[PATCH] Update i386 libm-test-ulps

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[PATCH] Update i386 libm-test-ulps

Sourceware - libc-alpha mailing list


0001-Update-i686-libm-test-ulps.patch (4K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] Update i386 libm-test-ulps

Sourceware - libc-alpha mailing list
On 7/9/20 4:52 PM, Patsy Griffin via Libc-alpha wrote:
> From e1687cc8bdd02371f09f28bcfdabac8bfde22dc0 Mon Sep 17 00:00:00 2001
> From: Patsy Franklin <[hidden email]>
> Date: Thu, 9 Jul 2020 16:06:17 -0400
> Subject: [PATCH] Update i686 libm-test-ulps

OK for 2.32. Please push. Thanks for updating the ULPS based on different
hardware.

Reviewed-by: Carlos O'Donell <[hidden email]>

> Without my ULP patch these 18 tests fail on i686:
>  https://koji.fedoraproject.org/koji/taskinfo?taskID=46467301
>
> + cat /proc/cpuinfo
> processor : 0
> vendor_id : GenuineIntel
> cpu family : 6
> model : 85
> model name : Intel Xeon Processor (Cascadelake)
>
> FAIL: math/test-double-j0
> FAIL: math/test-double-y0
> FAIL: math/test-float-erfc
> FAIL: math/test-float-j0
> FAIL: math/test-float-j1
> FAIL: math/test-float-lgamma
> FAIL: math/test-float-tgamma
> FAIL: math/test-float-y0
> FAIL: math/test-float32-erfc
> FAIL: math/test-float32-j0
> FAIL: math/test-float32-j1
> FAIL: math/test-float32-lgamma
> FAIL: math/test-float32-tgamma
> FAIL: math/test-float32-y0
> FAIL: math/test-float32x-j0
> FAIL: math/test-float32x-y0
> FAIL: math/test-float64-j0
> FAIL: math/test-float64-y0
>
> With my ULP patch applied these tests now pass:
>  https://koji.fedoraproject.org/koji/taskinfo?taskID=46436310
> ---
>  sysdeps/i386/fpu/libm-test-ulps | 42 ++++++++++++++++-----------------
>  1 file changed, 21 insertions(+), 21 deletions(-)
>
> diff --git a/sysdeps/i386/fpu/libm-test-ulps b/sysdeps/i386/fpu/libm-test-ulps
> index 9e228f90a9..be30bbc66c 100644
> --- a/sysdeps/i386/fpu/libm-test-ulps
> +++ b/sysdeps/i386/fpu/libm-test-ulps
> @@ -1102,7 +1102,7 @@ ldouble: 1
>  
>  Function: "erfc":
>  double: 3
> -float: 2
> +float: 3
>  float128: 2
>  ldouble: 3
>  
> @@ -1212,7 +1212,7 @@ ldouble: 4
>  
>  Function: "gamma":
>  double: 4
> -float: 3
> +float: 5
>  ldouble: 4
>  
>  Function: "gamma_downward":
> @@ -1222,12 +1222,12 @@ ldouble: 7
>  
>  Function: "gamma_towardzero":
>  double: 5
> -float: 4
> +float: 6
>  ldouble: 7
>  
>  Function: "gamma_upward":
>  double: 5
> -float: 5
> +float: 6
>  ldouble: 6
>  
>  Function: "hypot":
> @@ -1252,7 +1252,7 @@ ldouble: 1
>  
>  Function: "j0":
>  double: 2
> -float: 5
> +float: 8
>  float128: 2
>  ldouble: 2
>  
> @@ -1263,20 +1263,20 @@ float128: 4
>  ldouble: 4
>  
>  Function: "j0_towardzero":
> -double: 3
> -float: 2
> +double: 4
> +float: 5
>  float128: 2
>  ldouble: 5
>  
>  Function: "j0_upward":
> -double: 3
> -float: 3
> +double: 4
> +float: 5
>  float128: 5
>  ldouble: 6
>  
>  Function: "j1":
>  double: 2
> -float: 7
> +float: 9
>  float128: 4
>  ldouble: 5
>  
> @@ -1324,7 +1324,7 @@ ldouble: 5
>  
>  Function: "lgamma":
>  double: 4
> -float: 3
> +float: 5
>  float128: 5
>  ldouble: 4
>  
> @@ -1336,13 +1336,13 @@ ldouble: 7
>  
>  Function: "lgamma_towardzero":
>  double: 5
> -float: 4
> +float: 6
>  float128: 5
>  ldouble: 7
>  
>  Function: "lgamma_upward":
>  double: 5
> -float: 5
> +float: 6
>  float128: 8
>  ldouble: 6
>  
> @@ -1579,37 +1579,37 @@ ldouble: 4
>  
>  Function: "tgamma":
>  double: 5
> -float: 4
> +float: 8
>  float128: 4
>  ldouble: 5
>  
>  Function: "tgamma_downward":
>  double: 6
> -float: 5
> +float: 7
>  float128: 5
>  ldouble: 5
>  
>  Function: "tgamma_towardzero":
>  double: 6
> -float: 4
> +float: 7
>  float128: 5
>  ldouble: 5
>  
>  Function: "tgamma_upward":
>  double: 5
> -float: 4
> +float: 8
>  float128: 4
>  ldouble: 5
>  
>  Function: "y0":
> -double: 2
> -float: 1
> +double: 3
> +float: 8
>  float128: 3
>  ldouble: 1
>  
>  Function: "y0_downward":
>  double: 3
> -float: 4
> +float: 6
>  float128: 4
>  ldouble: 5
>  
> @@ -1621,7 +1621,7 @@ ldouble: 6
>  
>  Function: "y0_upward":
>  double: 3
> -float: 5
> +float: 6
>  float128: 3
>  ldouble: 5
>  
> --
> 2.21.1
>


--
Cheers,
Carlos.