Rage3D Discussion Area

Rage3D Discussion Area (http://www.rage3d.com/board/index.php)
-   Other Graphics Cards and 3D Technologies (http://www.rage3d.com/board/forumdisplay.php?f=65)
-   -   New benchmark shows 6800's true weaknesses...multitexturing! (http://www.rage3d.com/board/showthread.php?t=33768304)

mpFLUX Jul 7, 2004 06:26 PM

New benchmark shows 6800's true weaknesses...multitexturing!
 
Again!! Multi-texturing performance holds the NV back!!

Ok...

Here I wanna talk about something that we have talked about in the
past. Its something called "Multi-texturing" and after these past few
years of Nvidia's cards being DOMINATED by Radeons in the multi
texture fill-rate catagory, you would think that Nvidia would have
worked really hard to drastically improve NV40's multi-texturing
capabilities... but they failed to do so, and this is whats going to
make the 6800 look weak in 1 year when current fillrate requirements
double.

And even though their Pixel Shader performance has over
trippled, well over 300% better then 5900 series cards it is
still way inferior to the X800 and even the 9800XT.

Here is the proof that the 6800 reference design is seriously
lacking the "balls" to hold its own under the heaviest demand, and
the proof is in the numbers.

Lets take a look at a new Benchmark tool that is quickly beginning
to show up across the major hardware sites.

Its called "FillRateBenchmark" and its made by a company
called NDaw Interactive.

You can download FillRateBench 2004 from this
website... http://www.geocities.com/ndawinteractive/

I **HIGHLY** suggest that you all download a copy of this benchmark
tool and run it on your 9800 Pro's and 9800 XT's to get a baseline for
your own system.

Here are the settings that seem to be the most widely used
for benchmarking with FILLRATEBENCH.

1024x176 - D16 - No AA - No AF

Please take a look at this link here, it shows the results for
fillratebenchmark for the 6800 Non-Ultra reference 12 pipe
card running 350 mhz core.

http://www.nvnews.net/previews/gefor...6800_61.34.png

Ok.... Now pull up that image in the link above and then
compare it against the score you pulled off of your Radeon
9800 pro or 9800 XT.

See the "Dual Textures", "Triple Textures" and "Quad
Textures" test results? Now compare those against your
9800 Pro.

The 6800 falls behind the 9800 Pro by 30% when Dual
Textures are used on one pass... and falls 40% Behind when 3
textures are used during one pass, and falls OVER 60%!
behind when using quad texturing.

Compare the scores against your cards and see for
yourself...

And the referenced NV40 has 4 more pixel piplines
and more advanced GDDRIII !!

Now I look at the information above and wonder to myself
is high Pixel Shader performance worth more to me then
solid multitexturing performance? NO WAY!!!

Sure you happen to get 8 more "AVERAGE" fps from Aquamark3
with the 6800, but when the "Massive Overdraw" happends, the card
still bites the dust, while the X800 Pro manages to cut through it
quite nicely.

So little tricks here and their to push the framerate up on NV40
give it an appealing look in most benchmarks rough averages but
when the heat is truly on, the 6800 wimpers and cowers away.

If the Non-Ultra has such low multi-texturing performance im
doubting that the GT or Ultra are really going to raise this
bar?

Thanks for readingFlux0r
X800 Pro

ellingsen1 Jul 7, 2004 08:10 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
Quote:

Originally Posted by mpFLUX
Again!! Multi-texturing performance holds the NV back!!

Ok...

Here I wanna talk about something that we have talked about in the
past. Its something called "Multi-texturing" and after these past few
years of Nvidia's cards being DOMINATED by Radeons in the multi
texture fill-rate catagory, you would think that Nvidia would have
worked really hard to drastically improve NV40's multi-texturing
capabilities... but they failed to do so, and this is whats going to
make the 6800 look weak in 1 year when current fillrate requirements
double.

And even though their Pixel Shader performance has over
trippled, well over 300% better then 5900 series cards it is
still way inferior to the X800 and even the 9800XT.

Here is the proof that the 6800 reference design is seriously
lacking the "balls" to hold its own under the heaviest demand, and
the proof is in the numbers.

Lets take a look at a new Benchmark tool that is quickly beginning
to show up across the major hardware sites.

Its called "FillRateBenchmark" and its made by a company
called NDaw Interactive.

You can download FillRateBench 2004 from this
website... http://www.geocities.com/ndawinteractive/

I **HIGHLY** suggest that you all download a copy of this benchmark
tool and run it on your 9800 Pro's and 9800 XT's to get a baseline for
your own system.

Here are the settings that seem to be the most widely used
for benchmarking with FILLRATEBENCH.

