HardAttack: Review of Voodoo5 5500
Back when the Voodoo3 was released, there was much hype
about its performance. Some said that it would be many times better than the
Voodoo2, others spoke of its downs (ie. lack of 32-bit rendering and such). Overall,
the Voodoo3 was a great disappointment, especially to the die-hard 3dfx fans.
Being one myself, I wondered how this company which had brought about the
biggest milestones of modern 3d, Voodoo Graphics and the legendary Voodoo2,
could have come up with something so short of expectations. This article will
surface thoughts on the latest release in the 3dfx arsenal, one of its last
before selling off to nVidia (curses!).
The box of the Voodoo5 itself is quite an amusing one,
carrying the same old design as the boxes that the Voodoo3’s came in, the
Voodoo5 had an orange face that looks like it had a mud bath on the front. Same
old details on the back and so on. I pulled out the card from the nifty plastic
case it came with that reminded me of the Seashell from Seagate and plugged it
into my available board.
The card is a monster by itself,
barely able to fit in my casing, so I had to pull out the motherboard and put it
on the worktable to get some good pictures. Although the card is huge, it is
actually more like a barren desert. The huge card only houses several small
IC’s and other small items that could have easily been fitted in on a smaller
card, at least in my opinion, but what would I know about all these things.
Lets leave it to them hardware people.
I proceeded to install drivers
and such, using the ones that came with the card first, I later found that 3dfx
hasn’t exactly released a whole lot of drivers for the Voodoo5 which makes you
wonder how come nVidia seems to churn out new drivers every other week. Install
is through an exe program making life easier for the less initiated people
among us. The installation somehow made me wonder if I hadn’t installed it on a
clean hard disk maybe it would cause some conflicts with other display devices.
Anyway, moving on now, the install was pretty easy and it rebooted after that
readying it for some action.
I didn’t exactly have any
beautiful 1GHz Athlon’s lying around so I had to make do with a Duron system,
but I’m quite sure it’s an okay comparison give or take a couple of frames here
and there (more give actually). Here are the specs:
Processor
AMD Duron 850MHz
RAM
160MB PC100 SDRAM
HDD
Maxtor ATA100 15.0GB
MB
Abit KT7 RAID
Voodoo5 Driver
1.0.1
Before getting to the hardcore
stuff, I decided to warm up the card a little bit (not saying that it wasn’t
already hot, but more on the cooling later). 3dMark2000 came out of the cupboard,
and it ran with no problems, looked a little slow, but I didn’t think much of
it. At the end of the tests, it was quite obvious that it WAS a little slow.
The config above came up with a final score of 3357. Drivers I thought, so I
pulled the newest drivers off the 3dfx website, 1.0.4, but that didn’t help
very much. In fact the scores dropped even more. Maybe it was 3dmark? Never
mind, moving on to the core of the article, Quake3 performance.
At first I was considering
getting a GTS card to compare the Voodoo5 to, however, I am now glad that I
didn’t get it. The GTS would have easily run over the Voodoo5. I wouldn’t be
surprised if the Voodoo5 couldn’t even come up to the performance of a GeForce2
MX, but that’s a story for another time.
I tested Quake3 at three
different resolutions, the highest being 1024x768 because I don’t exactly bump
into gamers that use resolutions higher than that very often. The test is run
at Fastest, Normal, and High-Quality configurations with simple items and identify
target on. Everything else is off.
Q3
640x480
800x600
1024x768
Fastest
100.2
99.8
97.7
Normal
92.3
94.2
85.8
Hi-Quality
92.1
94.0
85.2
FSAA 2X
640x480
800x600
1024x768
Fastest
99.8
98.7
93.1
Normal
94.0
81.9
49.8
Hi-Quality
93.9
81.9
49.8
FSAA 4X
640x480
800x600
1024x768
Fastest
97.4
78.4
49.2
Normal
62.6
39.3
22.0
Hi-Quality
62.6
39.3
22.0
The most bizarre thing that happens
is in the no FSAA category. 800x600 resolution pumps out more frames than the
640x480 resolution in the normal and hi-quality settings. Can’t seem to explain
that. From the results I saw here, to say that I was disappointed would be an
understatement. Just maybe it had something to do with the hardware
configuration in this PC, so I ripped out the card and dumped it into my own
machine.
