de
> --- /dev/null
> +++ b/sysdeps/i386/i686/fpu/multiarch/e_exp2f.c
> @@ -0,0 +1,37 @@
> +/* Multiple versions of exp2f.
> +   Copyright (C) 2017 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library 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 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <http://www.gnu.org/licenses/>.  */
> +
> +extern float __redirect_exp2f (float);
> +
> +#define SYMBOL_NAME exp2f
> +#include "ifunc-sse2.h"
> +
> +libc_ifunc_redirected (__redirect_exp2f, __exp2f, IFUNC_SELECTOR ());
> +
> +#ifdef SHARED
> +# include <shlib-compat.h>
> +versioned_symbol (libm, __exp2f, exp2f, GLIBC_2_27);
> +#else
> +weak_alias (__exp2f, exp2f)
> +#endif
> +
> +strong_alias (__exp2f, __ieee754_exp2f)
> +strong_alias (__exp2f, __exp2f_finite)
> +
> +#define __exp2f __exp2f_ia32
> +#include <sysdeps/ieee754/flt-32/e_exp2f.c>
> diff --git a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
> index 5ff641f848..2d056a9260 100644
> --- a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
> +++ b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
> @@ -1921,7 +1921,9 @@ ildouble: 1
>  ldouble: 1
>
>  Function: "exp2_upward":
> +float: 1
>  float128: 2
> +ifloat: 1
>  ifloat128: 2
>  ildouble: 1
>  ldouble: 1
> --
> 2.13.6
>
  

Patch

diff --git a/sysdeps/i386/fpu/e_exp2f.S b/sysdeps/i386/fpu/e_exp2f.S
deleted file mode 100644
index 30623cd850..0000000000
--- a/sysdeps/i386/fpu/e_exp2f.S
+++ /dev/null
@@ -1,52 +0,0 @@ 
-/*
- * Written by J.T. Conklin <jtc@netbsd.org>.
- * Adapted for exp2 by Ulrich Drepper <drepper@cygnus.com>.
- * Public domain.
- */
-
-#include <machine/asm.h>
-#include <i386-math-asm.h>
-
-DEFINE_FLT_MIN
-
-#ifdef PIC
-# define MO(op) op##@GOTOFF(%ecx)
-#else
-# define MO(op) op
-#endif
-
-	.text
-ENTRY(__ieee754_exp2f)
-#ifdef  PIC
-	LOAD_PIC_REG (cx)
-#endif
-	flds	4(%esp)
-/* I added the following ugly construct because exp(+-Inf) resulted
-   in NaN.  The ugliness results from the bright minds at Intel.
-   For the i686 the code can be written better.
-   -- drepper@cygnus.com.  */
-	fxam				/* Is NaN or +-Inf?  */
-	fstsw	%ax
-	movb	$0x45, %dh
-	andb	%ah, %dh
-	cmpb	$0x05, %dh
-	je	1f			/* Is +-Inf, jump.  */
-	fld	%st
-	frndint				/* int(x) */
-	fsubr	%st,%st(1)		/* fract(x) */
-	fxch
-	f2xm1				/* 2^(fract(x)) - 1 */
-	fld1
-	faddp				/* 2^(fract(x)) */
-	fscale				/* e^x */
-	fstp	%st(1)
-	FLT_NARROW_EVAL_UFLOW_NONNEG_NAN
-	ret
-
-1:	testl	$0x200, %eax		/* Test sign.  */
-	jz	2f			/* If positive, jump.  */
-	fstp	%st
-	fldz				/* Set result to 0.  */
-2:	ret
-END (__ieee754_exp2f)
-strong_alias (__ieee754_exp2f, __exp2f_finite)
diff --git a/sysdeps/i386/fpu/libm-test-ulps b/sysdeps/i386/fpu/libm-test-ulps
index 60a721c000..5a319f147b 100644
--- a/sysdeps/i386/fpu/libm-test-ulps
+++ b/sysdeps/i386/fpu/libm-test-ulps
@@ -1921,7 +1921,9 @@  ildouble: 1
 ldouble: 1
 
 Function: "exp2_upward":
+float: 1
 float128: 2
+ifloat: 1
 ifloat128: 2
 ildouble: 1
 ldouble: 1
diff --git a/sysdeps/i386/fpu/w_exp2f.c b/sysdeps/i386/fpu/w_exp2f.c
deleted file mode 100644
index 583065d12a..0000000000
--- a/sysdeps/i386/fpu/w_exp2f.c
+++ /dev/null
@@ -1 +0,0 @@ 
-#include <sysdeps/../math/w_exp2f.c>
diff --git a/sysdeps/i386/i686/fpu/multiarch/Makefile b/sysdeps/i386/i686/fpu/multiarch/Makefile
index c5e8cfd689..2723f570af 100644
--- a/sysdeps/i386/i686/fpu/multiarch/Makefile
+++ b/sysdeps/i386/i686/fpu/multiarch/Makefile
@@ -1,6 +1,7 @@ 
 ifeq ($(subdir),math)
-libm-sysdep_routines += e_expf-sse2 s_sinf-sse2 s_cosf-sse2 \
+libm-sysdep_routines += e_exp2f-sse2 e_expf-sse2 s_sinf-sse2 s_cosf-sse2 \
                         s_sincosf-sse2
 
+CFLAGS-e_exp2f-sse2.c = -msse2 -mfpmath=sse
 CFLAGS-e_expf-sse2.c = -msse2 -mfpmath=sse
 endif
diff --git a/sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c b/sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c
new file mode 100644
index 0000000000..dc3138dc24
--- /dev/null
+++ b/sysdeps/i386/i686/fpu/multiarch/e_exp2f-sse2.c
@@ -0,0 +1,3 @@ 
+#define __exp2f __exp2f_sse2
+
+#include <sysdeps/ieee754/flt-32/e_exp2f.c>
diff --git a/sysdeps/i386/i686/fpu/multiarch/e_exp2f.c b/sysdeps/i386/i686/fpu/multiarch/e_exp2f.c
new file mode 100644
index 0000000000..0a5727f7de
--- /dev/null
+++ b/sysdeps/i386/i686/fpu/multiarch/e_exp2f.c
@@ -0,0 +1,37 @@ 
+/* Multiple versions of exp2f.
+   Copyright (C) 2017 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library 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 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+extern float __redirect_exp2f (float);
+
+#define SYMBOL_NAME exp2f
+#include "ifunc-sse2.h"
+
+libc_ifunc_redirected (__redirect_exp2f, __exp2f, IFUNC_SELECTOR ());
+
+#ifdef SHARED
+# include <shlib-compat.h>
+versioned_symbol (libm, __exp2f, exp2f, GLIBC_2_27);
+#else
+weak_alias (__exp2f, exp2f)
+#endif
+
+strong_alias (__exp2f, __ieee754_exp2f)
+strong_alias (__exp2f, __exp2f_finite)
+
+#define __exp2f __exp2f_ia32
+#include <sysdeps/ieee754/flt-32/e_exp2f.c>
diff --git a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
index 5ff641f848..2d056a9260 100644
--- a/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
+++ b/sysdeps/i386/i686/fpu/multiarch/libm-test-ulps
@@ -1921,7 +1921,9 @@  ildouble: 1
 ldouble: 1
 
 Function: "exp2_upward":
+float: 1
 float128: 2
+ifloat: 1
 ifloat128: 2
 ildouble: 1
 ldouble: 1