From mzaki@e-mail.ne.jp Wed Apr 21 05:17:04 1999
Return-Path: <mzaki@e-mail.ne.jp>
Received: from work.mzaki.nom (47.pool6.tokyo.att.ne.jp [165.76.23.62])
	by hub.freebsd.org (Postfix) with ESMTP id 5C80514EA6
	for <FreeBSD-gnats-submit@freebsd.org>; Wed, 21 Apr 1999 05:17:02 -0700 (PDT)
	(envelope-from mzaki@e-mail.ne.jp)
Received: from localhost (localhost [127.0.0.1])
	by work.mzaki.nom (8.9.2/8.9.2) with ESMTP id VAA29020
	for <FreeBSD-gnats-submit@freebsd.org>; Wed, 21 Apr 1999 21:14:38 +0900 (JST)
	(envelope-from mzaki@e-mail.ne.jp)
Message-Id: <19990421211438O.mzaki@e-mail.ne.jp>
Date: Wed, 21 Apr 1999 21:14:38 +0900
From: Motomichi Matsuzaki <mzaki@e-mail.ne.jp>
Reply-To: mzaki@e-mail.ne.jp
To: FreeBSD-gnats-submit@freebsd.org
Subject: suggestion to refine npx.4
X-Send-Pr-Version: 3.2

>Number:         11258
>Category:       docs
>Synopsis:       suggestion to refine npx.4
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-doc
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          change-request
>Submitter-Id:   current-users
>Arrival-Date:   Wed Apr 21 05:20:01 PDT 1999
>Closed-Date:    Sat Feb 5 04:00:20 PST 2000
>Last-Modified:  Sat Feb  5 04:00:50 PST 2000
>Originator:     Motomichi Matsuzaki
>Release:        FreeBSD 3.1-STABLE i386
>Organization:
>Environment:

	

>Description:

There is no description about "GPL_MATH_EMULATE".

In the 'BUGS' section, there is the description,
"Emulation of the transcendental functions instructions is incorrect."
But "MATH_EMULATE" has not implemented any transcendental functions,
and "GPL_MATH_EMULATE" has. This is a confused description, isn't it?
In fact, the transcendental funcions under "GPL_MATH_EMULATE" has
precision weakness.

I suggest to refine these descriptions.

Some should be written clearly:
 
* There are two options, "MATH_EMULATE" the original implementation,
  and "GPL_MATH_EMULATE" derived from Linux.

* "MATH_EMULATE" has no transcendental functions.

* "GPL_MATH_EMULATE" makes complete emulations, but
  the transcendental functions have precision weakness.

* "GPL_MATH_EMULATE" is GPL-licensed so that the GENERIC kernel
  uses "MATH_EMULATE".

>How-To-Repeat:

	

>Fix:
	
	


>Release-Note:
>Audit-Trail:

From: John and Jennifer Reynolds <jjreynold@home.com>
To: freebsd-gnats-submit@FreeBSD.org
Cc: mzaki@e-mail.ne.jp
Subject: Re: docs/11258: suggestion to refine npx.4
Date: Fri, 14 Jan 2000 00:00:26 -0700 (MST)

 I dug into this a little and Motomichi is quite right in his description of
 what MATH_EMULATE will not do (no transcendental functions are emulated).
 
 I cannot speak for the precision problems that the GPL'ed emulator from 
 Linux might suffer from, but the following patch (against 3.4-STABLE) to
 npx.4 takes Motomichi's basic comments and includes them. I believe this will
 suffice for one of you wonderful doc committers to close this PR :)
 
 --- /usr/src/share/man/man4/man4.i386/npx.4     Wed Sep  1 20:55:52 1999
 +++ npx.4       Thu Jan 13 23:52:10 2000
 @@ -38,6 +38,8 @@
  .Nd Numeric Processing Extension coprocessor and emulator
  .Sh SYNOPSIS
  .Cd "options MATH_EMULATE"
 +or
 +.Cd "options GPL_MATH_EMULATE"
  .\" XXX this is awful hackery to get it to work right... -- cgd
  .Cd "device npx0 at isa? port IO_NPX tty irq 13"
  .Sh DESCRIPTION
 @@ -62,6 +64,16 @@
  It will provide support for the instructions normally executed
  by the NPX.  If there is no NPX in the system and the kernel
  is not built with math emulation, the system will not boot.
 +.Pp
 +You can choose to use a much more fully-featured math emulator (under the
 +GPL license) by including the option "GPL_MATH_EMULATE" in your kernel
 +configuration file
 +.Sy instead
 +of the "MATH_EMULATE" option.
 +.Pp
 +The
 +.Sy GENERIC
 +kernel configuration uses the "MATH_EMULATE" option.
  .Sh CAVEATS
  The emulator is much slower than the NPX coprocessor.
  This will result in poor floating-point math performance
 @@ -72,5 +84,5 @@
  the CPU wired properly.  If this is the case, the emulator must be used
  if consistent system operation is desired.
  .Pp
 -Emulation of the transcendental functions instructions is incorrect.
 -Emulation of other instructions is suspect.
 +When using the default "MATH_EMULATE" code, transcendental function
 +instructions are not implemented. Emulation of other instructions is suspect.
 
 
 -Jr
 
 -- 
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 John Reynolds         Chandler Capabilities Engineering, CDS, Intel Corporation
 jreynold@sedona.ch.intel.com  My opinions are mine, not Intel's. Running
 jjreynold@home.com          FreeBSD 3.4-STABLE. FreeBSD: The Power to Serve.
 http://members.home.com/jjreynold/  Come join us!!! @ http://www.FreeBSD.org/
 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 
State-Changed-From-To: open->closed 
State-Changed-By: asmodai 
State-Changed-When: Sat Feb 5 04:00:20 PST 2000 
State-Changed-Why:  
Modified fixes applied in rev 1.9 of npx.4. 

Thanks. 
>Unformatted:
