Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

fastwrites and sideband addressing

65 views
Skip to first unread message

RJ

unread,
Jun 6, 2002, 11:13:47 AM6/6/02
to
How can I enable AGP Fastwrites and Sideband Addressing on my GeForce 3 Ti
200? Powerstrip and other diagnostics show them as "supported but
disabled," but enabling them is grayed out. I enabled Fastwrites in my
BIOS, but it has done no good. I have Windows XP, Detonator 27.42 WHQL, and
Athlon 800 KT133 system. Any help is appreciated. If more info is needed I
will post.


Mart

unread,
Jun 6, 2002, 2:06:12 PM6/6/02
to
The general consensus is that fastwrites will make little
difference to your performance and sidebanding will make
your system instable.

--
Mart
" All those who think they know the
3rd letter in the vowel chain say 'I' "


"RJ" <rj1...@optonline.net> wrote in message
news:L_KL8.41579$tK.94...@news02.optonline.net...

me

unread,
Jun 6, 2002, 3:44:21 PM6/6/02
to
First, get a program that detects sideband and fastwrite support status
properly.
Good programs are WCPUID at http://www.h-oda.com/ or Sisoft Sandra. WCPUID
is free. Sisoft Sandra is Shareware
but it needs not special registration in order to detect your AGP bus status
along with what features of your
AGP bus are supported like fastwrites and sidebanding. I use WCPUID because
it's free, it works, and it's small
and nonintrusive instead of thinks it owns your system and installs tons of
files.
Then, you need a special program which will either unmask the sideband write
mask on the Nvidia chip or
a program which will edit and save your Video BIOS. Through the Nvidia
specific video BIOS editing program you
tell the Video BIOS to always disable the Sideband mask, that way it lets
requests to enable Sidebanding pass through
to the sidebanding enable register. The reason the Sideband Mask is there is
that if you enable sidebanding and your
video card shows no video, then you have no way to see in order to turn
around and disable it again. This was the problem
when AMD released a certain series of so called intel compatible chipsets
which turned out to be AGP nightmares, which wouldn't
even work properly at AGP2x. This is at the same time Nvidia released the
first true AGP4x cards with true sidebanding and fastwrites
and Nvidia forced sidebanding off and forced it to where it would even need
a mask to be disabled to simply get sidebanding working, thus
thwarting old programs which used to enable sidebanding, by forcing a mask.
It's unfortunate because the messed up chipset by AMD really
screwed up Nvidia bad because they were a very popular chipset at the time,
and hundreds of thousands of people owned computers using the
chipset and every person who bought an nvidia card would end up with a
completely black screen because the Nvidia card wanted to use AGP
normally while the CPU chipset sucked so bad, it stated it did AGP4x and
sidebanding but was found out to not even be capable of doing AGP2x
less sidebanding and so on. The result is a massive money loss for nvidia,
and people turning on AMD for releasing such a horrible AGP untested
chip and worse, AMD implemented the AGP bus entirely wrong thus some people
with weak, smaller power supplies had their power supplies smoke up
on them and die because of the lack of filtering on the AGP bus causing very
fast but heavy duty cycled current changes on the power supply line
and worse, it wasn't even undershoot protected by series resistors like
normal control lines are, so there was undershoot and ring problems causing
a domino effect where components completely unrelated were getting
destroyed. Anyhow, Nvidias solution is to simply and completely disallow
sidebanding
period in their GeForce3 chips. There is a way around it though, and the key
is to disable that darned sidebanding mask register which will then allow
your program to enable sidebanding like the program is trying to do but the
chip refuses because the sidebanding mask register is rejecting your
attempts.
So, the only way to fix the problem is by altering your video BIOS with a
program called "NVIDIA BIOS EDIT". The only thing that has to be changed is
one simple radio click box, which says something like 'disable sideband
mask' or something like that on the second tab screen.

"RJ" <rj1...@optonline.net> wrote in message
news:L_KL8.41579$tK.94...@news02.optonline.net...

Shep©

unread,
Jun 6, 2002, 3:55:58 PM6/6/02
to
On Thu, 6 Jun 2002 19:06:12 +0100, Whilst thinking that the universe
can only be pragmatic "Mart" <nir...@seattle.invalid> wrote:

>The general consensus is that fastwrites will make little
>difference to your performance and sidebanding will make
>your system instable.

