Resorting to the strokers at Firing Squad, for they at least know how to run
benchmarks, if their editorializing is unsophisticated, here are some
interesting comparisons between the fastest 3 series and the fastest 4
series:
3D Mark 2001 Classic, 1024x768x32
GeForce4 Ti 4600 - Athlon 1.2G = 8689
GeForce3 Ti 500 - Athlon 1.2G = 7103
No surprises there...faster clock speeds give faster aggregate results.
3D Mark 2001 Classic, 1024x768x32 Car Chase, Low Detail
GeForce4 Ti 4600 - Athlon 1.2G = 120.7
GeForce3 Ti 500 - Athlon 1.2G = 110.4
Not as big an increase as the aggregate score. Not surprising either.
3D Mark 2001 Classic, 1024x768x32 Car Chase, High Detail
GeForce4 Ti 4600 - Athlon 1.2G = 38.2
GeForce3 Ti 500 - Athlon 1.2G = 38.2
Disappointing.
3D Mark 2001 Classic, 1024x768x32 Dragothic, Low Detail
GeForce4 Ti 4600 - Athlon 1.2G = 165.3
GeForce3 Ti 500 - Athlon 1.2G = 107.9
3D Mark 2001 Classic, 1024x768x32 Dragothic, High Detail
GeForce4 Ti 4600 - Athlon 1.2G = 94.5
GeForce3 Ti 500 - Athlon 1.2G = 60.9
A very significant gain.
3D Mark 2001 Classic, 1024x768x32 Lobby, Low Detail
GeForce4 Ti 4600 - Athlon 1.2G = 117.6
GeForce3 Ti 500 - Athlon 1.2G = 113.2
3D Mark 2001 Classic, 1024x768x32 Lobby, High Detail
GeForce4 Ti 4600 - Athlon 1.2G = 53.0
GeForce3 Ti 500 - Athlon 1.2G = 52.8
I wonder what it is about Dragothic that allowed such an increase?
3D Mark 2001 Classic, 1024x768x32 Nature
GeForce4 Ti 4600 - Athlon 1.2G = 46.9
GeForce3 Ti 500 - Athlon 1.2G = 37.5
Now on to the gaming FPS benchmarks.
Serious Sam, 1024x768x32
GeForce4 Ti 4600 - Athlon 1.2G = 109.4
GeForce3 Ti 500 - Athlon 1.2G = 109.4
Serious Sam, 1024x768x32
GeForce4 Ti 4600 - Athlon 1.2G = 109.4
GeForce3 Ti 500 - Athlon 1.2G = 109.4
Serious Sam, 1024x768x32x2AA
GeForce4 Ti 4600 - Athlon 1.2G = 108.1
GeForce3 Ti 500 - Athlon 1.2G = 74.9
Serious Sam, 1024x768x32xQuincunxAA
GeForce4 Ti 4600 - Athlon 1.2G = 108.1
GeForce3 Ti 500 - Athlon 1.2G = 58.7
Serious Sam, 1024x768x32x4AA
GeForce4 Ti 4600 - Athlon 1.2G = 66.2
GeForce3 Ti 500 - Athlon 1.2G = Greater than 56 FPS (there seems to be an
error in the reported benchmark...see link to verify)
The GeForce4 certainly can give quincunx for free, but it can't quite
give such a claim to x4AA
RTCW, 1024x768x32
GeForce4 Ti 4600 - Athlon 1.2G = 78.5
GeForce3 Ti 500 - Athlon 1.2G = 78.3
Quake III, 1024x768, High Quality
GeForce4 Ti 4600 - Athlon 1.2G = 185.8
GeForce3 Ti 500 - Athlon 1.2G = 180.4
Quake III, 1024x768x2AA, High Quality
GeForce4 Ti 4600 - Athlon 1.2G = 172.3
GeForce3 Ti 500 - Athlon 1.2G = 121.8
Quake III, 1024x768xQuincunx, High Quality
GeForce4 Ti 4600 - Athlon 1.2G = 172.5
GeForce3 Ti 500 - Athlon 1.2G = 99.1
Quake III, 1024x768x4AA, High Quality
GeForce4 Ti 4600 - Athlon 1.2G = 112.6(?)
GeForce3 Ti 500 - Athlon 1.2G = 77.1
The question mark is for the unreadable tens digit in the benchmark jpg.
It's higher than 100+ FPS.
Analysis of results:
It's quite clear that the GeForce4 Ti4600 has the ability to run x2 and
Quincunx AA with little to no cost. However, keep in mind you still have to
run anisotropic filtering to get the image quality you expect. Anisotrpic
filtering will lower your FPS.
However, x4 AA remains expensive, and could possibly reduce your frame rates
to unacceptable levels, especially when combined with aniso at higher
resolutions than 1024x768.
While it's clear that FPS' increase dramatically with no AA, the FPS are
already sufficient with a geforce3 ti 500 as to make that an unuseful
feature. What is useful is that quincunx is basically free. If you are a
high end gamer, fast CPU, large quality monitor, you understand how
important it is to have the highest visual settings possible and I wouldn't
be against the idea of quncunx being defaulted on in future driver
revisions. The 4600 also allows for x4 AA, when married with a fast CPU
(recall these benches are done with a 1.2G athlon, most high-ends will have
at least 1.53G plus oc'ing of the card itself) with frame rates that you
simply can't find fault with in almost any Direct3D or Open GL gaming
application.
I think high end gamers will appreciate the performance and the easy
upgrade.
Thanks for the bench marks.
My 4600 arrived today. I installed it at lunch and had only enough time to
run 3d mark 2001.
It's replacing an Elsa Gladiac 920.
I can't wait to get home and play some games. See what the 128 megs of ram
does for me.
"SecretSquirrel" <notac...@in.hell> wrote in message
news:%swf8.39297$%b6.91...@typhoon.ne.ipsvc.net...
AJ
"SecretSquirrel" <notac...@in.hell> wrote in message
news:%swf8.39297$%b6.91...@typhoon.ne.ipsvc.net...
marxbig
SecretSquirrel <notac...@in.hell> wrote in message
news:%swf8.39297$%b6.91...@typhoon.ne.ipsvc.net...
> If you have a GeForce 3 classic
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
Check out our new Unlimited Server. No Download or Time Limits!
-----== Over 80,000 Newsgroups - 19 Different Servers! ==-----
The GeForce4 certainly *CAN'T* give quincunx for free...
What the GeForce4 can do is provide performance in current games at 'normal'
resolutions that is sufficiently high that the CPU is the limiting factor -
even when Quincunx AA is turned on...
As more demanding games come out (and/or the processor is upgraded), then
enabling any AA setting will result in a drop in performance.
However, at 1024x768x32, there seems to be no appreciable loss when quincunx
is used in any application, hence the "For free".
Even at 1600x1200x32 you go from 46.1 (no AA) to 38.3 (Quincunx) in Serious
Sam
But in Quake III 1600x1200x32 (no aa) 138 to 62.6 (Quincunx) That isn't for
free...
Certainly, 1.2G, stock video card, 1600x1200x32xQuincunx = 62.6 is
impressive, regardless of your opinion of Quake III!
"Graham Triggs" <gra...@nopsam.biomedcentral.nospam.com> wrote in message
news:a5nk66$a...@dispatch.concentric.net...
--
First of One, Primary Adjunctive Grid Alpha-01
Mech Arena: http://daxin.netfirms.com/
FormulaSAE Racing: http://fsae.utoronto.ca/
"SecretSquirrel" <notac...@in.hell> wrote in message
news:%swf8.39297$%b6.91...@typhoon.ne.ipsvc.net...