🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Thread View: alt.games.quake3
6 messages
6 total messages Started by noise Fri, 10 Dec 1999 00:00
Wanted: Voodoo3 performance tests
#226
Author: noise
Date: Fri, 10 Dec 1999 00:00
153 lines
7575 bytes
Hi,

I'm going to publish a website full of tips and tweaks designed to help
Voodoo 3 owners get the most out of their cards, especially for Quake 3.
Since I've had some success with improving the display quality in Q3, I
had decided to make a Voodoo 3 page, but there's a lot more I could put
into it, eg. charts of the speed difference all kinds of settings and
tweaks make (in % points). There's some tweaks that improve visual display
quality at the expense of speed and vice versa, and there's others that
make improvements without costing you elsewhere.

I'd like some other Voodoo 3 users to run similar tests, and I'll put
your results in my website and give you recognition for it (I can't
give you any money for it though! Volunteers only please!). As I've
got a K6-III system, I need owners of other systems to do the tests
too, to help discover what differences in hardware do to affect the
way Quake 3 runs with your Voodoo 3. Maybe we'll discover that some
feature of a particular CPU is either well suited or bad for Quake 3,
or that some motherboards hold you back, and so on.
But most of all, it's to help Voodoo3 users to make the very most of
their cards... they may seem to be behind the pack for Quake 3 compared
to modern cards, but optimising things might just make the difference
you're looking for.

People set up their PC's differently. Yesterday, someone replied to a
post of mine and told me that setting "Force Mipmap Dithering" cost
him almost 10 FPS on his PIII 600mhz system with Voodoo3. But on my
PC, this only cost me about a frame or two. The main difference between
his and my set-up was that he uses WickedGL and I don't, as it's not
well suited to the K6-2 or 3.
So variations like these cause all kinds of differences depending on
what's different about the rest of your system. The way to find out
how what and how much, is to try changing one setting, and then run
a series of speed tests.

As an example, yesterday I posted "Improving Voodoo3 Display Quality for
Quake 3", which concentrated on Windows settings. It was about visual
quality, not speed, but speed tweaks will be there too. Anyway, I ran
timedemo tests of a standard configuration of Windows Display settings,
and Quake 3 settings too, and then I ran a series of tests for all the
important resolutions and recorded the FPS (Frames Per Second) results.
Then, after changing just one setting, I'd repeat the tests. For each
test, I'd return the settings to their original, default state, and then
change only the setting I wanted to test.
The goal here was to speed-test all of the tweaks that might improve
the way the display looks. Those settings that don't have too much of
a speed hit, I use, to make the scenes look better. It lets me know how
I can best balance speed with display quality and features.

OK, well down to business. At the end of this post is a section you
can copy & paste into a new post, and fill in your details and results.
Then, either email them to me or post them here, or even better,
address the form to both this group and my email address.

Yesterday, that person who saw a 10 FPS drop in frame rates when he
switched on "Force Mipmap Dithering" prompted me to test for myself.
I set up a standard (or, control) configuration, and ran a series of
timedemo runs. Then, I went to Windows and altered JUST that one
setting, and ran the tests again. Here's my results, not so much as
results but an example of what I'm hoping you can do for me:

Basic Configuration (Configuration A):

WINDOWS SETTINGS: (Control Panel/Display/Settings/Advanced/3dfxTweaks)
Disable Splash Screen: ----------> On
Force Mipmap Dithering: ---------> Off
Limit Texture Memory to 2MB: ----> Off
Force Triple Buffering: ---------> Off

GAME OPTIONS: (Quake 3)
Simple Items: -------------------> Off
Marks On Walls: -----------------> On
Ejecting Brass: -----------------> On
Dynamic Lighting: ---------------> On
Identify Target: ----------------> On
High Quality Sky: ---------------> On
V-Sync: -------------------------> Off
Force Models: -------------------> Off
Team Overlay: -------------------> Off

GRAPHICS SETTINGS:
Graphics Mode: ------------------> 1024x768
Display Depth: ------------------> 32 bit
Fullscreen: ---------------------> On
Lighting: -----------------------> Lightmap
Geomtric Detail: ----------------> High
Texture Detail: -----------------> (3/4)
Texture Quality: ----------------> 16-bit
Filter: -------------------------> Bilinear

----------------------------------------------------------------
Res:  512x384   640x480   800x600  1024x768  1152x960  1280x1024

FPS:   42.0      41.3      40.3      38.8      35.0      28.7
----------------------------------------------------------------

Configuration B: As per Config. A but Force Mipmap Dithering - Enabled.


