public inbox for newlib@sourceware.org
 help / color / mirror / Atom feed
* [PATCH] Fix error in exp in magnitude [2e-32,2e-28]
@ 2020-03-06 14:49 Fabian Schriever
  2020-03-09  9:12 ` Corinna Vinschen
  0 siblings, 1 reply; 2+ messages in thread
From: Fabian Schriever @ 2020-03-06 14:49 UTC (permalink / raw)
  To: newlib; +Cc: Fabian Schriever

While testing the exp function we noticed some errors at the specified
magnitude. Within this range the exp function returns the input value +1
as an output. We chose to run a test of 1m exponentially spaced values
in the ranges [-2^-27,-2^-32] and [2^-32,2^-27] which showed 7603 and
3912 results with an error of >=0.5 ULP (compared with MPFR in 128 bit)
with the highest being 0.56 ULP and 0.53 ULP.

It's easy to fix by changing the magnitude at which the input value +1
is returned from <2^-28 to <2^-32 and using the polynomial instead. This
reduces the number of results with an error of >=0.5 ULP to 485 and 479
in above tests, all of which are exactly 0.5 ULP.

As we were already checking on exp we also took a look at expf. For expf
the magnitude where the input value +1 is returned can be increased from
<2^-28 to <2^-23 without accuracy loss for a slight performance
improvement. To ensure this was the correct value we tested all values
in the ranges [-2^-17,-2^-28] and [2^-28,2^-17] (~92.3m values each).
---
 newlib/libm/math/e_exp.c  | 2 +-
 newlib/libm/math/ef_exp.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/newlib/libm/math/e_exp.c b/newlib/libm/math/e_exp.c
index 81ea64dfb..d23b1162b 100644
--- a/newlib/libm/math/e_exp.c
+++ b/newlib/libm/math/e_exp.c
@@ -142,7 +142,7 @@ P5   =  4.13813679705723846039e-08; /* 0x3E663769, 0x72BEA4D0 */
 	    }
 	    x  = hi - lo;
 	} 
-	else if(hx < 0x3e300000)  {	/* when |x|<2**-28 */
+	else if(hx < 0x3df00000)  {	/* when |x|<2**-32 */
 	    if(huge+x>one) return one+x;/* trigger inexact */
 	}
 
diff --git a/newlib/libm/math/ef_exp.c b/newlib/libm/math/ef_exp.c
index e817370ac..fb3e2ffe6 100644
--- a/newlib/libm/math/ef_exp.c
+++ b/newlib/libm/math/ef_exp.c
@@ -77,7 +77,7 @@ P5   =  4.1381369442e-08; /* 0x3331bb4c */
 	    }
 	    x  = hi - lo;
 	} 
-	else if(hx < 0x31800000)  {	/* when |x|<2**-28 */
+	else if(hx < 0x34000000)  {	/* when |x|<2**-23 */
 	    if(huge+x>one) return one+x;/* trigger inexact */
 	}
 
-- 
2.24.1.windows.2


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] Fix error in exp in magnitude [2e-32,2e-28]
  2020-03-06 14:49 [PATCH] Fix error in exp in magnitude [2e-32,2e-28] Fabian Schriever
@ 2020-03-09  9:12 ` Corinna Vinschen
  0 siblings, 0 replies; 2+ messages in thread
From: Corinna Vinschen @ 2020-03-09  9:12 UTC (permalink / raw)
  To: newlib

[-- Attachment #1: Type: text/plain, Size: 2318 bytes --]

On Mar  6 15:46, Fabian Schriever wrote:
> While testing the exp function we noticed some errors at the specified
> magnitude. Within this range the exp function returns the input value +1
> as an output. We chose to run a test of 1m exponentially spaced values
> in the ranges [-2^-27,-2^-32] and [2^-32,2^-27] which showed 7603 and
> 3912 results with an error of >=0.5 ULP (compared with MPFR in 128 bit)
> with the highest being 0.56 ULP and 0.53 ULP.
> 
> It's easy to fix by changing the magnitude at which the input value +1
> is returned from <2^-28 to <2^-32 and using the polynomial instead. This
> reduces the number of results with an error of >=0.5 ULP to 485 and 479
> in above tests, all of which are exactly 0.5 ULP.
> 
> As we were already checking on exp we also took a look at expf. For expf
> the magnitude where the input value +1 is returned can be increased from
> <2^-28 to <2^-23 without accuracy loss for a slight performance
> improvement. To ensure this was the correct value we tested all values
> in the ranges [-2^-17,-2^-28] and [2^-28,2^-17] (~92.3m values each).
> ---
>  newlib/libm/math/e_exp.c  | 2 +-
>  newlib/libm/math/ef_exp.c | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/newlib/libm/math/e_exp.c b/newlib/libm/math/e_exp.c
> index 81ea64dfb..d23b1162b 100644
> --- a/newlib/libm/math/e_exp.c
> +++ b/newlib/libm/math/e_exp.c
> @@ -142,7 +142,7 @@ P5   =  4.13813679705723846039e-08; /* 0x3E663769, 0x72BEA4D0 */
>  	    }
>  	    x  = hi - lo;
>  	} 
> -	else if(hx < 0x3e300000)  {	/* when |x|<2**-28 */
> +	else if(hx < 0x3df00000)  {	/* when |x|<2**-32 */
>  	    if(huge+x>one) return one+x;/* trigger inexact */
>  	}
>  
> diff --git a/newlib/libm/math/ef_exp.c b/newlib/libm/math/ef_exp.c
> index e817370ac..fb3e2ffe6 100644
> --- a/newlib/libm/math/ef_exp.c
> +++ b/newlib/libm/math/ef_exp.c
> @@ -77,7 +77,7 @@ P5   =  4.1381369442e-08; /* 0x3331bb4c */
>  	    }
>  	    x  = hi - lo;
>  	} 
> -	else if(hx < 0x31800000)  {	/* when |x|<2**-28 */
> +	else if(hx < 0x34000000)  {	/* when |x|<2**-23 */
>  	    if(huge+x>one) return one+x;/* trigger inexact */
>  	}
>  
> -- 
> 2.24.1.windows.2
> 

Pushed.


Thanks,
Corinna

-- 
Corinna Vinschen
Cygwin Maintainer
Red Hat

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2020-03-09  9:13 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-03-06 14:49 [PATCH] Fix error in exp in magnitude [2e-32,2e-28] Fabian Schriever
2020-03-09  9:12 ` Corinna Vinschen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).