New Voodoo5 drivers and benchmarks

[Wingznut]

Under the options, did you have "P3 optimizations" selected? Make sure you didn't have FSAA or anything like that enabled. Also keep in mind that when you updated the drivers, all the settings get reset back to default. So maybe previously you had something set for faster performance, and now it's defaulted.

Those are the only things that come to mind.

 

[nippyjun]

My performance increased by 20-25% with the new voodoo drivers. Yes!

 

[Wingznut]

(Man, I'm so bored tonight...)

Well, lsd... I accept your challenge. I dialed my cpu down to 480mhz (6.0 x 80mhz), and reran the Q3 benchmark.

At 1280x1024x32, with all details maxed, I get 53.3 fps.

"bwhahaha you voodoo guys make me laugh. "
Who's laughing now?

 

[BFG10K]

Interesting to see the same guys posting benchmarks who only a few days ago were saying that benchmarks are useless and you can't judge anything by them. Why are you posting benchmarks? I thought everyone has to buy the card in order to know how good it is.

 

[Wingznut]

I posted them, for your benefit. Also, because I was asked to by people who were curious.

Btw brainiac, don't put words in my mouth. Never did I say they were useless. I said there's more to buying a card, then just looking at benchmarks. And also that benchmarks are overrated. I still stand by that opinion.

 

[RoboTECH]

BFG, don't be so silly. Benchmarks obviously aren't pointless, but basing your purchasing decision *solely* on them is rather backwards, don't you think? Once you've made your purchase, nothing wrong with optimizing your system.

Benchmarking your own system after a change obviously is a good idea, especially when you change drivers. I know you know this, because of your reply in my (loaded) thread from earlier last week.

I know you're just being facetious here, but be realistic. There are 3 reasons to release new drivers:

1) New features
2) Faster speed
3) Better stability

Seems that the new 5500 drivers show all of the above (apparently fixes the 1600x1200 single processor bug in D3d)

This is good stuff here, don't be so negative.

and Wing....mellow out. BFG is playing you...

 

[Wingznut]

Yeah, I'm kind of a sucker for that type of thing.

I have a real problem with irrationality, and how one person can post some inane opinion, passing it off as fact. There are people who aren't as experienced with PC hardware, who read these boards. And since 90% of the posts here are intelligent, someone unfamiliar with the likes of BFG10K and Hardware might buy into their lines of crap.

Really, I don't care much what kind of video card somebody buys. Pick a GTS card... Pick a V5.... They are both excellent pieces of hardware. You won't be unhappy with either one, no matter what the trolls say.

 

[Eug]

BFG, if you want a speed increase in Q3 on the V3, install the WickedGL drivers. The 3dfx drivers suck in comparison.

Do you guys actually play at those resolutions? I find Q3 extremely annoying to play with anything less than 60 fps in demo001. In fact, I'm running somewhere around 67-69 fps and I still find it irritating at times.

 

[BFG10K]

BFG, don't be so silly. Benchmarks obviously aren't pointless, but basing your purchasing decision *solely* on them is rather backwards, don't you think? Once you've made your purchase, nothing wrong with optimizing your system.

Yes I was being silly but only to prove a point. I agree that purchases should be based on benchmarks and all of the other features. What I don't like to see is people like Wingnut dismissing my statements as false simply because I haven't tried the cards. Website information is accurate, especially of the majority of website have the same results.

and how one person can post some inane opinion, passing it off as fact.

There is nothing opinionated about reading website reviews. Yet again you find the idea that somehow the websites are invalid because I haven't tried the cards. Whether or not I try them the websites' results are equally true.

BFG, if you want a speed increase in Q3 on the V3, install the WickedGL drivers. The 3dfx drivers suck in comparison.

Eug, where can I get these drivers? Are they stable/good quality? Do they run a wide variety of OpenGL games or are they just limited to Quake3?

 

[EMAN]

Damn BFG, I thought you knew what you were talking about but guess not. Boy was I fooled. I won't make that mistake again with a 10 year old.

BFG if I told you I just bought a porsche would you believe me?

I thought so, your so damn gullible! You believe every web site huh, pretty pathetic.



Seriously you never heard of wicked GL?

I thought everybody knew about those. They used to increase but not that much anymore. 3dfx is getting pretty good writing OPEN GL drivers now. Before, not really but it's constantly getting better and better every time.

 

[Wingznut]

Like I mentioned previously, I'm not in the mood to banter back and forth with you anymore, BFG10K, but I will say one thing. My point was that you have said that the V5 sucks. And that's not fair to say, judging soley by benchmarks.

By the way, what do you think of the V5 now, with the latest drivers? Nah, nevermind.

The WickedGL drivers are excellent for the V3. You can get them from Wicked3d.com I noticed both performance and image quality improvements, on my V3. They aren't applicable with every game... Check out the website for more info.

Eug... Yes, I do play at 1280x1024 resolution in Q3. But I use 16bit color. I'm not real sure why, but I really don't notice a difference, with the exception of volumetric lighting (colored smoke, colored fog, water, etc....). But I have volumetric lighting turned off in FPS games, anyway. (All the better to see you with, my dear!) Maybe there is some substance to 3dfx's "22bit" rendering. Either way, 32bit is definitely not worth the performance hit.

In UT, I play at 1024x768 2xFSAA, in Glide. NFS5 gets 1024x768x16 4xFSAA. Same with Madden2001. We'll see about NHL2001, this Friday (I can't wait!)

