Odd problem with K7S5A BIOS/POST

A mate gave me his PC to fix - it had intermittent blue screens and file
corruption in Windows. Now sorted out, but while working with it, I
noticed something odd.

The box has a K7S5A motherboard and Duron 1300 processor. The
intermittent nature of the file corruption meant I spent quite some time
swapping out components to diagnose the problem. One of those was the
video card.

I found that simply swapping the video card affected the speed of the
POST, particularly the memory count. Putting in a GeForce2 Ti causes
the memory check to speed up so much that the machine "loses" the hard
disk, i.e. it reaches the IDE bus scan before the hard disk has spun up
and initialised, so the drive is not seen and the machine fails to boot
the OS. In fact, with the GF2Ti in, the POST is incredibly quick -
about 3 seconds from power on to showing the Win2000 "Starting Windows"
screen.

Putting in a GeForce2 MX400 slows down the POST memory check to about 10
seconds and allows the hard disk to initialise and be found each and
every time.

I tried SDRAM and DDR RAM (one 256Mb stick each time) with no difference
in behaviour. The BIOS has an option, "Quick Boot", which is set to
disabled. The BIOS settings are not touched between video card changes.
Both video cards are AGP.

In each case, the POST counts the full 256MB memory. It's just a lot
faster with the GF2Ti in the system. (Actually, the machine as a whole
is a lot quicker with the Ti in.)

My question: why does changing the video card affect the POST memory
check speed?

--
(\__/)
(='.'=) This is Bunny. Copy and paste Bunny into your
(")_(") signature to help him gain world domination.

"Mike Tomlinson" <> wrote in message
news:...
>
> A mate gave me his PC to fix - it had intermittent blue screens and file
> corruption in Windows. Now sorted out, but while working with it, I
> noticed something odd.
>
> The box has a K7S5A motherboard and Duron 1300 processor. The
> intermittent nature of the file corruption meant I spent quite some time
> swapping out components to diagnose the problem. One of those was the
> video card.
>
> I found that simply swapping the video card affected the speed of the
> POST, particularly the memory count. Putting in a GeForce2 Ti causes
> the memory check to speed up so much that the machine "loses" the hard
> disk, i.e. it reaches the IDE bus scan before the hard disk has spun up
> and initialised, so the drive is not seen and the machine fails to boot
> the OS. In fact, with the GF2Ti in, the POST is incredibly quick -
> about 3 seconds from power on to showing the Win2000 "Starting Windows"
> screen.
>
> Putting in a GeForce2 MX400 slows down the POST memory check to about 10
> seconds and allows the hard disk to initialise and be found each and
> every time.
>
> I tried SDRAM and DDR RAM (one 256Mb stick each time) with no difference
> in behaviour. The BIOS has an option, "Quick Boot", which is set to
> disabled. The BIOS settings are not touched between video card changes.
> Both video cards are AGP.
>
> In each case, the POST counts the full 256MB memory. It's just a lot
> faster with the GF2Ti in the system. (Actually, the machine as a whole
> is a lot quicker with the Ti in.)
>
> My question: why does changing the video card affect the POST memory
> check speed?
>
Simple answer is, its a K7S5A, the biggest POS mobo made in my opinion, so
anything is possible. Personally I would dump it and replace it with
ANYTHING else, before he loses any more data.

SteveH

"Mike Tomlinson" <> wrote in message
news:
> A mate gave me his PC to fix - it had intermittent blue screens and
> file corruption in Windows. Now sorted out, but while working with
> it, I noticed something odd.
>
> The box has a K7S5A motherboard and Duron 1300 processor. The
> intermittent nature of the file corruption meant I spent quite some
> time swapping out components to diagnose the problem. One of those
> was the video card.
>
> I found that simply swapping the video card affected the speed of the
> POST, particularly the memory count. Putting in a GeForce2 Ti causes
> the memory check to speed up so much that the machine "loses" the hard
> disk, i.e. it reaches the IDE bus scan before the hard disk has spun
> up and initialised, so the drive is not seen and the machine fails to
> boot the OS. In fact, with the GF2Ti in, the POST is incredibly
> quick - about 3 seconds from power on to showing the Win2000
> "Starting Windows" screen.

I don't understand what you are saying here - if the disk isn't recognised
then where is the "Starting Windows" screen coming from?


> Putting in a GeForce2 MX400 slows down the POST memory check to about
> 10 seconds and allows the hard disk to initialise and be found each
> and every time.
>
> I tried SDRAM and DDR RAM (one 256Mb stick each time) with no
> difference in behaviour. The BIOS has an option, "Quick Boot", which
> is set to disabled. The BIOS settings are not touched between video
> card changes. Both video cards are AGP.
>
> In each case, the POST counts the full 256MB memory. It's just a lot
> faster with the GF2Ti in the system. (Actually, the machine as a
> whole is a lot quicker with the Ti in.)
>
> My question: why does changing the video card affect the POST memory
> check speed?