----------------------------------------------------------------
Res:  512x384   640x480   800x600  1024x768  1152x960  1280x1024

FPS:   42.0      41.3      40.0      38.1      33.8      27.3
----------------------------------------------------------------

So from these figures, this tells me that Force Mipmap Dithering costs very
little performance, but the higher the resolution, the more of an effect it
has. The largest difference it makes is at 1280x1024: 28.8-27.3=1.5 FPS,
or 5.2% slower.


That's the idea anyway.
First up, what;s everyone's suggestion for a BASE configuration? Which
features ought to be on or off? What other settings are needed? Clearly,
nothing that is only controllable from the command line.

Anyway, if you're ready to go, copy this and change the values to suit
your settings and results, then post or email.
TIA,
noise@iinet.net.au

WINDOWS SETTINGS: (Control Panel/Display/Settings/Advanced/3dfxTweaks)
Disable Splash Screen: ----------> 
Force Mipmap Dithering: ---------> 
Limit Texture Memory to 2MB: ----> 
Force Triple Buffering: ---------> 

GAME OPTIONS: (Quake 3)
Simple Items: -------------------> 
Marks On Walls: -----------------> 
Ejecting Brass: -----------------> 
Dynamic Lighting: ---------------> 
Identify Target: ----------------> 
High Quality Sky: ---------------> 
V-Sync: -------------------------> 
Force Models: -------------------> 
Team Overlay: -------------------> 

GRAPHICS SETTINGS:
Graphics Mode: ------------------> 
Display Depth: ------------------> 
Fullscreen: ---------------------> 
Lighting: -----------------------> 
Geomtric Detail: ----------------> 
Texture Detail: -----------------> 
Texture Quality: ----------------> 
Filter: -------------------------> 

----------------------------------------------------------------
Res:  512x384   640x480   800x600  1024x768  1152x960  1280x1024

FPS:   00.0      00.0      00.0      00.0      00.0      00.0
----------------------------------------------------------------
Re: Wanted: Voodoo3 performance tests
#224
Author: THOR[AGQx]
Date: Thu, 09 Dec 1999 00:00
16 lines
1140 bytes
Add possibly a driver company and version?  That may help differenciate
also.


--
                      <<>>THOR[AGQx]<<>>
"One cannot attain pleasure without first enduring pain."


"noise" <noise@iinet.net.au> wrote in message
news:384FF24B.88C7C25A@iinet.net.au...
> Hi,
>
> I'm going to publish a website full of tips and tweaks designed to help
> Voodoo 3 owners get the most out of their cards, especially for Quake 3.
SNIP
SV: Wanted: Voodoo3 performance tests
#225
Author: misko
Date: Thu, 09 Dec 1999 00:00
26 lines
1525 bytes
noise <noise@iinet.net.au> skrev i
diskussionsgruppsmeddelandet:384FFADA.B23BF4C2@iinet.net.au...
>
>
> "THOR[AGQx]" wrote:
> >
> > Add possibly a driver company and version?  That may help differenciate
> > also.
>
> Damn. I was going to put that. Yes, deffinitely mention driver - 3dfx?
> WickedGL? And the version number too.
>
> (I nearly finished that post then had to go to work... when I got home I
> just didn't have the strength to finish it properly 8)

and I'd like an option:

V3 model: (to choose between 1000/2000/3000 and PCI/AGP)
V3 core speed: (for us overclockers...)


if I see those two included, I will mail you the benchmark right away....


misko
Re: Wanted: Voodoo3 performance tests
#227
Author: noise
Date: Fri, 10 Dec 1999 00:00
13 lines
1062 bytes

"THOR[AGQx]" wrote:
> 
> Add possibly a driver company and version?  That may help differenciate
> also.

Damn. I was going to put that. Yes, deffinitely mention driver - 3dfx?
WickedGL? And the version number too.

