TessMark 0.2.2: New OpenGL 4 Tessellation Benchmark

Quote:

---

Originally Posted by ElSel10

I want to see 5870 vs GTX480!

---

Just look above - the GTX480- OWNS- the HD5870 in Tesselation ;)

Quote:

---

Originally Posted by tombman

Just look above - the GTX480- OWNS- the HD5870 in Tesselation ;)

---

This is indeed true. nVidia did a very good job on their dynamic tessellation configuration.

So I thought it would be interesting to run Furmark and Tessmark at the same time to see what the result would be.

Tessmark result with both tessmark and furmark running in a window

http://i132.photobucket.com/albums/q...essFurmark.png


Tessmark runs on top of furmark

http://i132.photobucket.com/albums/q...sFurmark_2.png

So my scores drop from 150 tessmarks running in a window to 51 with furmark running in the background.

Settings used in Tessmark and Furmark :

http://i132.photobucket.com/albums/q...r/tess_set.png http://i132.photobucket.com/albums/q...r/furm_set.png



So it would be interesting to see what kind of performance GTX460/470/480 owners would get running furmark and tessmark.

Quote:

---

Originally Posted by Toysoldier

So it would be interesting to see what kind of performance GTX460/470/480 owners would get running furmark and tessmark.

---

GTX 460 800/2000. Using settings from your screenshots.

TessMark - 381 fps
TessMark + FurMark - 124 fps

Quote:

---

Originally Posted by trinibwoy

GTX 460 800/2000. Using settings from your screenshots.

TessMark - 381 fps
TessMark + FurMark - 124 fps

---

Thank You.
So the performance drop seems to be about the same on both type of GPU's.

Charlie was right... err... I mean wrong!

Quote:

---

Originally Posted by hurleybird

Charlie was right... err... I mean wrong!

---

??????????

About Fermi only being great at tessellation when it has nothing else to do.

Quote:

---

Originally Posted by Toysoldier

This is indeed true. nVidia did a very good job on their dynamic tessellation configuration.

So I thought it would be interesting to run Furmark and Tessmark at the same time to see what the result would be.

Tessmark result with both tessmark and furmark running in a window


Tessmark runs on top of furmark

http://i132.photobucket.com/albums/q...sFurmark_2.png

---

now that is Xtreme! :up:

Following everyone's lead; 1024x768(FS), AA-off, PostFX-off:

http://i291.photobucket.com/albums/l...etqwnormal.jpg

http://i291.photobucket.com/albums/l...tqwextreme.jpg

http://i291.photobucket.com/albums/l...etqwinsane.jpg

GTX460 SLI

EDIT: Yes, SLI was not working. I was told to use etqw.exe and it seems to have done the trick!

^ seems SLI not working.. you can check my results at previous page..

Quote:

---

Originally Posted by tombman

Just look above - the GTX480- OWNS- the HD5870 in Tesselation ;)

---

are you surprised? :confused:

its just too bad there arent really any games using tesselation in a way it actually enhances iq...

Quote:

---

Originally Posted by saaya

are you surprised? :confused:

its just too bad there arent really any games using tesselation in a way it actually enhances iq...

---

lets just hope ATI step up their game with the new 6000 series..;)

Quote:

---

Originally Posted by hurleybird

About Fermi only being great at tessellation when it has nothing else to do.

---

Yes, a GTX460 scoring 120 with TessMark with FurMark in the background vs. a 5870 scoring only ~50 under the same configuration is a big victory for Charlie :ROTF:

Quote:

---

Originally Posted by Johnny87au

lets just hope ATI step up their game with the new 6000 series..;)

---

so they score higher in benchmarks and people slap their fanboy epeens around?

wrong thinking! :P
lets hope GAME DEVELOPERS STEP UP THEIR GAME WITH TESSELATION AND USE IT IN A WAY IT ACTUALLY MAKES IT WORTH IT TO GO DX11!!!! :D

Quote:

