Tessellation Hero Challenge!! Bring it!!

[howard stern]

NINJA EDIT: One of the tessellation sliders in this run was only set to 1. Howard Stern has new results later in this thread. - ragejg

Unigine
Heaven Benchmark v2.1
FPS: 49.3
Scores: 1241
Min FPS: 22.4
Max FPS: 110.4

Hardware
Binary: Windows 32bit Visual C++ 1500 Release May 21 2010
Operating system: Windows 7 (build 7600) 64bit
CPU model: Intel(R) Core(TM) i7 CPU 950 @ 3.07GHz
CPU flags: 4337MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 SSE42 HTT
GPU model: NVIDIA GeForce GTX 470 8.17.12.6099 1280Mb

Settings
Render: direct3d11
Mode: 1920x1200 8xAA fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 16x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
Replication: disabled
Tessellation: extreme

[newparad1gm]

I'm thinking I could do better than this. I'm running it in Surround mode, but on a single monitor, so I have no idea what sort of driver shenanigans are going on and if that impacts performance. I'll try running it in single monitor SLI mode and see what that does.

[ragejg]

New results chart is up in the News Articles thread. Check it out!

[ExtremeGrandpa]

Originally Posted by AdamK47

Also, since the results are simple HTML, they can be edited. The results by CraptacularOne appear to be Normal tessellation results with the results file edited with "extreme". There is no physical way three GTX 470s at those speeds could get that result.

I agree whole heartedly and thats exactly why I am disputing the results table. If the #1 score if false then the entire table should be taken with a grain of salt. To even entertain the idea of keeping the score will only encourage people to give false results in the future. Thus making it harder to get a good mix of true test runs to get a true comparison.

I understand its all in the name of fun, but when people purposely skew there test results for a little forum glory, its no longer fun.

EG FoldingForTheFuture

[ragejg]

Originally Posted by ExtremeGrandpa

I agree whole heartedly and thats exactly why I am disputing the results table. If the #1 score if false then the entire table should be taken with a grain of salt. To even entertain the idea of keeping the score will only encourage people to give false results in the future. Thus making it harder to get a good mix of true test runs to get a true comparison.

I understand its all in the name of fun, but when people purposely skew there test results for a little forum glory, its no longer fun.

EG FoldingForTheFuture

From http://www.xtremesystems.org/forums/...eply&p=4629220

Originally Posted by CraptacularOne

Nobody is promoting false anything. You people are really just blowing this way out of proportion. Setting image quality in the nvidia control panel to maximum performance is hardly cheating, nor is setting a slightly lower LOD bias. Almost everyone who behchmarks uses small LOD adjustments as well as the "maximum performance" setting the control panel.

I was asked via PM to re run the benchmark, without any LOD or anything else. However not having GTX470's anymore makes that impossible. So instead here is my results with 3 GTX480's @ 925/1850/4400 and an i7 920 at 4.4Ghz.

Originally Posted by grey_1

This bench doesn't seem to respond much to overclocking. I've only picked up 1-2 fps between 725 core and 775 core. Not a huge overclock, but I expected a little more. Is my res (1680x1050) limiting my 480? Would a 1080 monitor bring out more of what the 480 is capable of?

BUMP!

Anyone?

[Roadhog]

Originally Posted by grey_1

BUMP!

Anyone?

1920x1080 still isnt that high. Get at least 1920x1200, 2560x1440, or 2560x1600.

[Redeemed]

Originally Posted by grey_1

BUMP!

Anyone?

I'm rocking 1920 x 1200 w/ a 5770 and I can play every game I own (including Crysis) with 100% maxed in-game settings... err actually I think in Metro I have to turn off that advanced DOF or whatever it is. That's the only exception. Most games I'm rocking a fair amount of AA as well.

So yeah, your 480 would likely do very well at any higher res.

Thanks for the replies guys. I was just wondering if my current res was keeping me a little cpu limited with this card.

I think I need a higher rez to play with, and try this bench again.

[howard stern]

I made a mistake on the benchmark I only moved one of the sliders to 2.0. I took a 8 point hit from my first run. My bad I was not trying to cheat.

Unigine
Heaven Benchmark v2.1
FPS: 41.8
Scores: 1052
Min FPS: 15.0
Max FPS: 116.6

Hardware
Binary: Windows 32bit Visual C++ 1500 Release May 21 2010
Operating system: Windows 7 (build 7600) 64bit
CPU model: Intel(R) Core(TM) i7 CPU 950 @ 3.07GHz
CPU flags: 4312MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 SSE42 HTT
GPU model: NVIDIA GeForce GTX 470 8.17.12.6099 1280Mb

Settings
Render: direct3d11
Mode: 1680x1050 8xAA fullscreen
Shaders: high
Textures: high
Filter: trilinear
Anisotropy: 16x
Occlusion: enabled
Refraction: enabled
Volumetric: enabled
Replication: disabled
Tessellation: extreme

Unigine Corp. © 2005-2010

[ragejg]

Grey, that res definitely limits a 480. 1920x1080 is a perfect match for assuring that every game can have every setting maxed plus 4x or 8x AA. 1920x1200, same, except likely just 4x... and 2560x1440, same except 2x to 4x.

[AdamK47]

I've really don't give a "crap" anymore, but I still think CraptacularOne's scores are a bit suspicious. If I did everything he did by overclocking my system to it's limits and setting everything in the control panel to performance, there would still be no way for me to get to 88.5 FPS. That windmill section he took a screenshot of is where the FPS are at their highest. It shows only 80.8 there. Perhaps it's a dip because the screenshot was taken at the time, but if not an average frame rate of 88.5 would mean that section would be into the 100s.