Sign in with
Sign up | Sign in
Your question

A7N8X - Gainward 6800GT sound and video distortion

Last response: in Motherboards
Share
Anonymous
a b V Motherboard
September 9, 2005 2:26:11 PM

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

Hello,

I have recently put a Gainward 6800GT card in an A7N8X v.106
motherboard. I am now noticing crackles on the audio, both with the
onboard sound and a Soundblaster lie card I swopped in. In addition,
the video peridicaaly breaks up in places on the screen. With a
digital TV unit, while the audio starts fine in play mode, switching
to record produces the above errors plus the picture "lags" every few
seconds (like a brake was applied).

Is this likely to be a result of this cards interaction with this
motherboard? I am going to swop in another make to test it but I am
rather peeved to say the least if this proves to be the case.

TIA
September 9, 2005 8:08:30 PM

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

In article <ruk2i1hmn8rrhp6keamje2sqd5uluno0vo@4ax.com>, No-one
<No-one@nowhere.notime> wrote:

> Hello,
>
> I have recently put a Gainward 6800GT card in an A7N8X v.106
> motherboard. I am now noticing crackles on the audio, both with the
> onboard sound and a Soundblaster lie card I swopped in. In addition,
> the video peridicaaly breaks up in places on the screen. With a
> digital TV unit, while the audio starts fine in play mode, switching
> to record produces the above errors plus the picture "lags" every few
> seconds (like a brake was applied).
>
> Is this likely to be a result of this cards interaction with this
> motherboard? I am going to swop in another make to test it but I am
> rather peeved to say the least if this proves to be the case.
>
> TIA

I've seen a couple of suggestions for fixing stuff like this.
One was to disable APIC on Nforce2. Another is to play with the
PCI Latency settings on the various bus masters. The PCI Latency
setting can be programmed with Powerstrip from entechtaiwan.com,
or this tool "PCI Latency 2.x" was suggested.

http://downloads.guru3d.com/download.php?det=951

There is another setting which would help on other boards, and that
is "Delayed Transaction" [Enabled]. That allows a transaction to
a slow-to-respond device to be temporarily "parked", while other
fast transactions continue to occur. On some chipsets, accesses
to devices in the Southbridge, tie up the PCI bus until the answer
comes back. Enabling "Delayed Transaction" allows the bus to
be released, and the Southbridge operation retried after a short
interval. The end result is better bus efficiency, sometimes
enough to stop audio crackling.

But, on the Nforce2, the feature "Concurrent PCI" implies that
Delayed Transaction is always enabled. So there is no corresponding
BIOS setting. You should already be receiving whatever benefits
that optimization can provide.

From manual e1170 for A7N8X:

"Concurrent PCI
Concurrent PCI maximizes system performance with simultaneous
CPU, PCI and ISA bus activities. It includes multi-transaction
timing, enhanced write performance, a passive release mechanism
and support for PCI 2.1 compliant delayed transactions. Concurrent
PCI provides increased bandwidth, reduced system latencies,
improves video and audio performance, and improves processing
of host based applications."

When I look through some Intel datasheets, I see many pieces of
hardware have Latency registers. Some of the registers are fake
(hardwired to 0x00, function only as scratchpad registers etc),
while others are real. Generally what I find, is an eight bit
register, where the upper five bits are writable. The max value
is 0xF8 and the min value is 0x00. 0xF8 corresponds to 248 cycles.

The bus latency value, is the maximum time a bus master can
occupy the bus, until releasing the bus to another master.
Now, architecturally, not all hardware devices in your computer
sit on the exact same bus, but it is possible for a "bus hog",
no matter where it is located, to have an impact on other
devices. If a device somewhere grabs the bus, and does a 248
cycle long data burst, that is time that other cards will not
be able to get/put any data they've got.

Video cards are generally set quite high. Expect to find 248
for them. Dropping the video card allocation can sometimes
make the machine smoother, at the expense of perhaps dropping
the benchmark for the videocard (i.e. adjust max latency,
rerun 3DMark and see what you are losing).

The BIOS has a PCI Latency Timer setting, and it will generally be
set to 32. I don't know whether the BIOS writes this value
into all the cards, or the PCI Latency setting is being
applied to a single register somewhere in the chipset.

http://www.rojakpot.com/showFreeBOG.aspx?Lang=0&bogno=1...

The deal with any max latency register is, the higher one of
these is set, the greater the percentage of bandwidth the
card can grab. But, a card which has a real time requirement,
like a sound card, can be starved waiting for data, if some
other card(s) are too greedy. Rather than custom tune every
system, that is why settings like 32 are generally used, as
a compromise between monster benchmarks for individual hardware
devices, versus smooth overall computer operation. You may
find tweaking is nothing but a PITA, and not a permanent
fix.

Also, you didn't mention the brand/model of your "digital
TV" unit, but be aware that when these record, they
create double the bus traffic. The data is being recorded, but
the data is also decompressed and sent to the frame buffer
at the same time. In some cases, the video card is even
being used to do IDCT (inverse discrete cosine transform)
as part of the decompression process. Both the processor
in the computer, and the system busses, can be pretty busy
when this happens. It might help a bit, to drop the size
of the monitoring window down a notch, to reduce the amount
of decompression generated traffic.

So, have a look at your PCI Latency register settings, and
see if cards other than the video card, have been given
a very high setting.

HTH,
Paul
Anonymous
a b V Motherboard
September 9, 2005 9:33:54 PM

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

On Fri, 09 Sep 2005 16:08:30 GMT, nospam@needed.com (Paul) wrote:

>I've seen a couple of suggestions for fixing stuff like this.

[snip]

Thanks for the suggestions. I am also experiencing the same problem
with a BFGTech 6800GT I swopped in. (Changed drivers as well)

What gets me is that everything was working perfectly until a couple
of days back.

Best regards.
Anonymous
a b V Motherboard
September 11, 2005 1:43:10 AM

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

On Fri, 09 Sep 2005 17:33:54 +0100, No-one <No-one@nowhere.notime>
wrote:

I nuked the partition, reinstalled and this appears to have cleared a
number of problems.

Regards.
!