Differences between cards due to onboard vga bios and memory differences.
The MX400 is communicating more slowly leading to a general slowdown of the
system. Whilst the system boots OK it is actually running slower than it
should. Your difficulties with post/booting with the GF2 Ti may be
unrelated to the speed of the post - it is more likely that the board is
sometimes having difficulty properly initialising with this card in place.
It could be that the PSU is marginal and the GF2 Ti is drawing a bit more
power, causing problems at post because this is when power draw peaks.
However, the K7S5A was well known for being picky with graphics cards. Some
boards just wouldn't post at all with some graphics cards, some would
post/boot intermittently, others would hang during intial post but work fine
if the reset button was pressed and others would hang every time whilst
posting. A board that would hang with one card would often work fine with
another, leading to much confusion. When working well these were excellent
boards but when behaving erratically they could be a bastard to sort out.
There were a variety of mods available to improve performance and stability,
including soldering resistors to the board and flashing with modified bios
files.

--
Tony
'04 XL883C, '95 LS650
OMF#24

"SteveH" <> wrote in message
news:vT7ih.19143$. uk
> "Mike Tomlinson" <> wrote in message
> news:...
>>
>> A mate gave me his PC to fix - it had intermittent blue screens and
>> file corruption in Windows. Now sorted out, but while working with
>> it, I noticed something odd.
>>
>> The box has a K7S5A motherboard and Duron 1300 processor. The
>> intermittent nature of the file corruption meant I spent quite some
>> time swapping out components to diagnose the problem. One of those
>> was the video card.
>>
>> I found that simply swapping the video card affected the speed of the
>> POST, particularly the memory count. Putting in a GeForce2 Ti causes
>> the memory check to speed up so much that the machine "loses" the
>> hard disk, i.e. it reaches the IDE bus scan before the hard disk has
>> spun up and initialised, so the drive is not seen and the machine
>> fails to boot the OS. In fact, with the GF2Ti in, the POST is
>> incredibly quick - about 3 seconds from power on to showing the
>> Win2000 "Starting Windows" screen.
>>
>> Putting in a GeForce2 MX400 slows down the POST memory check to
>> about 10 seconds and allows the hard disk to initialise and be found
>> each and every time.
>>
>> I tried SDRAM and DDR RAM (one 256Mb stick each time) with no
>> difference in behaviour. The BIOS has an option, "Quick Boot",
>> which is set to disabled. The BIOS settings are not touched between
>> video card changes. Both video cards are AGP.
>>
>> In each case, the POST counts the full 256MB memory. It's just a lot
>> faster with the GF2Ti in the system. (Actually, the machine as a
>> whole is a lot quicker with the Ti in.)
>>
>> My question: why does changing the video card affect the POST memory
>> check speed?
>>
> Simple answer is, its a K7S5A, the biggest POS mobo made in my
> opinion, so anything is possible.

Except when it worked well, in which case it was the "best bang for the
buck" in its day. Personally I ran one for two years, overclocked to 150MHz
fsb without any problems whatsoever and there were many other satisfied
users. However, some boards were a bit flaky. One problem with issues like
this is that, say, ten thousand problem boards can seem like a major
problem - until it is considered as a percentage of the total number of
boards out there (probably a couple of million). People with problems are
the ones that tend to post messages about their motherboards so there are a
whole load of posts about K7S5A problems but they are not representative of
the entire user base.

--
Tony
'04 XL883C, '95 LS650
OMF#24

In article <>, TMack <REMOVETHECAPStony
> writes

>I don't understand what you are saying here - if the disk isn't recognised
>then where is the "Starting Windows" screen coming from?

Sorry, I could have been clearer there. If the machine is powered on
from cold with the Ti in, it POSTs so quickly that the disk isn't found
- it's not had time to spin up and initialise by the time POST does the
IDE bus scan. So the machine, after displaying the BIOS configuration
page, halts with "No system disk or disk error" (or words to that
effect).

If it's warm-booted, the disk is found and boots OK every time. This
wouldn't cause me a problem, but the user of the machine is somewhat
technophobic, and panics if anything out of the ordinary happens.

>Differences between cards due to onboard vga bios and memory differences.
>The MX400 is communicating more slowly leading to a general slowdown of the
>system. Whilst the system boots OK it is actually running slower than it
>should.

I hear what you're saying, but am unconvinced. At POST, the BIOS is
talking to the video card using the VGA interrupts. It's not even aware
at this point if the card is PCI or AGP. I don't think the marked
difference in the memory count speed can be attributed to the speed of
the different video cards.

> Your difficulties with post/booting with the GF2 Ti may be
>unrelated to the speed of the post - it is more likely that the board is
>sometimes having difficulty properly initialising with this card in place.

There's no screen artifacts/corruption or other intermittent
strangeness, which I would expect to see if there were problems talking
to the card.

>It could be that the PSU is marginal and the GF2 Ti is drawing a bit more
>power, causing problems at post because this is when power draw peaks.

I've tried a couple of power supplies, including a good 350W FSP unit,
with no change in behaviour.

>However, the K7S5A was well known for being picky with graphics cards. Some
>boards just wouldn't post at all with some graphics cards, some would
>post/boot intermittently, others would hang during intial post but work fine
>if the reset button was pressed and others would hang every time whilst
>posting.