---

Originally Posted by Arctucas

Following everyone's lead; 1024x768(FS), AA-off, PostFX-off:

http://i291.photobucket.com/albums/l...etqwnormal.jpg

http://i291.photobucket.com/albums/l...etqwinsane.jpg

GTX460 SLI

---

480 vs ^460sli^ @ same clocks

normal runs hotter than insane

http://i56.tinypic.com/91c4s4.png

http://i52.tinypic.com/rvh92w.png

stock cooled @ 910/1820/2200

http://i56.tinypic.com/2v16syp.png

http://i52.tinypic.com/b3sbrs.png

Tried to see how far I could go:

Normal

http://i291.photobucket.com/albums/l...rkNormal-1.png


Extreme

http://i291.photobucket.com/albums/l...kExtreme-1.jpg


Insane

http://i291.photobucket.com/albums/l...rkInsane-1.png


Notice that the more intense the tessellation, the lower the clocks.

Quote:

---

Originally Posted by NapalmV5

stock cooled @ 910/1820/2200

http://i56.tinypic.com/2v16syp.png

http://i52.tinypic.com/b3sbrs.png

---

I wonder how much effect the difference between your Quad @ 4500 vs. my Dual @ 3200 makes?

http://img99.imageshack.us/img99/4412/36266078.png

http://img338.imageshack.us/img338/4143/94895403.png

I see ONE person with custom drivers based off 10.7 running a benchmark which itself reports innoperable with most AMD high end configurations due to bugs and crashes. Also one claim that a 5770 is doing better than a 5870 for the same reasons above.
I would wait for the problems to be ironed out before giving a overall opinion about this benchmark. So far all I can say is nice for having xp support, but kinda pointless if it only works as intended on nvidia so far.

Most likely problem is ATI/AMD's inability to write good OpenGL drivers , looking back in history they always sucked in that area.
And considering we don't have alot of new if any games in OpenGL , ATI might be keeping it a low low priority.

here's some 5970 results for you folks, all in 2560x1600 because I have a big screen:

Also, some details about my setup... Everything is watercooled, running 10.10 catalysts, did a fresh re-install of win7 64bits 2 days ago, the 5970 is slightly overclocked 775/1075, it goes higher but I don't really care, the cpu is at 4.6ghz for 24/7 use but can boot upwards of 4.8 so I just need some phase change. I'm running a double monitor setup (2560x1600, 1920x1200) so perhaps this influences the benchmark results a bit.

Moderate, no AA
http://img254.imageshack.us/img254/8...oderate.th.jpg

Normal, no AA
http://img521.imageshack.us/img521/2...0normal.th.jpg

Normal, 4x MSAA
http://img3.imageshack.us/img3/7481/...al4xmsa.th.jpg

Normal, 8x MSAA
http://img684.imageshack.us/img684/9...al8xmsa.th.jpg

Extreme, no AA
http://img717.imageshack.us/img717/6...extreme.th.jpg

Insane, no AA
http://img600.imageshack.us/img600/6...0insane.th.jpg

http://i159.photobucket.com/albums/t135/BigCharb/3.jpg

http://i159.photobucket.com/albums/t135/BigCharb/2.jpg

http://i159.photobucket.com/albums/t135/BigCharb/1.jpg

from the looks of the Insane tess run, ati have to pull up their socks.

Anyone got 6870 runs to compare to the 5870's? Dont know if tessmark will take advantage of the new method that 6870's have, but worth a try.

Quote:

---

Originally Posted by STEvil

Anyone got 6870 runs to compare to the 5870's? Dont know if tessmark will take advantage of the new method that 6870's have, but worth a try.

---

Everything at stock.

Moderate:
http://lh6.ggpht.com/_S6CHN8qPQnA/TO...k-Moderate.jpg

Normal:
http://lh6.ggpht.com/_S6CHN8qPQnA/TO...ark-Normal.jpg

Extreme:
http://lh3.ggpht.com/_S6CHN8qPQnA/TO...rk-Extreme.jpg

Insane:
http://lh5.ggpht.com/_S6CHN8qPQnA/TO...ark-Insane.jpg

*the links picasa was wanting me to use weren't working right let me know if you can see these.

Edit: Just realized in my pics the memory was at 900 where the actual stock value is 1050, forgot i dropped that down i guess, the benchmark numbers the same on insane, +17 for extreme, +120 or so for Normal, +1040ish for Moderate.