 

[oldfart]

Get the Wicked3D. Really makes Q3 look good. Use the high resolution install. I just switched to a Radeon Here is a good setup for Q3 on a Voodoo 3. Setup an autoexec.cfg that looks like this:

set r_intensity "1.75"
set r_gamma "0.9"
set r_ext_gamma_control "0"
set r_sawpinterval "1"
set r_ignorehwgamma "1"
set r_overbrightbits "2"

In the 3DFX tools options, use "high" and "smoother". Use 16 bit color depth in the Q3 video options. Will look pretty darn good with this config.

 

[BFG10K]

BFG if I told you I just bought a porsche would you believe me? I thought so, your so damn gullible! You believe every web site huh, pretty pathetic.

WTF are you talking about? Do you think all the websites are there just to trick you? If I read an Anandtech review, am I being gullible? Either you are stupid or paranoid, or both.

Why are you even here? Don't be so "gullible" and read these forums. After all you can't trust anything you read right? Its a conspicracy to trick you.

My point was that you have said that the V5 sucks. And that's not fair to say, judging soley by benchmarks

Perhaps unfair but certainly not invalid.

Anyway thanks for the link. I am just downloading them now.

 

[Wingznut]

"Perhaps unfair but certainly not invalid"

Wait, ... what? (I hear a voice in my head.)

"Wing....mellow out. BFG is playing you..." -- RoBo


*phew*... Thanks, man! I almost got sucked into his nonsense again.

 

[BFG10K]

I am not trying to play anything.

I just tried the WickedGL drivers. They are slower than the 3dfx beta drivers and WQHL drivers with both the high res and low res versions in Quake 3. The image quality is the same but now I see an annoying spinning icon in the top right hand corner.

And yes I have installed them correctly and I am using the same settings in Quake 3 I had before. When I go to "driver info" Quake is definitely seeing them. I have unistalled them now.

Thanks for the link anyway Wingnut. Perhaps 3dfx have actually been optimising their drivers.

 

[merlocka]

Hmmm. Firingsquad's results aren't as good as the 20-25% some are seeing. Weird.

Firingsquad Q3HQ 1.03.00

 

[Geeforcer]

The Tech Report didn't see that kind of improvement either.

The Damage Report: 3dfx's 1.03.00 Voodoo 5 drivers

 

[Eug]

<< I just tried the WickedGL drivers. They are slower than the 3dfx beta drivers and WQHL drivers with both the high res and low res versions in Quake 3. The image quality is the same but now I see an annoying spinning icon in the top right hand corner. >>



Interesting. I had noticed that the WickedGL drivers were still better, although the more recent V3 drivers were definitely better than previously. I can't remember what settings I used to test though, and didn't include the most recent beta drivers.

The spinning icon disappears during gaming.

Wingnut,

What kind of numbers do you get? Just wondering.

 

[Goi]

I noticed that when I was CPU limited, the reference 3dfx drivers perform better than the WickedGL drivers. This happened at 640x480 with all the minimum options/settings on my V3 3000 and old C300@504MHz. However, after I upgraded to a C566@850MHz, the WickedGL drivers are quite a bit faster than reference 3dfx drivers, even the latest one.

 

[Davegod75]

some give me a hollar when the w2k new ones come out

 

[BFG10K]

I noticed that when I was CPU limited, the reference 3dfx drivers perform better than the WickedGL drivers.

That sounds like me! I have a Celeron 500.

 

[Goi]

BFG10K, well there you have it, there's your problem then
It will be especially worse if your 500 is running at a lower FSB. I had my 504MHz running at 4.5x112MHz and it was still limiting the V3.

 

[Eug]

Ah, I see.

I am running a Celery 920 (115 FSB, 77 AGP), CAS 2 memory, BX mobo, V3 at 179.

FYI:

Q3 demo001 800x600 High Quality, WickedGL (says 32-bit colour for some reason).

960 --> 70.0 fps (+1.2 = +1.7% over 920)
920 --> 68.8 fps (+1.3 = +1.9% over 880)
880 --> 67.5 fps (+1.4 = +2.1% over 840)
840 --> 66.1 fps (+2.0 = +3.1% over 800)
800 --> 64.1 fps

I would have tested 533 MHz, but that would require me to open the case. I'm too lazy at the moment. My guess is that the sweet spot for the V3 at 800x600 is about a Celeron 850, give or take. Above that increases become fairly small, and below that it's not worth it paying for the chip (since the Celeron 566 overclocked to 850 is so cheap anyway).




EDIT

At 533 MHz, I get 46.4 fps with the same settings in Q3. UT became virtually unplayable with a little over 21 fps at 800x600 with minimum frame rate of zero and textures at medium.

BFG, it's time to get a new CPU.

 

[Wingznut]

I ran one Q3 benchmark with the Wicked3D drivers, at 1280x1024x32 (HQ). I lost about 3 fps.

One benchmark isn't exactly scientific... But one was all I was in the mood for.

 

[cool]

P3-650E
V3-3000AGP (v4/5 1.03.00-drivers)
160MB-RAM

Q3A in 1024x768x16 everything to max ('3dfx advanced features' Depth Prcision (16bit) at 'faster')

demo001: 44.1 (with WGL and depth precision(16bit) at fast: 44.9)
demo002: 45.7 (&quot;&quot; 47.0)
quaver: 39.3 (&quot;&quot;no diff)