BIOS issue (at least, that's what I think)

My PC has a weird performance issue. A lot of people in this group
have been trying to help (thanks!). The help lead to locating the
problem, not solving it.

A BIOS ISSUE?
From a "cold" boot, the graphics performance of my system is approx.
40% lower than after any reboot (tested with 3DMark05, 3Dmark06 and
F.E.A.R.). Sisoft Sandra benchmarks show that the CPU and memory
speeds are normal.

Rebooting right after POST, will also result in a 40% performance
increase. So, it does not seem to have to do with Windows or drivers.
But something with the BIOS and/or its settings - I assume.

Windows does not report any hardware conflicts or faulty devices.
According to the SiSoft Sandra read-outs, there are no BIOS
mis-detections and CPU, memory, GPU and VRAM are running at their
designated MHz.

WHAT I'VE TRIED - AND DID NOT WORK
* Drivers, patches, about everything that was recommended to me in an
earlier thread, but all Windows related.
* Enabling or disabling "cool & quiet": there is no difference.
* Changing the AGP apperture size: no difference.
* Setting the busspeed manually to 200 MHz in stead of the AUTO
setting: no difference.

MY SYSTEM
* Asus A8V, VIA K8T800Pro chipset, BIOS version 0219 (latest)
* AMD Athlon64 X2 4200+
* 2 x 1GB dual channel Corsair Value RAM
* AOpen 6800LE AGP8x
* 150 GB WD Raptor & 200 GB WD Caviar
* everything at stock speed

PLEASE HELP
Is this some setting in BIOS? Faulty hardware? What can I do or try?

Thanks in advance,
Arend

Arend van Beek wrote:
My PC has a weird performance issue. A lot of people in this group
have been trying to help (thanks!). The help lead to locating the
problem, not solving it.

A BIOS ISSUE?
From a "cold" boot, the graphics performance of my system is approx.
40% lower than after any reboot (tested with 3DMark05, 3Dmark06 and
F.E.A.R.). Sisoft Sandra benchmarks show that the CPU and memory
speeds are normal.

Rebooting right after POST, will also result in a 40% performance
increase. So, it does not seem to have to do with Windows or drivers.
But something with the BIOS and/or its settings - I assume.



PLEASE HELP
Is this some setting in BIOS? Faulty hardware? What can I do or try?

Thanks in advance,
Arend

I would think that it's a hardware problem than a BIOS issue. Have you
tried the card in a different machine, what about the power supply, are
you able to change it?

It's almost reminiscent of the trouble with the ATI Radeon 9800 pro's
and some motherboards. The motherboard didn't allow enough power to the
card during post and the capacitors on the card drew most of the power
during that period, so the card didn't fully initialize. People tried
all sorts of remedies, from hot plugging the molex connector, to timing
delay circuits.
What worked for me, until I pulled the card, was to power up the system,
wait a few minutes after post until I was sure that all caps were full,
shut the power off at the back and restart, worked every time.

This is almost what you're doing, although not as extreme, which makes
me wonder about availability of power at the initial post.

Good luck, it's very frustrating, I know.

Ray

On Fri, 05 May 2006 09:47:45 -0400, Ray
wrote:

It's almost reminiscent of the trouble with the ATI Radeon 9800 pro's
and some motherboards. The motherboard didn't allow enough power to the
card during post and the capacitors on the card drew most of the power
during that period, so the card didn't fully initialize.

Ray,

Thanks. How can I check whether this is the cause?

By the way: the card worked flawlessly on a Asus A7N-8X board before
with an AOpen 350W PSU. The current PSU is a Zalman 460W.

Arend van Beek wrote:
On Fri, 05 May 2006 09:47:45 -0400, Ray
wrote:

It's almost reminiscent of the trouble with the ATI Radeon 9800 pro's
and some motherboards. The motherboard didn't allow enough power to the
card during post and the capacitors on the card drew most of the power
during that period, so the card didn't fully initialize.

Ray,

Thanks. How can I check whether this is the cause?

By the way: the card worked flawlessly on a Asus A7N-8X board before
with an AOpen 350W PSU. The current PSU is a Zalman 460W.


Now that's a toughy, I found that when I contacted the component
manufactures, each one blamed the other. I was eventually able to RMA
the card back to ATI and they checked it over (no report back to me
though) but it was no better.
After a while the MB maker, Soyo, posted a compatibility chart on their
site and said that the Radeon series of cards weren't compatible with
this board.
I think I would try tech support, Asus and Aopen. Actually, just
occurred to me, try here

http://mental18.phpbbweb.com/index.php?mforum=mental18

A very good forum for Asus boards.

Ray

In article , Arend van Beek
wrote:

On Fri, 05 May 2006 09:47:45 -0400, Ray
wrote:

It's almost reminiscent of the trouble with the ATI Radeon 9800 pro's
and some motherboards. The motherboard didn't allow enough power to the
card during post and the capacitors on the card drew most of the power
during that period, so the card didn't fully initialize.

Ray,

Thanks. How can I check whether this is the cause?

