From rb@gid.co.uk  Mon Apr 14 14:06:12 1997
Received: from isbalham.ist.co.uk ([192.31.26.1])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA14182
          for <FreeBSD-gnats-submit@freebsd.org>; Mon, 14 Apr 1997 14:04:53 -0700 (PDT)
Received: from gid.co.uk (uucp@localhost)
          by isbalham.ist.co.uk (8.8.4/8.8.4) with UUCP
	  id VAA20380 for freebsd.org!FreeBSD-gnats-submit; Mon, 14 Apr 1997 21:44:49 +0100 (BST)
Message-Id: <20851.199704142045@seagoon.gid.co.uk>
Date: Mon, 14 Apr 1997 21:45:58 +0100
From: Bob Bishop <rb@gid.co.uk>
Reply-To: rb@gid.co.uk
To: FreeBSD-gnats-submit@freebsd.org
Subject: Cyrix 486 performance problem
X-Send-Pr-Version: 3.2

>Number:         3292
>Category:       kern
>Synopsis:       Cyrix 486 performance problem
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-bugs
>State:          closed
>Quarter:
>Keywords:
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Mon Apr 14 14:10:01 PDT 1997
>Closed-Date:    Mon Apr 28 16:53:07 PDT 1997
>Last-Modified:  Mon Apr 28 16:53:35 PDT 1997
>Originator:     Bob Bishop
>Release:        FreeBSD 3.0-CURRENT i386
>Organization:
GID ltd
>Environment:

	Cyrix 486DX2/80, ISA M/B, 20Mb, Adaptec1542CF

>Description:

	With kernels built after around 20 March, CPU runs an
	order of magnitude slower than it should. Eg, kernel
	build takes 3.25 hours against 28 min; tiny program
	containing nested for loops takes 51 sec against 4
	sec.

	Old (ie good) kernels identify the CPU as:
		CPU: Cy486DLC (486-class CPU)
		   Origin = "Cyrix"
	Newer (broken) kernels:
		CPU: Cyrix 486DX2 (486-class CPU)
		   Origin = "CyrixInstead"  Device ID = 0x321b
		    Stepping=3  Revision=2

	I surmise that the CPU initialisation code got broken
	sometime around 20 March.

>How-To-Repeat:

	Run anything on this machine with a current-ish
	kernel.

>Fix:
	
	Work around by using an old kernel :-(

>Release-Note:
>Audit-Trail:

From: "Justin T. Gibbs" <gibbs@plutotech.com>
To: rb@gid.co.uk
Cc: FreeBSD-gnats-submit@freebsd.org
Subject: Re: kern/3292: Cyrix 486 performance problem 
Date: Mon, 14 Apr 1997 16:34:05 -0600

 >	With kernels built after around 20 March, CPU runs an
 >	order of magnitude slower than it should. Eg, kernel
 >	build takes 3.25 hours against 28 min; tiny program
 >	containing nested for loops takes 51 sec against 4
 >	sec.
 
 Have you played around with all of the new fangled CPU options?
 They're documented in lint.  My guess is that your cache is disabled.
 
 --
 Justin T. Gibbs
 ===========================================
   FreeBSD: Turning PCs into workstations
 ===========================================
 
 

From: Bob Bishop <rb@gid.co.uk>
To: "Justin T. Gibbs" <gibbs@plutotech.com>
Cc: FreeBSD-gnats-submit@freebsd.org
Subject: Re: kern/3292: Cyrix 486 performance problem
Date: Thu, 17 Apr 1997 23:19:06 +0100

 At 11:34 pm +0100 14/4/97, Justin T. Gibbs wrote:
 >Have you played around with all of the new fangled CPU options?
 >They're documented in lint.  My guess is that your cache is disabled.
 
 You're right, this CPU needs options CYRIX_CACHE_WORKS. I've never heard of
 cache problems with Cyrix 486DX; except perhaps when using them on old
 Cyrix-unaware motherboards, and then they often won't run period. I have a
 machine around with a (similar IIRC) TI486DX2/80, guess I'd better check
 that too.
 
 On a 486DX2/80 with the cache disabled you'd need over 3 hours to build a
 kernel to fix it. On the evidence so far I'd say the cache should be
 enabled by default for Cx486; I appreciate there are other gotchas with
 Cx586 and Cx686.
 
 
 --
 Bob Bishop              (0118) 977 4017  international code +44 118
 rb@gid.co.uk        fax (0118) 989 4254  between 0800 and 1800 UK
 
 
State-Changed-From-To: open->closed 
State-Changed-By: gpalmer 
State-Changed-When: Mon Apr 28 16:53:07 PDT 1997 
State-Changed-Why:  
Submitter reports a recent change by KATO Takenori fixed the problem. 
>Unformatted:
