[PATCH] regenerate ulps for x86_64 after cf7ffdd

[PATCH] regenerate ulps for x86_64 after cf7ffdd

[Paul Zimmermann]

---
 math/auto-libm-test-out-hypot     | 25 +++++++++++++++++++++++++
 sysdeps/x86_64/fpu/libm-test-ulps |  1 +
 2 files changed, 26 insertions(+)

diff --git a/math/auto-libm-test-out-hypot b/math/auto-libm-test-out-hypot
index 92ab48df84..105307d29a 100644
--- a/math/auto-libm-test-out-hypot
+++ b/math/auto-libm-test-out-hypot
@@ -14765,3 +14765,28 @@ hypot -0x1.fa7deap+0 0x1.a761bab383ac8p+0
 = hypot tonearest ibm128 -0x1.fa7deap+0 0x1.a761bab383ac8p+0 : 0x2.942416eb158a10019b3e022619p+0 : inexact-ok
 = hypot towardzero ibm128 -0x1.fa7deap+0 0x1.a761bab383ac8p+0 : 0x2.942416eb158a10019b3e022619p+0 : inexact-ok
 = hypot upward ibm128 -0x1.fa7deap+0 0x1.a761bab383ac8p+0 : 0x2.942416eb158a10019b3e02261ap+0 : inexact-ok
+hypot -0x1.003222p-20 -0x1.6a2d58p-32
+= hypot downward binary32 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222p-20 : inexact-ok
+= hypot tonearest binary32 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222p-20 : inexact-ok
+= hypot towardzero binary32 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222p-20 : inexact-ok
+= hypot upward binary32 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003224p-20 : inexact-ok
+= hypot downward binary64 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffffp-20 : inexact-ok
+= hypot tonearest binary64 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffffp-20 : inexact-ok
+= hypot towardzero binary64 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffffp-20 : inexact-ok
+= hypot upward binary64 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003223p-20 : inexact-ok
+= hypot downward intel96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff408p-20 : inexact-ok
+= hypot tonearest intel96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40ap-20 : inexact-ok
+= hypot towardzero intel96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff408p-20 : inexact-ok
+= hypot upward intel96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40ap-20 : inexact-ok
+= hypot downward m68k96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff408p-20 : inexact-ok
+= hypot tonearest m68k96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40ap-20 : inexact-ok
+= hypot towardzero m68k96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff408p-20 : inexact-ok
+= hypot upward m68k96 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40ap-20 : inexact-ok
+= hypot downward binary128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfca5p-20 : inexact-ok
+= hypot tonearest binary128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfca5p-20 : inexact-ok
+= hypot towardzero binary128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfca5p-20 : inexact-ok
+= hypot upward binary128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfca6p-20 : inexact-ok
+= hypot downward ibm128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfc8p-20 : inexact-ok
+= hypot tonearest ibm128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfc8p-20 : inexact-ok
+= hypot towardzero ibm128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfc8p-20 : inexact-ok
+= hypot upward ibm128 -0x1.003222p-20 -0x1.6a2d58p-32 : 0x1.003222fffffff40977c94f4dfdp-20 : inexact-ok
diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps
index 0ef27deea8..9e47fdfa87 100644
--- a/sysdeps/x86_64/fpu/libm-test-ulps
+++ b/sysdeps/x86_64/fpu/libm-test-ulps
@@ -1574,6 +1574,7 @@ ldouble: 6
 
 Function: "hypot":
 double: 1
+float: 1
 float128: 1
 ldouble: 1
 
-- 
2.39.1

Re: [PATCH] regenerate ulps for x86_64 after cf7ffdd

[Florian Weimer]

* Paul Zimmermann via Libc-alpha:

> ---
>  math/auto-libm-test-out-hypot     | 25 +++++++++++++++++++++++++

I wouldn't expect this change in a “regenerate ulps” commit?

Thanks,
Florian

Re: [PATCH] regenerate ulps for x86_64 after cf7ffdd

[Paul Zimmermann]

       Hi Florian,

sorry, probably this change should have been part of cf7ffdd.
Should I modify the commit message, or split in two parts?

Paul

> From: Florian Weimer <fweimer@redhat.com>
> Cc: Paul Zimmermann <Paul.Zimmermann@inria.fr>
> Date: Tue, 14 Feb 2023 10:47:11 +0100
> 
> * Paul Zimmermann via Libc-alpha:
> 
> > ---
> >  math/auto-libm-test-out-hypot     | 25 +++++++++++++++++++++++++
> 
> I wouldn't expect this change in a “regenerate ulps” commit?
> 
> Thanks,
> Florian
> 
> 

Re: [PATCH] regenerate ulps for x86_64 after cf7ffdd

[Florian Weimer]

* Paul Zimmermann:

>        Hi Florian,
>
> sorry, probably this change should have been part of cf7ffdd.
> Should I modify the commit message, or split in two parts?

I'd suggest splitting it, given that other architectures might need
further adjustments after the second update.

Thanks,
Florian