Processor
AMD Duron 700MHz
RAM
256MB PC100 ECC SDRAM
HDD
Seagate UW-SCSI2 18.0GB
MB
ASUS A7Pro
Voodoo5 Driver
1.0.4
Again, clean install, nothing
inside except Windows, the drivers and the test material. I didn’t think that
it would make much of a difference but it was just about time to format anyway.
Q3
640x480
800x600
1024x768
Fastest
112.3
111.0
107.8
Normal
105.7
104.3
100.1
Hi-Quality
101.4
100.8
83.7
FSAA 2X
640x480
800x600
1024x768
Fastest
111.0
106.7
90.3
Normal
103.3
97.7
70.6
Hi-Quality
96.6
73.2
43.2
FSAA 4X
640x480
800x600
1024x768
Fastest
100.0
71.2
43.9
Normal
83.6
55.5
34.1
Hi-Quality
55.0
34.0
17.8
The results from both tests were amusing.
I was expecting frame rate much higher than this, considering the fact that a
Voodoo3 3000 can reach within 10 frames of the Voodoo5 in the no FSAA tests.
Enough of talking of this sad performance, hopefully the FSAA will cheer some
people up.
FSAA
To say the very least, FSAA makes
a HUGE difference when playing games. Even 2X FSAA makes everything look
amazing, circles really look circular, and no more stair-stepping or
pixel-popping. Seeing is believing, and I saw.
I don’t think I need to say which one is the one with FSAA
enabled, but just in case here are a few hints. Look at the pole, the hill, and
the bridge. Just amazing I tell you, and the frames you get are more than enough
to play smoothly. FSAA is just perfect for racing and flight simulation games.
The FPS hit on First Person Shooter games is just too high to warrant enabling
FSAA, but that doesn’t mean that First Person Shooter games don’t look better
either.
No FSAA is on the left and 2X FSAA is on the right. Rather
big difference if you ask me, but I don’t think I’ll be noticing any jaggies or
pixel-popping when I’m running my ass off being chased by some maniac with an
oversized rocket launcher down a tight walkway.
Heat
The biggest non-performance problem that the Voodoo5 faces
other than its obvious behemoth size is heat. During tests the processors
heated up so much that it was pretty much the same temperature as the Duron
processors, meaning up to 50° Celsius. The fans that were provided by 3dfx are
obviously not enough to cool the two VSA-100 processors on board. I decided to
try something funny to the card and see whether it helped. The temperature
dropped to a maximum of about 40° which is quite cool, and the card looked like a real
killer, however, I don’t suggest you try this with your Voodoo5.
Conclusion
The performance of the Voodoo5 5500 is far from what I
would expect out of this company, and not at all near to what I would expect from
what I would be paying for this card if I did buy it. This is all of course
from a Quake3 player, I’m sure that fans of the Need For Speed series, and so
on would be very happy with a card like this. Maybe the card is just targeted
at that group. In short, if you want more bang for your buck, go for something
else like a GeForce2 MX maybe.
1.) Little-Known DOS Commands That Have Saved My Ass - madirish
2.) Rampact Piracy on the Sea of Information - xearthed
3.) Physics utilizing comp technology or computer tech utilizing Physics (Part 2) - josette
4.) State of the Hack Awards #2 - madsaxon
5.) Games Industry Syndicate of the web? - OZONE
6.) A brief look at Quantum Mechanics - josette
7.) HardAttack: Review of Voodoo5 5500 - biatch0
8.) A year in the box - L33tdawg