1024x176 - D16 - No AA - No AF

Please take a look at this link here, it shows the results for
fillratebenchmark for the 6800 Non-Ultra reference 12 pipe
card running 350 mhz core.

http://www.nvnews.net/previews/gefor...6800_61.34.png

Ok.... Now pull up that image in the link above and then
compare it against the score you pulled off of your Radeon
9800 pro or 9800 XT.

See the "Dual Textures", "Triple Textures" and "Quad
Textures" test results? Now compare those against your
9800 Pro.

The 6800 falls behind the 9800 Pro by 30% when Dual
Textures are used on one pass... and falls 40% Behind when 3
textures are used during one pass, and falls OVER 60%!
behind when using quad texturing.

Compare the scores against your cards and see for
yourself...

And the referenced NV40 has 4 more pixel piplines
and more advanced GDDRIII !!

Now I look at the information above and wonder to myself
is high Pixel Shader performance worth more to me then
solid multitexturing performance? NO WAY!!!

Sure you happen to get 8 more "AVERAGE" fps from Aquamark3
with the 6800, but when the "Massive Overdraw" happends, the card
still bites the dust, while the X800 Pro manages to cut through it
quite nicely.

So little tricks here and their to push the framerate up on NV40
give it an appealing look in most benchmarks rough averages but
when the heat is truly on, the 6800 wimpers and cowers away.

If the Non-Ultra has such low multi-texturing performance im
doubting that the GT or Ultra are really going to raise this
bar?

Thanks for readingFlux0r
X800 Pro

Stop spreading FUD.The Pro/XT lose every time against the GT/Ultras in all of the single,dual, triple, and quad fillrate and texturing speed tests. I personally think the test you are using or your settings are screwed.

Take a look below to see the real results between the cards. By the time they get to quads the Ultra is beating the XT by 50%+.
http://www.xbitlabs.com/articles/vid...y/r420_26.html

Here are the results I received using your fillrate tester and settings with a Ultra.
http://pages.prodigy.net/ellingsen1/Fillratre.JPG

night Jul 7, 2004 09:27 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
wtf did u quote him for :hmm:

scottweil4nd Jul 7, 2004 09:43 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
a benchmark utility from a geocities website? nt

DVS Darkside Jul 7, 2004 09:43 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
Score is greatly affected if you have AA enabled :hmm: turn it and AF off just in case and scores will be normal :bleh:
that little utility is pretty cool actually.

_leech_ Jul 7, 2004 09:47 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
I thought multi-texturing was on it's way out anyways in favour of shaders ever since the whole "three textures per pass"-thing on the original Radeon tanked...

:confused:

mpFLUX Jul 7, 2004 11:07 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
I dont know the difference between the two benchmarks
the one I posted and the one thats on xbit, but... Xbit in
my honest oppionion has ALWAYS been Nvidia biased, you
can beleive what you want.

I have run this particular benchmark on many systems
now and believe its numbers are accurate, the card I
also pitted against was a Non-Ultra, I take these things
into consideration.

And for your artical from xbit, 3/4 of the tests go to
the X800, it started out lookin good... but the more math
involved your Geforce 6 choaks.

:lol:

Flux

mpFLUX Jul 7, 2004 11:11 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
Quote:

Originally Posted by ellingsen1
Here are the results I received using your fillrate tester and settings with a Ultra.
http://pages.prodigy.net/ellingsen1/Fillratre.JPG

Wow... This is JUST THE OPPOSITE of the
6800 Non-Ultra. WTF....

here is the original artical I pulled this off of... that changed my
mind about the real performance of the base architecture.

http://www.nvnews.net/vbulletin/show...2&page=1&pp=15

Your scores are pretty nice. I wonder what makes
the Non-Ultra choke so damn bad.

Thats a pitty

Flux

Spitzfiya Jul 8, 2004 12:43 AM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
Well from what this thread is telling me my 9800Pro owns the 6800U in some areas? I doubt it -_-

It was a good read but is it the truth?

badboy Jul 8, 2004 01:05 AM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
so the GF6800 series own the Radeon 9800 in every game out there beacuse of magic then? yeah k thats good :bleh:
Look at the actual god damn game performance not on stupid benchmark numbers. :rolleyes:
I personally am buying a new graphic card to play games maybe you are buying one to benchmark 24/7 :bleh:

therapture Jul 8, 2004 07:04 AM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
Quote:

Originally Posted by badboy
so the GF6800 series own the Radeon 9800 in every game out there beacuse of magic then? yeah k thats good :bleh:
Look at the actual god damn game performance not on stupid benchmark numbers. :rolleyes:
I personally am buying a new graphic card to play games maybe you are buying one to benchmark 24/7 :bleh:


OWNED...


I play real games too, not just sit around running dumb ass synthetic benches that have nothing in common with real world game play...

