Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 3 May 2000 11:39:34 +0930
From:      Greg Lehey <grog@lemis.com>
To:        Chris Wasser <cwasser@v-wave.com>, Olaf Hoyer <Olaf.Hoyer@netsurf.de>
Cc:        questions@FreeBSD.ORG
Subject:   Re: Fwd: Major AMD K7 Problems under 4.0-S
Message-ID:  <20000503113934.A8585@freebie.lemis.com>
In-Reply-To: <20000409220240.A54143@area51.v-wave.com>
References:  <4.1.20000502145857.00a1f7e0@mail.rz.fh-wilhelmshaven.de> <20000502192207.A549@freebie.lemis.com> <4.1.20000502145857.00a1f7e0@mail.rz.fh-wilhelmshaven.de> <20000503084811.A1654@freebie.lemis.com> <4.1.20000503025803.009d06c0@mail.rz.fh-wilhelmshaven.de> <20000409220240.A54143@area51.v-wave.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday,  3 May 2000 at  3:11:41 +0200, Olaf Hoyer wrote:
> At 08:48 03.05.00 +0930, you wrote:
>> On Tuesday,  2 May 2000 at 15:01:26 +0200, Olaf Hoyer wrote:
>>> At 19:22 02.05.00 +0930, you wrote:
>>>> I've just got a new Athlon machine with an Epox EP-7KXA motherboard.
>>>> It comes highly recommended, but I have some severe problems with the
>>>> timekeeping: it seems that the BIOS does not allow me to disable APM,
>>>> and the 8254 timekeeping goes crazy, running up to 10 times as fast as
>>>> real time.  Does anybody else have one of these boards?  Have you seen
>>>> the problem?  (How) did you fix it?
>>>
>>> Have talked to some guys, which also sell this board, and they told
>>> me that they'd first flash the BIOS to the current one, dated 25/4,
>>> IIRC.  There were some issues in the past that were fixed with BIOS
>>> upgrades.
>>>
>>> They found it very stable, also running some benches a complete
>>> weekend to test stability (even under windows, it survived a whole
>>> weekend...)
>>
>> Yes, but have you tried it under FreeBSD?  I have the BIOS here, but I
>> can't flash it because it insists on flashing from floppy.  But
>> looking at the list of fixes, I don't see anything there about APM.
>> My BIOS is dated 17 April, so there's probably not much difference.
>
> Well, they are mostly M$ folks... (Mostly some small retailers in
> US) But Epox has a history of BIOS issues for years, and those guys
> told me that BIOS updates fix in most cases the troubles you have
> with the mainboards, since the mechanical quality is quite good.  So
> I'd flash the lastes BIOS from some bootable floppy (there should be
> some M$ machine being around somewhere). Not every error or glitch
> is documented...

No, but when it's something as blatant as a missing option in the BIOS
menus, you'd think they'd mention it.

> Or be the one who files a bug report to Epox  about  APM ;-))

Done yesterday:

> Customer email: grog@lemis.com
> Subject: Error in EP-7KXA BIOS of 17 April 2000
> Case # 5224636
> 
> I have just purchased an EP-7KXA motherboard with BIOS dated 17 April
> 2000.  It is not possible to disable power management.  According to
> the manual, the menu selection Power Management Setup/Power Management
> should give the four choices "Disabled", "Min saving", "Max saving"
> and "User defined".  In fact, the "Disabled" menu item is missing.
> 
> This board is running very unstably.  The time runs ahead up to ten
> times the real time, causing the system to hang repeatedly.
> 
> I cannot identify a serial number on the motherboard.  The number
> supplied is invented to keep your web software happy.

> Of course, there may be the slight statistical chance that you got a
> bad mainboard...

No, we've had other reports:

 On Sun, 9 Apr 2000 21:52:55 -0600, Chris Wasser <cwasser@v-wave.com> wrote:
> To: questions@FreeBSD.ORG
>
> I'm sure this has been hashed over and over countless times, but I
> can't seem to resolve this problem locally and figured I'd shoot it
> out to the mailing list here and see if anyone has any ideas. This
> machine is having critical problems preventing it from being put
> into production and I'd _really_ appriciate any
> information/insight/ideas people can shoot over to me to try.
>
> Please respond on questions/bugs@FreeBSD.ORG or via email to the above
> address (not subscribed to freebsd-hardware)
>
> Please note, I have an identical AMD K7-700 system running, the only
> difference is it has a ASUS K7M motherboard and 3c905B-TX nics and a
> Mach64-GB AGP video card. It does not exibit this behavior. It uses the
> AMD-751 and VIA 82C686 chipset.
>
> Here's the system:
>
> AMD K7-700 (0.18fab) running @ 700MHz (100x7.0)
> 256MB PC133 (Running @ 100MHz)
> 1 3Com 3c905C-TX connected @ 10Base-T to a cablemodem (xl0)
> 1 3Com 3c905C-TX connected @ 100Base-T full-duplex to LAN (xl1)
> EpoX 7KXA motherboard (VIA KX133 chipset/82C686)
> 2 x Quantum KX 20GB ATA-66 (running @ ATA-66)
> ASUS 50x ATA-33 IDE CDROM
> ATI Rage 128-RF AGP running @ AGP2x
> 300W power supply
>
> The system and kernel were cvsup'd and built on March 30th/2000 from
> stock 4.0-RELEASE .. xl0 was enabled but there was no active connection
> during this test. The only active card w/link was xl1 @ 100Base-T
> full-duplex (switched).
>
> The actual time and my date below for the dmesg output is inaccurate
> (forgot to set the system date & time in the bios, nothing major.), it
> was re-compiled today [not cvsup'd, just recompiled the kernel when I
> added new options, the source on the system is from March30th, the only
> change I made to the actual source tree was from PR17831 which I submitted
> a few days ago.] (April 9th)
>
> When doing file xfers on the local lan (xl1) the system will eventually
> produce error messages up microuptime() going backwards:
>
> microuptime() went backwards (3821.740589 -> 3821,720223)
> microuptime() went backwards (3821.774554 -> 3821,731215)
> ..etc etc
>
> I had to terminate syslogd because it was spending so much time logging
> that it was dragging down the system when the microuptime() messages
> started appearing (choppy keyboard response over ssh and locally at
> console and the load avg was a solid 2.01 across the board)
>
> This happens under heavy network load (transferring +7GB files over the
> local LAN) and processes start showing negative run times or completely
> outrageous run time:
>
> root   0  0.0  0.0     0    0  ??  DLs  10:19AM  11:36.96  (swapper)
> root   5  0.0  0.0     0    0  ??  DL   10:19AM 359:53.83  (syncer)
> root   2  0.0  0.0     0    0  ??  DL   10:19AM  11:48.00  (pagedaemon)
>
> I set sysctl to use accurate time measurements thinking that might provide
> a solution (after cruising through older mailing list archives and made
> no difference):
>
> kern.timecounter.method: 1
> kern.timecounter.hardware: i8254

This all corresponds exactly to what I saw with the same motherboard.

Anyway, the good news is that it doesn't happen under -CURRENT, and
since I was planning to run -CURRENT on it anyway, this is now a
non-issue.

Greg
--
When replying to this message, please copy the original recipients.
For more information, see http://www.lemis.com/questions.html
Finger grog@lemis.com for PGP public key
See complete headers for address and phone numbers


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000503113934.A8585>