Fillrate benchmark, 6800

phial

Splendid
Oct 29, 2002
6,757
0
25,780
Seen this at <A HREF="http://www.nvnews.net/vbulletin/showthread.php?t=29862" target="_new">nvnews</A>




theres this new benchmark out where you can test the pixel pushing powers of your gpu , and its very excellent in teh way it tests the card. <A HREF="http://www.nvnews.net/previews/geforce_6800/fillrate_benchmark_geforce_6800_61.34.png" target="_new">Screenshot</A> of the results a 6800NU gets.

<A HREF="http://ftp://fillrate:fillrate@219.237.118.29/FillrateBenchmark_V092_b.rar" target="_new">Download it Here</A>

heres the test restults my 9800pro gets clocked at 411/365

System Information
-----------------------------------------------------------
CPU : Unknown CPU Type
GFX : RADEON 9800 PRO
OS : Microsoft Windows XP
Settings : 1024x768 32 bits D16 No AA

Benchmark Result
-----------------------------------------------------------
FrameBuffer Clear : 5657.6 FPS
Color Fill : 1847.172 M-Pixel/s
Z Fill : 2582.013 M-Pixel/s
Color + Z Fill : 1361.471 M-Pixel/s
Single Texture : 1834.589 M-Pixel/s
Single Texture Alpha Blend : 1842.138 M-Pixel/s
Dual Textures : 1452.068 M-Pixel/s
Triple Textures : 1004.116 M-Pixel/s
Quad Textures : 767.5576 M-Pixel/s
1 Floating Poing Texture : 1449.551 M-Pixel/s
Render to Self : 2573.835 M-Pixel/s
PS 1.1 Simple : 1774.191 M-Pixel/s
PS 1.4 Simple : 1776.707 M-Pixel/s
PS 2.0 Simple : 1774.191 M-Pixel/s
PS 2.0 PP Simple : 1776.707 M-Pixel/s
Customized Pixel Shader : 1565.314 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


the 6800NU, a 12pipeline architecture, gets nearly 3 times the performance in some areas. some very important areas.


i should have kept my 9500pro and saved up for a new generation card after this summer ;(




-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
one thing to note is that things may change when DX9c is released, and the 6800 will more than likely support all the tests while improving the performance of the now supported ones



also, the "Customized Pixel Shader" test is kind of interesting. 6800 gets 372mpixels, while my 9800pro gets 1565mpixels. thats a large difference, and i wonder if it has to do with dx9c and drivers

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
we will see what?

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

entium

Distinguished
May 23, 2004
961
0
18,980
nvidia's whole line is really nice, the lowest end card the 6200 is probably going to be just as fast as the 5900 or 5950. I'm sure ATi will do the same.
 

cefoskey

Distinguished
Jan 20, 2003
440
0
18,780
eighteen gazillion mega ultra texels, four hundred super multi linear verticies....all youre looking at are numbers from a Nvidia developed benchmark program. Id rather see some game IQ and performance tests. Seeing a 30% increase in some benchmark with some ambigious numbers doesnt do anything for me.

"Who is General Failure, and why is he reading my drive?"
P4 3.0 HT, Intel D865GBF, 512MB Crucial PC3200 DDR, WD 36GB Raptor 10,000RPM, BBA Radeon 9600PRO, SB Audigy, Hauppage WinTV
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
how do you know its developed by nvidia?

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
bump



how do you know its developed by nvidia?

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

entium

Distinguished
May 23, 2004
961
0
18,980
When did you get your card, and from who, some companies will let you send it back within 90 days or so. might want to check out the return policy
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
LOL

you missed the thread i made about it?



i know a guy who works in a warehouse, and this box of 9800pros fell out of the back of one of the trucks, crushing the packaging. they cant put crushed cardboard boxes on retail display, so he sold me one at cost for 200$ canadian

i cant return it, because it was a "under the table sale". of course there is a warranty tho



ill probably just end up selling it for 150$ or something in a few months

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
proved what point?

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

entium

Distinguished
May 23, 2004
961
0
18,980
look any of you go buy both cards run the tests on your own machines. Also go around other forums if you like and talk to people that have both cards. They will tell you the 6800 is better.