(I nearly finished that post then had to go to work... when I got home I
just didn't have the strength to finish it properly 8)
noise
Re: Wanted: Voodoo3 performance tests
#228
Author: FuelnFire[cD]
Date: Fri, 10 Dec 1999 00:00
97 lines
4148 bytes
P2-400 (not o'clocked)
192MB pc100
Voodoo3 3000 @ 166Mhz
Wicked3d ver 1.1 (Regular Install)
Voodoo Settings : 
		Alpha = Auto
		Video = Normal
		Gamma = 1.09
		Force Triple buffering
Game Options :  	
		Everything off except Simple Items and Identify Target
Special settings in my cfg : 
		cg_errordecay "0"
 		cg_gibs "0"
 		cg_marks "0"
		cg_shadows "0"
		cg_drawGun "0"
Graphic Options : 	(1)
	           Color Depth - 	32 Bit
		Lighting - 		Vertex
		Geometric Detail - 	High
		Texture Detail - 	3/3
		Texture Quality - 	32 Bit
		Texture Filter -  	Trilinear
 			(2)
	           Color Depth - 	32 Bit
		Lighting - 		Vertex
		Geometric Detail - 	High
		Texture Detail -  	2/3
		Texture Quality - 	32 Bit
		Texture Filter -  	Trilinear
			(3)
		Game Options all ON
		Color Depth - 	32 Bit
		Lighting - 		Lightmap
		Geometric Detail - 	High
		Texture Detail -  	3/3
		Texture Quality - 	32 Bit
		Texture Filter -  	Trilinear

		            (1)		           (2)
(3)
			demo1	demo2		demo1	demo2
demo1	demo2
640x480		53.7	53.8		52.5	53.6		38.6
39.8
800x600		52.8	53.0		51.9	52.2		35.5
34.6
960x720		53.3	53.6		51.9	52.4		34.9
35.3
1024x768		50.8	51.2		49.5	50.1
32.5	33.0

Notes:
I didn't feel it was worth my while to test other resolutions.
I found that with Lighting set to Lightmap that there was a 5-12 fps 
decrease depending on the setting, so I just dimmed the brightness 
control  some and deal with it.
Enabling Force mipmap dithering freezes my pc, so that is not posted
here.
I found that a regular install of wickedgl improves the framerate over
a high res install.
Needless to say I'm sticking with my hard work and going with #1 :)
Anything else needed I'll be glad to supply.
MarkF

On Fri, 10 Dec 1999 noise > wrote:

>Hi,
>
>I'm going to publish a website full of tips and tweaks designed to help
>Voodoo 3 owners get the most out of their cards, especially for Quake 3.
>Since I've had some success with improving the display quality in Q3, I
>had decided to make a Voodoo 3 page, but there's a lot more I could put
>into it, eg. charts of the speed difference all kinds of settings and
>tweaks make (in % points). There's some tweaks that improve visual display
>quality at the expense of speed and vice versa, and there's others that
>make improvements without costing you elsewhere.
>
>I'd like some other Voodoo 3 users to run similar tests, and I'll put
>your results in my website and give you recognition for it (I can't
>give you any money for it though! Volunteers only please!). As I've
>got a K6-III system, I need owners of other systems to do the tests
>too, to help discover what differences in hardware do to affect the
>way Quake 3 runs with your Voodoo 3. Maybe we'll discover that some
>feature of a particular CPU is either well suited or bad for Quake 3,
>or that some motherboards hold you back, and so on.
>But most of all, it's to help Voodoo3 users to make the very most of
>their cards... they may seem to be behind the pack for Quake 3 compared
>to modern cards, but optimising things might just make the difference
>you're looking for.
>
ICQ #53579855
http://members.xoom.com/Rollrgirl/     - The mIRC Remains
http://people.ne.mediaone.net/markkus/ - The Quake Battery
http://www.clancd.com/                 - Clan Cruel Death
http://members.xoom.com/Bill2354/      - Alchemy Garden Ponds
Re: Wanted: Voodoo3 performance tests
#229
Author: FuelnFire cD MarkF
Date: Fri, 10 Dec 1999 00:00
109 lines
4554 bytes
oops, switch 1 and 2 configurations