Odd. The K7S5A we're talking about here used to be mine. I used it for a
couple of years with an XP1800+, then an XP2400+ before handing it down
to my friend. In that time I must have used it with a dozen different
video cards and never had a problem.

> A board that would hang with one card would often work fine with
>another, leading to much confusion. When working well these were excellent
>boards but when behaving erratically they could be a bastard to sort out.
>There were a variety of mods available to improve performance and stability,
>including soldering resistors to the board and flashing with modified bios
>files.

Yes, I used to keep an eye on the K7S5A web forum (was it based at
ochardware.com or something like that?) and tried a few of the modded
BIOSes. The HoneyX ones in particular were very good.

However, as the board never gave me any trouble (it was, and still is,
the best board I've ever used), I didn't need to perform any of the
hardware mods.

Thanks for the feedback Tony. It's much appreciated.

In article <vT7ih.19143$ >, SteveH
<> writes

>Simple answer is, its a K7S5A, the biggest POS mobo made in my opinion, so
>anything is possible.

It used to be mine. I used it for 2 years before passing it on and it's
the best board I've ever used. And I've used a LOT of boards.

> Personally I would dump it and replace it with
>ANYTHING else, before he loses any more data.

The data loss problem is fixed. I was just asking about the POST memory
count speed changing with different video cards.

Mike Tomlinson wrote:
>
> I've tried a couple of power supplies, including a good 350W FSP unit,
> with no change in behaviour.


My daughter's PC has this mobo and when we first got it it was a bit
flaky. Replaced the 300w PSU for a 400w PSU and all the troubles went
away. It's been solid ever since. It's got an ATi card of some sort.
I remember replacing the original 8500 for something better but I can't
remember what off the top of my head. It's never been over-clocked just
set to the max. Still has the manufacturer's BIOS, though it was the
latest they offered at the time. I changed the Win98SE to XP Home a
while back without any problems, though I did do a complete new format
and install.

Regards

Mark

"Mike Tomlinson" <> wrote in message
news:...
|
| A mate gave me his PC to fix - it had intermittent blue screens and
file
| corruption in Windows. Now sorted out, but while working with it, I
| noticed something odd.
|
| The box has a K7S5A motherboard and Duron 1300 processor. The
| intermittent nature of the file corruption meant I spent quite some
time
| swapping out components to diagnose the problem. One of those was
the
| video card.
|
| I found that simply swapping the video card affected the speed of
the
| POST, particularly the memory count. Putting in a GeForce2 Ti
causes
| the memory check to speed up so much that the machine "loses" the
hard
| disk, i.e. it reaches the IDE bus scan before the hard disk has spun
up
| and initialised, so the drive is not seen and the machine fails to
boot
| the OS. In fact, with the GF2Ti in, the POST is incredibly quick -
| about 3 seconds from power on to showing the Win2000 "Starting
Windows"
| screen.
|
| Putting in a GeForce2 MX400 slows down the POST memory check to
about 10
| seconds and allows the hard disk to initialise and be found each and
| every time.
|
| I tried SDRAM and DDR RAM (one 256Mb stick each time) with no
difference
| in behaviour. The BIOS has an option, "Quick Boot", which is set to
| disabled. The BIOS settings are not touched between video card
changes.
| Both video cards are AGP.
|
| In each case, the POST counts the full 256MB memory. It's just a
lot
| faster with the GF2Ti in the system. (Actually, the machine as a
whole
| is a lot quicker with the Ti in.)
|
| My question: why does changing the video card affect the POST memory
| check speed?


Vid card memory speeds can dramatically change when one upgrades a vid
card. I suspect the difference in video memory speed, plus a faster
GPU, can make a difference in post times. If you want to slow things
down more during post, disable caching of vid card memory, and disable
BIOS caching and BIOS shadowing. These items have little or no impact
on system speed once Windows is running since substantially all BIOS
functions are replaced by protected mode drivers (with the possible
exception of some power management functions controlled under ACPI)
once the OS is up and running.

Is the HD a Western Digital? I've had similar problems with a WD 80
gig IDE HD (I think its a 800JB model) and my k7s5a (I think I have
anyway, b/c this all sounds familiar). I still run a k7s5a system
with a GF TI200 vid card and have seen this type of problem on
occasion. Since I never turn my k7s5a system off (which helps avoid
the cold boot CMOS memory loss syndrome) I don't see a cold boot very
often. One tip on HD corruption, it is possible for dirt and crud on
the mobo to affect HD data transfers, use a brush to clean the mobo
once in a while, particularly brushing off the leads of the ICs.
Observe safe static grounding techniques while doing this. Cleaning
the mobo crud resolved a HD corruption problem for me, tho I'm sure
this is a rather odd failure scenario.
--
Best regards,
Kyle

Mike Tomlinson wrote:
>
> I found that simply swapping the video card affected the speed of the
> POST, particularly the memory count.

I strongly suspect that this is down to text display speed. It's
probably writing the memory check point to video memory every n KB
rather than every n milliseconds.

Odd that it's so much slower with the GF2-MX, but then you do say that
the machine is generally slower with that card. Might be an AGP
compatibility issue, or some kind of VGA BIOS glitch.


--
John Jordan