Ditto :D


me

unread,
Jun 6, 2002, 3:56:28 PM6/6/02
to
That used to be true.
Now the concensus is that fastwrites do make a difference for programs which
are aware of it, otherwise the program itself should disable it because
fastwrites can slow things down unless a program can make use of it. Also,
fastwrites is excellent for 2D games and excellent for old DOS games which
use
your CPU for calculating 3D rendering. The speedups are in the order of over
1000 times faster CPU to Video memory write speeds especially for block
transfers. This is also very useful for 3D programs, but it's really meant
for unaccelerated programs because with Fastwrites enabled for accelerated
programs, the program will use fastwrites by default in which there is
faster, accelerated ways to do block transfers.
THIS is where fastwrites gets a bad name because indeed, Quake3 will use
fastwrites instead of block transfers and yes, you will lose frames because
a slower, CPU dependant process is being used instead of simply your video
accelerator being used. Come to think of it though, Nvidia really could
implement in their drivers a simple way to disable fastwrites on accelerated
programs only which could work faster with fastwrites disabled, thus
enabling the faster hardware block transfer modes. On the other hand, if
Nvidia will do this I'm not sure.
As far as Sidebanding, sidebanding will not cause instability unless your
chipset doesn't fully support the proper, full AGP implementation. In that
case,
it's the fault of your chipset, not your video card. I've not had one
person, who used a recent intel chipset ever have stability problems with
sidebanding. Also, most people don't even know what sidebanding is. In
reality, sidebanding is simply the demultiplexing of address and data busses
into their own lines across AGP plus a separate sideband set of lines which
sends specific video rendering commands to/from the video card. These
commands are not standard but you can see that it is much faster to be able
to send commands at the same time as you send the data and video memory
address needed by a command. You can get a great speed increase that way, if
the command does a massive feat like copies a massive chunk of memory, or
shades or antialiases an entire screen. Sidebanding should ALWAYS be enabled
on Nvidia video cards because Nvidia can and DOES implement it greatly when
you allow it. It might not make an earth shattering difference in some of
your games but I can tell you know that it can with some of your other games
and especially games and other programs in the future which makes much more
use of the massive Nvidia features you have just waiting for the dumb game
programmers to actually use instead of the programmers being dull, boring
and ignoring all these neat new features Nvidia provides and is dying to be
used, which present very little performance hit if any at all. Anyhow,
enough blathering, bye!


"Mart" <nir...@seattle.invalid> wrote in message
news:ado8be$k31$1...@news8.svr.pol.co.uk...

mhicaoidh

unread,
Jun 6, 2002, 11:20:14 PM6/6/02
to

I can confirm that very little performance gain is evident, but I've
never had stability issues with either on a GF2 GTS or GF4 Ti4400....


Mark

unread,
Jun 7, 2002, 2:24:08 AM6/7/02
to
"mhicaoidh" <mhic_...@NOSPAMhotmail.com> wrote in message
news:BCVL8.129903$ux5.1...@rwcrnsc51.ops.asp.att.net...
I can confirm that fastwrites do make a big difference in some games. MOHAA
was really slow and choppy on my Asus V8200 GeForce 3 and I discovered that
like the poster, fastwrites were disabled on the Nvidia chipset. I enabled
it with a bios utility and what an impact it had. Other games such as Quake
3, UT, SOF 1&2 didn't show any major difference. I can't remember the scores
but 3DMark also showed a slight improvement by a couple of hundred points. I
also tried sideband addressing and found that windows froze as soon as I
booted.

HTH, Regards, Mark


Shep©

unread,
Jun 7, 2002, 2:55:10 PM6/7/02
to
On Fri, 7 Jun 2002 06:24:08 +0000 (UTC), Whilst thinking that the
universe can only be pragmatic "Mark" <mr_cr...@btinternet.com>
wrote:

FW Enabled/Disabled made no difference to MOHAA on my Win98 system :)

Tripper

unread,
Jun 8, 2002, 3:52:30 AM6/8/02
to
"me" <m...@here.org> wrote in message news:<pYOL8.8816$2n4....@nwrddc04.gnilink.net>...

> So, the only way to fix the problem is by altering your video BIOS with a
> program called "NVIDIA BIOS EDIT". The only thing that has to be changed is
> one simple radio click box, which says something like 'disable sideband
> mask' or something like that on the second tab screen.

Worked for me with NVmax v.4.01.00, just clicked on the box. Now
Sandra, Riva Tuner and gFTweaker all show FW/SA enabled. Amazing.



> "RJ" <rj1...@optonline.net> wrote in message
> news:L_KL8.41579$tK.94...@news02.optonline.net...
> > How can I enable AGP Fastwrites and Sideband Addressing on my GeForce 3 Ti
> > 200? Powerstrip and other diagnostics show them as "supported but

> > disabled," but enabling them is grayed out...

OLIRC

unread,
Jun 8, 2002, 3:14:57 PM6/8/02
to
"RJ" <rj1...@optonline.net> wrote in message news:<L_KL8.41579$tK.94...@news02.optonline.net>...

Get Nvidia BIOS editor from here http://www.x-bios.3dgames.ru/ and
make the changes you want in the cards BIOS.

Download George Breeses "PCI Latency" patch for VIA chipsets here
http://www.networking.tzo.com/net/software/ and install it to get
better speed from your PCI-Cards.

OLIRC

David J. Boniuk

unread,
Jun 8, 2002, 4:15:01 PM6/8/02
to

"OLIRC" <ol...@my-deja.com> wrote in message
news:d54018a9.0206...@posting.google.com...

Just a note about the PCI Latency patch. I have found in my personal
experiments that while this will increase your
hard drive throughput, it also increases "hesitation" (small jerkiness) in
3D apps. Some folks have reported less problems
with Soundblaster Live! cards using this patch, but I don't have those
problems. I eventually decided to uninstall the patch
and loose a bit of harddrive performance to get back 3D performance.
YMMV as always.

DJB

0 new messages