By the way: the card worked flawlessly on a Asus A7N-8X board before
with an AOpen 350W PSU. The current PSU is a Zalman 460W.

I doubt that is the problem. Only some models of cards in
the 9700 series had that problem. The result was a failure to
POST. With some other cards, it was Shuttle computer owners
who had problems, and a Shuttle doesn't have a very big supply
to begin with.

I would stick with the BIOS theory for now.

And by all means, file a problem report with Asus. Their tech
support get to see it all, and know stuff that we don't. For
example, when one guy had a channel fail to work on his audio
on a particular motherboard, the tech knew right away that he
had put a standoff underneath the board where there was no hole.
So I would give Asus tech support a try. Phoning tech support
may be more useful than email, as with email, you don't know
if your problem will be elevated to a real tech or not.

In each case where the performance is 2400 or 3700, have you used
CPUZ to review all the clock speeds and other parameters ?

You are running BIOS 0219. Have you tried the beta 0221 BIOS ?

If you were running an ATI video card, they set the AGP
transfer rate when the Windows desktop appears. The SmartGART
software tests for interface speed, and if AGP 8X is not stable,
the driver switches to 4X speed. As far as I know, Nvidia
(your 6800LE) don't do that, so the BIOS setting of 8X should
be a constant. You can try setting the AGP transfer rate in the
BIOS manually to AGP 4X, and see if the performance level is a
constant that way.

At AGP 8X, you could also enable the DBI option (dynamic bus
inversion). That feature is mandatory for AGP 3.0 spec, so if
the card is definitely an AGP8X card, you can enable that feature
in the BIOS. The following text is from the AGP 3.0 standard:

"2.1.5 Dynamic Bus Inversion

In order to mitigate the effects of simultaneous switching
outputs, AGP3.0 adopts a scheme called Dynamic Bus Inversion
(DBI) to limit the maximum number of simultaneous transitions
on source synchronous data transfers. DBI impacts only AD[31:0]
and is used during source synchronous and common clock transfers."

Now, all that setting should do, is improve transfer stability to
the card. It should not affect performance in any way. However, if
the Nvidia driver has some way of testing the AGP interface, like
the ATI driver does, then perhaps improving stability by turning
on DBI, will help.

At a time like this, it is important to use as many utilities as
you can find, to review clocks, timings, and AGP settings. It could
be that one of those settings is changing between your tests.

Paul

Paul,

Thank you. I'm trying your suggestions now.

And by all means, file a problem report with Asus.

Done. As soon as they reply to me, I will post their answer.

In each case where the performance is 2400 or 3700, have you used
CPUZ to review all the clock speeds and other parameters ?

Yes, in both cases the CPU runs at 2200 MHz, and memory runs at 200
MHz.

In my post I wrote: "[...] there are no BIOS mis-detections and CPU,
memory, GPU and VRAM are running at their designated MHz [...]". But
my measurement for the speed for the GPU is based solely on the
endresults of 3DMark.

Is there a tool like CPU-Z that checks GPU and VRAM speed
continuously?

Thanks,
Arend

Arend, you could try RivaTuner, I believe that has a monitoring
function with graphing.

Keep us posted as to how you make out, I know I'm interested, it
sounded so much like the problem I had with the Radeon. But go with
what Paul says, 'cause he is the man who knows these things.

Ray

On 6 May 2006 09:29:32 -0700, "Ray" wrote:

Arend, you could try RivaTuner, I believe that has a monitoring
function with graphing.

Keep us posted as to how you make out, I know I'm interested, it
sounded so much like the problem I had with the Radeon. But go with
what Paul says, 'cause he is the man who knows these things.

Ray


Paul, Ray, everyone else,

Thanks. Let me give you an update - no solution yet by the way.

GPU AND VRAM SPEED
RivaTuner has indeed a hardware monitor feature to check the GPU and
graphics card memory during operation (e.g. during benchamrking). I
used it and the speeds are stable and run at their designated MHz.

BIOS AGP SETTINGS
Besides the DBI setting, I've tried multiple other settings. Between
brackets the different settings. None of the settings made a change.
* Primary graphics adapter [pci][agp]
* Search for MDA resources [yes][no]
* VLink 8x supported [enabled][disabled]
* AGP mode [4x][8x]
* AGP 3.0 calibration cycle [enabled][disabled]
* DBI output for AGP trans [enabled][disabled]
* AGP aperture [32][64][128][256]

BIOS UPDATE
I was running 0219 (latest final version) and I have replaced it with
0221 (latest beta version). No difference, the problem's still there.

CONTACT ASUS
I mailed Asus with the same problem description I used in my first
post of this thread. This is their response:

"Thank you for contacting ASUS Customer Service. We have not received
this problem before. When you installed VIA 4in1 Driver,try not to
install the AGP driver,or update to the latest BIOS from ASUS website,
test again."

I have already replied, explaining the driver-thing will not do. I'm
hoping they get back to me with alternative solutions but I'm
sceptical.

TILL HELL FREEZES OVER
Kill me now please... This is consuming sooooo much time. Why does
PC-**** never simply WORK?

Anyone any other suggestion?

Arend