My engine isn't optimized for either of the two cards (actually its optimized for shader 2.0 for Parrallex bump mapping and this benfits ATi alot more, without the optimizations the x800 was crawling).

During PS 1 and 2 tests, we get an average fill rate of ;


120 million polys/sec on both cards, 4 million polygons per screen @30 fps

using SM 3.0 (nV Ogl specific paths) since we aren't using Dx for our engine yet. nV's performance goes up 100% we are now doing 240 million polygons per/sec 4 million at 60 fps.

Added to that we add tesselation it goes up more (ati's cards can't do this unless we use true form and true form doesn't support displacment of the new polys)

Our Terra Frima: Havok demo on the Cry engine the 6800 pulls ahead we have a butt load of shadows in the game from multiple light sources, both cards have equal quality. Forcing Ogl, it doesn't look as good on either cards (the Ogl port was only for PS2, all the shaders are disabled for the most part) but nV performance is higher by a huge amount.
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
yes all that does sound very enticing. i too think that once dx9c is released and some things to show it off, then people will probably be more impressed with the 6800 than they are now..

if games are even released soon, and its hard to tell. ...SM2.0 took FOREVER to take off. the gaming industry needs to catch up very fast


-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
man that barely made any sense, hahahah wow im baked from work

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

entium

Distinguished
May 23, 2004
961
0
18,980
lol

SM 3.0 is more of an extension to 2.0 so its very easy to add in support. But performance gains are huge. Beceuase of branching. That where the secret lies for SM 3.0. is not that the code is faster. Its just able to do more at the same time. And the programmers don't need to worry about it. The hardware takes care of it automatically.
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
yes i am aware of the branching, and it seems like it would be as powerful as hardware t&l itself


but we havent really seen it yet. I WISH THEY WOULD RELEASE DX9C!!!!!!!!!

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
whoa, look what the 4.6 Catalysts did to my card!

System Information
-----------------------------------------------------------
CPU : Unknown CPU Typ
GFX : RADEON 9800 PRO
OS : Microsoft Windows XP
Settings : 1024x768 32 bits D16 No AA

Benchmark Result
-----------------------------------------------------------
FrameBuffer Clear : 5721.6 FPS
Color Fill : 2748.108 M-Pixel/s
Z Fill : 2811.022 M-Pixel/s
Color + Z Fill : 2121.479 M-Pixel/s
Single Texture : 2768.241 M-Pixel/s
Single Texture Alpha Blend : 1935.252 M-Pixel/s
Dual Textures : 1628.229 M-Pixel/s
Triple Textures : 1069.547 M-Pixel/s
Quad Textures : 800.2731 M-Pixel/s
1 Floating Poing Texture : 1600.546 M-Pixel/s
Render to Self : 2654.575 M-Pixel/s
PS 1.1 Simple : 2798.439 M-Pixel/s
PS 1.4 Simple : 2795.923 M-Pixel/s
PS 2.0 Simple : 2798.439 M-Pixel/s
PS 2.0 PP Simple : 2798.439 M-Pixel/s
Customized Pixel Shader : 1660.944 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


its kinda wierd isnt it

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
yea it is! thats from upgrading from Catalyst 4.4, to 4.6

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 
The single textures are impressive, but once you add more then the score come closer in line.

Remember that the GF6800 series benifits from 32x0/24x0 architecture and will process those less complex pixel for incredible raw single texture numbers and high z-fill. It's not as important as the multitexture numbers, since there is very few games that will use simple texels. The one exception will be for shadows in D]|[ which should give the GF6800 series a considerable advanatage if the shadows ae heavy enough. Interesting to see it under the D]|[ engine with/without heavy shadows.

So considering that, it's not surprising that it gets about 3 times as much because it can push 24 simple pixels through per clock cycle, which is 3 times as much, of course there are also frequency differences which favour your card somewhat.

When looking at more complex shaders the gap narrows, and it becomes primarily a 12 vs 8pipe race.


Results are similar though to <A HREF="http://www.digit-life.com/articles2/gffx/nv40-rx800-2-p1.html" target="_new">those from Digit-Life</A> which show trememndous 0 and 1 texture results, and then down to normal.


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK :evil: