Quake III Arena Benchmark Thread

Discussion about old graphics cards, monitors and video related things.

Re: Quake III Arena Benchmark Thread

Postby appiah4 » 2017-7-06 @ 22:47

leileilol wrote:None of those 32-bit color or 32-bit textures will matter on V3 as they're clamped to 16-bit automatically due to what the GL ICD reports back to Q3.


"22-bit" is just a 2x2 box filter smudging your dither patterns. The framebuffer is the same


I can say for sure that the resulting image on the screen is not and the 32 bit option makes it a million times better.
A500:+512K|ACA500+|C1084S
i386:Am386SX-25|4M|TVGA9000B|ES688
i486:U5S-33|8M|GD5428|YMF719|MUNT-Pi
i586:P133|32M|T64+/MX2|V1|CT3980/32M|SD-35
i686:K6-2/400|64M|V2/SLI|CT4500/32M
S370:P3-1200|384M|GF4Ti4200|MX300
S754:A3700+|2G|X1950PRO|SB0350
User avatar
appiah4
l33t
 
Posts: 3878
Joined: 2017-2-19 @ 07:36

Re: Quake III Arena Benchmark Thread

Postby zerker » 2017-7-07 @ 22:25

So I got my second video card installed and benchmarked it too. Results follow:

1346 frames, 27.7 seconds: 48.6 fps

Intel Pentium II-350
Intel SE440BX-2 Motherboard
Windows 98SE
192 MB RAM
ABIT Siluro GeForce 2 MX-400 (64 MB VRAM)

EDIT: Re-tested on Windows XP. Exact same specs otherwise:
1346 frames, 29.8 seconds: 45.2 fps
User avatar
zerker
Member
 
Posts: 231
Joined: 2016-2-11 @ 01:34

Re: Quake III Arena Benchmark Thread

Postby slivercr » 2017-7-13 @ 18:54

One of my new P3-S 1400 arrived today, so I figured I'd test it out and post my results. I was also in the mood to play with a Radeon 9100—a lower clocked 8500 that came out later—so I'm curious under which year you'll put the entry;

144.8 FPS, slivercr, Radeon 9100 @ 250/500 MHz (5.8 Catalyst), Pentium III-S 1400 @ 1400 MHz, intel OR840 (i840), 2 GB RDRAM PC800-45, XP-SP3

When the second Tualatin arrives I'll submit the dual CPU results, as well as those of a GF3 Ti 500 and a GF4 Ti 4600 ;)
Last edited by slivercr on 2017-7-13 @ 20:42, edited 1 time in total.
Outrigger: an ongoing adventure with the OR840
QuForce FX 5800: turn your Quadro into a GeForce
User avatar
slivercr
Member
 
Posts: 306
Joined: 2017-2-16 @ 18:00
Location: Groningen

Re: Quake III Arena Benchmark Thread

Postby appiah4 » 2017-7-13 @ 19:39

I just realized I may have benched with vsync o .. I need to check that in the 3dfx panel and rebench if necessary this weekend.
A500:+512K|ACA500+|C1084S
i386:Am386SX-25|4M|TVGA9000B|ES688
i486:U5S-33|8M|GD5428|YMF719|MUNT-Pi
i586:P133|32M|T64+/MX2|V1|CT3980/32M|SD-35
i686:K6-2/400|64M|V2/SLI|CT4500/32M
S370:P3-1200|384M|GF4Ti4200|MX300
S754:A3700+|2G|X1950PRO|SB0350
User avatar
appiah4
l33t
 
Posts: 3878
Joined: 2017-2-19 @ 07:36

Re: Quake III Arena Benchmark Thread

Postby dexvx » 2017-7-14 @ 04:53

How well does SMP work with this version of Quake 3?
dexvx
Oldbie
 
Posts: 725
Joined: 2017-3-07 @ 03:32
Location: USA

Re: Quake III Arena Benchmark Thread

Postby slivercr » 2017-7-14 @ 10:02

dexvx wrote:How well does SMP work with this version of Quake 3?


Its buggy, to say the least. Hangs are common if you edit the cfg file with r_smp 1: most of the time you will be greeted by a blank screen when launching the game. Using the method above in this thread, editing the shortcut, has yet to fail me. It consistently launches the game for me.

Performance wise, you'll get a nice boost. In my system, with a couple of CuMine 1000 EB and a GeForce3 Ti 500 [GeForce4 Ti 4600] you are looking at 92.5 [93.3] fps without SMP, versus 115.2 [120.6] with SMP enabled.
Outrigger: an ongoing adventure with the OR840
QuForce FX 5800: turn your Quadro into a GeForce
User avatar
slivercr
Member
 
Posts: 306
Joined: 2017-2-16 @ 18:00
Location: Groningen

Re: Quake III Arena Benchmark Thread

Postby zerker » 2017-7-24 @ 22:54

So my Pentium III-450 just arrived, and I swapped out the Pentium II and re-benchmarked it on both my cards (and OSes). I also reformatted my results up to now to match the format on the first page.

Here are the 3DFX results on both processors:
35.5 FPS, zerker, 3Dfx Voodoo III 2000 (4.12.01.0666) @ 143.2 MHz, Pentium II 350 @ 349.19 MHz, Intel SE440BX-2 (82440BX), 192MB SDRAM PC100 CL3-2-2, 98SE

38.9 FPS, zerker, 3Dfx Voodoo III 2000 (4.12.01.0666) @ 143.2 MHz, Pentium III 450 @ 448.96 MHz, Intel SE440BX-2 (82440BX), 192MB SDRAM PC100 CL3-2-2, 98SE



Next for the GeForce 2 MX. Didn't retest Win98 + GeForce on the Pentium III because I don't actually want to keep Windows 98 drivers enabled for this card. Last time I reverted to a backup after testing, which isn't my preference.

48.6 FPS, zerker, ABIT Siluro GeForce 2 MX-400 (81.98) @ 100/182.55 MHz, Pentium II 350 @ 349.19 MHz, Intel SE440BX-2 (82440BX), 192MB SDRAM PC100 CL3-2-2, 98SE

45.2 FPS, zerker, ABIT Siluro GeForce 2 MX-400 (93.71) @ 100/182.55, Pentium II 350 @ 349.19 MHz, Intel SE440BX-2 (82440BX), 192MB SDRAM PC100 CL3-2-2, XP-SP2

51.0 FPS, zerker, ABIT Siluro GeForce 2 MX-400 (93.71) @ 100/182.55 MHz, Pentium III 450 @ 448.96 MHz, Intel SE440BX-2 (82440BX), 192MB SDRAM PC100 CL3-2-2, XP-SP2


Incidentally, can anyone recommend a Windows-XP tool for grabbing the clock speed of a Video card? The Windows 98 tool I already had just bluescreens on Windows XP, and CPU-Z could supposedly grab that information, but is just blank for me right now (hence the TBCs above).
Last edited by zerker on 2017-7-25 @ 22:53, edited 1 time in total.
User avatar
zerker
Member
 
Posts: 231
Joined: 2016-2-11 @ 01:34

Re: Quake III Arena Benchmark Thread

Postby dexvx » 2017-7-25 @ 01:50

You can use Powerstrip, RivaTuner (nvidia only), or for newer cards (at least GeForce 4 Ti/Radeon 9700) GPU-Z.
dexvx
Oldbie
 
Posts: 725
Joined: 2017-3-07 @ 03:32
Location: USA

Re: Quake III Arena Benchmark Thread

Postby zerker » 2017-7-25 @ 22:54

Thanks, RivaTuner got me the numbers. Edited my post accordingly.
User avatar
zerker
Member
 
Posts: 231
Joined: 2016-2-11 @ 01:34

Re: Quake III Arena Benchmark Thread

Postby silikone » 2017-8-02 @ 08:42

Comparing Quake 3 benchmarks to Quake 2, is the difference in GPU capped performance mainly caused by the significant increase in polygons, or are all the "special" blending effects to blame?

leileilol wrote:None of those 32-bit color or 32-bit textures will matter on V3 as they're clamped to 16-bit automatically due to what the GL ICD reports back to Q3.


"22-bit" is just a 2x2 box filter smudging your dither patterns. The framebuffer is the same


Is 16-bit texturing inherent in the Voodoo 3? Are there no texture formats that offer higher quality on it?
Do not refrain from refusing to stop hindering yourself from the opposite of watching nothing other than that which is by no means porn.
User avatar
silikone
Member
 
Posts: 258
Joined: 2012-3-21 @ 19:53

Re: Quake III Arena Benchmark Thread

Postby dexvx » 2017-8-02 @ 19:05

silikone wrote:Comparing Quake 3 benchmarks to Quake 2, is the difference in GPU capped performance mainly caused by the significant increase in polygons, or are all the "special" blending effects to blame?


Polygons, increase texture size, and lighting.

Also, OP needs to update this thread
dexvx
Oldbie
 
Posts: 725
Joined: 2017-3-07 @ 03:32
Location: USA

Re: Quake III Arena Benchmark Thread

Postby silikone » 2017-8-02 @ 19:37

dexvx wrote:
Polygons, increase texture size, and lighting.


What lighting? The lightmap density remains the same, but the dynamic lighting has regressed to blobs that forget to take the third dimension into account, as if they thought they were in the Doom engine. In addition, addition is not what they use. They instead blend with the lightmaps multiplicatively.
I understand that this was done to abolish texture uploads, which Id was a big opponent of, but still, the dlights might as well not be there, as they fail to do their one job: illuminate the darkness.
Do not refrain from refusing to stop hindering yourself from the opposite of watching nothing other than that which is by no means porn.
User avatar
silikone
Member
 
Posts: 258
Joined: 2012-3-21 @ 19:53

Re: Quake III Arena Benchmark Thread

Postby leileilol » 2017-8-03 @ 01:26

Q3's big non-graphical speed bottleneck is actually the QVM system. The slowness of QVMs meant excising a lot of things and putting hard limits on dynamic polygons and marks because it can't handle that much. This is mainly why Q3 didn't have a particle system.

You should see faster results when using DLLs instead, though getting those going would require MSVC6'ing your own from the Q3 SDK source or the original 1.32c gpl source release.

The dlights are a more expensive modulation blend (which cannot light up dark areas)

(some "pro" players also alleged the sound system in Q3's also slow, though i don't feel that's true first-hand)
by the way, DOSBox is not for running Windows 9x
User avatar
leileilol
l33t++
 
Posts: 9767
Joined: 2006-12-16 @ 18:03

Re: Quake III Arena Benchmark Thread

Postby silikone » 2017-8-03 @ 07:33

More expensive? Than additive blending? Then what is the benefit?

Did they not update the sound system at some point? I recall that the rockets did not have the doppler effect in early versions (probably delegating this to the then available A3D support).
Do not refrain from refusing to stop hindering yourself from the opposite of watching nothing other than that which is by no means porn.
User avatar
silikone
Member
 
Posts: 258
Joined: 2012-3-21 @ 19:53

Re: Quake III Arena Benchmark Thread

Postby havli » 2017-8-03 @ 17:08

Nice scores everyone :)
OP updated.
HW museum.cz - my collection of PC hardware
User avatar
havli
Oldbie
 
Posts: 789
Joined: 2014-11-07 @ 16:51
Location: Czech Republic

Re: Quake III Arena Benchmark Thread

Postby leileilol » 2017-8-03 @ 19:45

silikone wrote:More expensive? Than additive blending? Then what is the benefit?


Addiitve would wash out the texture detail. If you want an id tech3 example of additive dynamic lighting, play Alice.

silikone wrote:Did they not update the sound system at some point? I recall that the rockets did not have the doppler effect in early versions (probably delegating this to the then available A3D support).

Starting from 1.25 the soundsystem junked A3D support out as well as regress the channel handling (causing annoying overlapped sounds). This was deliberately done to avoid the "hitch" when going in and out of menus
by the way, DOSBox is not for running Windows 9x
User avatar
leileilol
l33t++
 
Posts: 9767
Joined: 2006-12-16 @ 18:03

Re: Quake III Arena Benchmark Thread

Postby silikone » 2017-8-03 @ 21:34

leileilol wrote:
silikone wrote:More expensive? Than additive blending? Then what is the benefit?

Addiitve would wash out the texture detail. If you want an id tech3 example of additive dynamic lighting, play Alice.


I presume this is a limitation of multitexturing. If they were done in two passes, could the dynamic lights be added so that the albedo would not be affected?
Actually, forget I asked. A veteran shouldn't have to lecture someone for free.
Do not refrain from refusing to stop hindering yourself from the opposite of watching nothing other than that which is by no means porn.
User avatar
silikone
Member
 
Posts: 258
Joined: 2012-3-21 @ 19:53

Re: Quake III Arena Benchmark Thread

Postby leileilol » 2017-8-04 @ 01:27

Unfortunately Q3's shader system is very unfriendly to any idea of combining textures with the lightmap. Jedi Academy did it but its method isn't compatible with other ideas for shaders like detail texturing...


FWIW the Q3tests had the old update-the-lightmap method instead (and this was incompatible with texture compression!!)
by the way, DOSBox is not for running Windows 9x
User avatar
leileilol
l33t++
 
Posts: 9767
Joined: 2006-12-16 @ 18:03

Re: Quake III Arena Benchmark Thread

Postby falloutboy » 2017-8-04 @ 19:17

leileilol wrote:(some "pro" players also alleged the sound system in Q3's also slow, though i don't feel that's true first-hand)

Makes a big difference on my Super Socket 7 system. Have not testet with others.
44.8 fps with sound ; no a3d sound
62.9 fps without sound

Quake3 1.32 with custom settings (quite low)
demo four

K6-III+ 550MHz
Tyan S1590
GeForce3-Ti200 Det. 14.10
Vortex 2
Win98SE

In Quake 2 it's only 5% difference on that system.
falloutboy
Member
 
Posts: 108
Joined: 2014-10-24 @ 19:50

Re: Quake III Arena Benchmark Thread

Postby Fusion » 2017-8-11 @ 04:58

Running the Q3A Demo v1.11, demo001 I get 50.3fps at 1024x768 high settings. With my Voodoo 3 3000 @ 195/195.

With a few tweaks I can hit near 60fps with little image quality loss. :cool:
WIP Retro Files
P3 800 | 512MB PC100| PCI V3 3000 16MB @ 195/195 | CT4780 SB Live! Value| WinME
P3 450 | 128MB PC100| AGP TNT2 Pro 32MB | CT4170 SB16 | MS-DOS 7.10
A64 2.4 | 2GB DDR1 | PCI-E Radeon X800XT @ 575/575 | Win2K
User avatar
Fusion
Member
 
Posts: 192
Joined: 2017-3-06 @ 04:49
Location: Ontario, Canada

PreviousNext

Return to Video

Who is online

Users browsing this forum: No registered users and 2 guests