Fahim Jul 8, 2004 12:10 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
This benchmark is nice, maybe effective too. But how can my 9800 PRO beat 6800NU?? Guys please post more result so we can understand the efficiency of this benchmark. Here goes mine:


PowerColor 9600XT Bravo 128MB (500,675) CAT 04.6


FillrateBenchmark(tm) 2004 - "easy benchmark series"

Benchmark Main Program Version: FRB_V092
Benchmark Date/Time : 27/06/2004 6:15:56 PM

System Information
-----------------------------------------------------------
CPU : AMD Athlon(tm) 64 Processor 3000+
GFX : C.P. Technologies Inc. RADEON 9600XT BRAVO 128MB
OS : Microsoft Windows XP
Settings : 1024x768 32 bits D24S8 No AA

Benchmark Result
-----------------------------------------------------------
FrameBuffer Clear : 2752 FPS
Color Fill : 1331.272 M-Pixel/s
Z Fill : 1985.584 M-Pixel/s
Color + Z Fill : 1331.272 M-Pixel/s
Single Texture : 1361.471 M-Pixel/s
Single Texture Alpha Blend : 1051.931 M-Pixel/s
Dual Textures : 989.0168 M-Pixel/s
Triple Textures : 654.3114 M-Pixel/s
Quad Textures : 490.7336 M-Pixel/s
1 Floating Poing Texture : 971.4008 M-Pixel/s
Render to Self : 1290.587 M-Pixel/s
PS 1.1 Simple : 1366.504 M-Pixel/s
PS 1.4 Simple : 1366.504 M-Pixel/s
PS 2.0 Simple : 1366.504 M-Pixel/s
PS 2.0 PP Simple : 1363.988 M-Pixel/s
Customized Pixel Shader : 1006.633 M-Pixel/s
PS 2.0 Complex : (Unsupported)
PS 2.0 PP Complex : (Unsupported)
PS 2.0 Massive Register : (Unsupported)
PS 2.0 PP Massive Register : (Unsupported)
PS 2.0 Sincos Procedure Tex : (Unsupported)
PS 2.0 Per-Pixel Lighting : (Unsupported)
-----------------------------------------------------------
* End of FillrateBenchmark Result


MSI GeForce FX 5900XT 128MB (450,850) 61.71


FillrateBenchmark(tm) 2004 - "easy benchmark series"

Benchmark Main Program Version: FRB_V092
Benchmark Date/Time : 29/06/2004 12:26:58 PM

System Information
-----------------------------------------------------------
CPU : AMD Athlon(tm) 64 Processor 3000+
GFX : NVIDIA GeForce FX 5900XT
OS : Microsoft Windows XP
Settings : 1024x768 32 bits D24S8 No AA

Benchmark Result
-----------------------------------------------------------
FrameBuffer Clear : 5747.2 FPS
Color Fill : 1814.456 M-Pixel/s
Z Fill : 3548.381 M-Pixel/s
Color + Z Fill : 1814.456 M-Pixel/s
Single Texture : 1736.442 M-Pixel/s
Single Texture Alpha Blend : 1202.926 M-Pixel/s
Dual Textures : 1215.509 M-Pixel/s
Triple Textures : 822.9225 M-Pixel/s
Quad Textures : 581.3306 M-Pixel/s
1 Floating Poing Texture : (Unsupported)
Render to Self : 1707.187 M-Pixel/s
PS 1.1 Simple : 1790.548 M-Pixel/s
PS 1.4 Simple : 1789.29 M-Pixel/s
PS 2.0 Simple : 586.3636 M-Pixel/s
PS 2.0 PP Simple : 1791.807 M-Pixel/s
Customized Pixel Shader : 444.1768 M-Pixel/s
PS 2.0 Complex : (Unsupported)
PS 2.0 PP Complex : (Unsupported)
PS 2.0 Massive Register : (Unsupported)
PS 2.0 PP Massive Register : (Unsupported)
PS 2.0 Sincos Procedure Tex : (Unsupported)
PS 2.0 Per-Pixel Lighting : (Unsupported)
-----------------------------------------------------------
* End of FillrateBenchmark Result

gts007 Jul 8, 2004 08:19 PM

Re: New benchmark shows 6800's true weaknesses...multitexturing!
 
Quote:

Originally Posted by badboy
so the GF6800 series own the Radeon 9800 in every game out there beacuse of magic then? yeah k thats good :bleh:
Look at the actual god damn game performance not on stupid benchmark numbers. :rolleyes:
I personally am buying a new graphic card to play games maybe you are buying one to benchmark 24/7 :bleh:

lol no sh**. Seems like most people here buy these cards to run benchmarks and argue about them over the internet :nuts:


All times are GMT -5. The time now is 01:03 PM.

Powered by vBulletin® Version 3.6.5
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
All trademarks used are properties of their respective owners. Copyright 1998-2011 Rage3D.com