(FuelnFire[cD]) wrote:

>P2-400 (not o'clocked)
>192MB pc100
>Voodoo3 3000 @ 166Mhz
>Wicked3d ver 1.1 (Regular Install)
>Voodoo Settings : 
>		Alpha = Auto
>		Video = Normal
>		Gamma = 1.09
>		Force Triple buffering
>Game Options :  	
>		Everything off except Simple Items and Identify Target
>Special settings in my cfg : 
>		cg_errordecay "0"
> 		cg_gibs "0"
> 		cg_marks "0"
>		cg_shadows "0"
>		cg_drawGun "0"
>Graphic Options : 	(1)
>	           Color Depth - 	32 Bit
>		Lighting - 		Vertex
>		Geometric Detail - 	High
>		Texture Detail - 	3/3
>		Texture Quality - 	32 Bit
>		Texture Filter -  	Trilinear
> 			(2)
>	           Color Depth - 	32 Bit
>		Lighting - 		Vertex
>		Geometric Detail - 	High
>		Texture Detail -  	2/3
>		Texture Quality - 	32 Bit
>		Texture Filter -  	Trilinear
>			(3)
>		Game Options all ON
>		Color Depth - 	32 Bit
>		Lighting - 		Lightmap
>		Geometric Detail - 	High
>		Texture Detail -  	3/3
>		Texture Quality - 	32 Bit
>		Texture Filter -  	Trilinear
>
>		            (1)		           (2)
>(3)
>			demo1	demo2		demo1	demo2
>demo1	demo2
>640x480		53.7	53.8		52.5	53.6		38.6
>39.8
>800x600		52.8	53.0		51.9	52.2		35.5
>34.6
>960x720		53.3	53.6		51.9	52.4		34.9
>35.3
>1024x768		50.8	51.2		49.5	50.1
>32.5	33.0
>
>Notes:
>I didn't feel it was worth my while to test other resolutions.
>I found that with Lighting set to Lightmap that there was a 5-12 fps 
>decrease depending on the setting, so I just dimmed the brightness 
>control  some and deal with it.
>Enabling Force mipmap dithering freezes my pc, so that is not posted
>here.
>I found that a regular install of wickedgl improves the framerate over
>a high res install.
>Needless to say I'm sticking with my hard work and going with #1 :)
>Anything else needed I'll be glad to supply.
>MarkF
>
>On Fri, 10 Dec 1999 noise > wrote:
>
>>Hi,
>>
>>I'm going to publish a website full of tips and tweaks designed to help
>>Voodoo 3 owners get the most out of their cards, especially for Quake 3.
>>Since I've had some success with improving the display quality in Q3, I
>>had decided to make a Voodoo 3 page, but there's a lot more I could put
>>into it, eg. charts of the speed difference all kinds of settings and
>>tweaks make (in % points). There's some tweaks that improve visual display
>>quality at the expense of speed and vice versa, and there's others that
>>make improvements without costing you elsewhere.
>>
>>I'd like some other Voodoo 3 users to run similar tests, and I'll put
>>your results in my website and give you recognition for it (I can't
>>give you any money for it though! Volunteers only please!). As I've
>>got a K6-III system, I need owners of other systems to do the tests
>>too, to help discover what differences in hardware do to affect the
>>way Quake 3 runs with your Voodoo 3. Maybe we'll discover that some
>>feature of a particular CPU is either well suited or bad for Quake 3,
>>or that some motherboards hold you back, and so on.
>>But most of all, it's to help Voodoo3 users to make the very most of
>>their cards... they may seem to be behind the pack for Quake 3 compared
>>to modern cards, but optimising things might just make the difference
>>you're looking for.
>>
>ICQ #53579855
>http://members.xoom.com/Rollrgirl/     - The mIRC Remains
>http://people.ne.mediaone.net/markkus/ - The Quake Battery
>http://www.clancd.com/                 - Clan Cruel Death
>http://members.xoom.com/Bill2354/      - Alchemy Garden Ponds
>


-
ICQ #53579855
http://members.xoom.com/Rollrgirl/     - The mIRC Remains
http://people.ne.mediaone.net/markkus/ - The Quake Battery
http://www.clancd.com/                 - Clan Cruel Death
Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads