PDA

View Full Version : Front Page - Tessellation Hero Challenge: Results are in...


Pages : [1] 2

ragejg
11-15-10, 02:00 AM
While browsing and looking for some GTX 580 overclocking tips at Xtremesystems.org (http://www.xtremesystems.org/forums/showthread.php?t=262073), I happened upon a thread by a fellow named Ket, who issued a benchmarking challenge using Unigine Heaven 2.1 (http://www.xtremesystems.org/forums/showthread.php?t=259862).

http://img221.imageshack.us/img221/2131/heavenst.jpg


The challenge:

* Resolution: 1680x1050
* DX11 mode
* All settings highest
* 8xAA
* 16xAF
* Extreme Tessellation
* Set the three tessellation sliders to 2.0 within the benchmark


I found this "test of manhood" to be quite interesting, and I thought what might make it even more fun would be to issue the challenge to some other hardware communities. So I called it the "Tessellation Hero Challenge" and asked folks at the NVIDIA forums, XtremeSystems.org, techpowerup!, rage3d, [H]ardforum, Guru3D, Elitebastards and Beyond3D to participate. As of 11/15/2010 here are the results. I apologize in advance for any typos, and also offer up that any of these results are up for debate or discussion.

EDIT: New chart added with CPU and GPU info when accessible.

http://img571.imageshack.us/img571/9472/tesshero20101116.jpg (http://img571.imageshack.us/i/tesshero20101116.jpg/)

As this test stresses using geometry, AA and AF, this test isn't simply a tessellation benchmark, I'd suppose you could call it a "DX11-overdone GPU test". Of course, since AMD's current tech does not perform DX11 tessellation as good as NVIDIA's, you won't see many AMD results on the high end of the list. Furthermore, only GPUs are included on the list. I did this just for fun, as I though it'd be pretty neat to get a few communities involved in the same challenge. If you'd like to visit the respective threads to check the results out in more detail, HERE is the link to all the threads (http://www.nvnews.net/vbulletin/showthread.php?t=157017).

NOTE: First place holder CraptacularOne told me via PM that he used optimized low image quality LOD and had texture filtering quality set to the lowest setting. Though I can't say this is really "cheating", it is highly likely that almost everyone else used either the Quality or Highest Quality setting.

Johnny C
11-15-10, 08:05 AM
I'm amazed to see two 5970's so far up the list...

newparad1gm
11-15-10, 12:13 PM
I'm taking this challenge as soon as I install my two GTX 580s today. Can't wait :D

grey_1
11-15-10, 12:34 PM
I'm amazed to see two 5970's so far up the list...

Wow, I just noticed that and the tri-xfire 5850s are the only AMD cards over 21.1fps

Drolfrawd
11-15-10, 12:36 PM
Sniff ...Drols single gtx 260 won't play that :(

grey_1
11-15-10, 12:38 PM
Sniff ...Drols single gtx 260 won't play that :(

It's ok, my single 480 barely plays it :lol:

I think it will be a while before we see anywhere near that amount of tessellation in a game.

Crysis 4 or 5 maybe? :)

Drolfrawd
11-15-10, 12:46 PM
Maybe by then Mrs Drol will have a lapse in judgement i get a new card :D

doubt it though.

Redeemed
11-15-10, 12:49 PM
I feel like Bob, bringin' up the rear. :(


:p :lol:

weevil
11-15-10, 01:32 PM
Hey JC Ming here, good to see you mate :)

Just got a Gigabyte 580. Wondering if I'll blow it up doing this test lol I've got to get some use out of it first :)

bacon12
11-15-10, 04:51 PM
I'm amazed to see two 5970's so far up the list...

I call bs that they beat two GTX 580s. Same guy, though I don't know if its the same rig.

Redeemed
11-15-10, 04:56 PM
I call bs that they beat two GTX 580s. Same guy, though I don't know if its the same rig.

Isn't the 5970 faster than the 580?

Madpistol
11-15-10, 05:26 PM
Isn't the 5970 faster than the 580?

They're about the same in average performance. If I had to guess, I'd say the 580 is a little stronger on Tessellation though.

=assassin=
11-15-10, 05:30 PM
Looks like my 5850 would struggle if I tried this, but that was expected. Unfortunately my tiny monitor only goes to 1280x1024.

ragejg
11-15-10, 06:35 PM
I have a revised chart that will be ready soon; it fixes a few mistakes and adds more granular detail. :)

Johnny C
11-15-10, 08:41 PM
I call bs that they beat two GTX 580s. Same guy, though I don't know if its the same rig.

Stock the 5970 and the GTX580 trade blows. I'm guessing it's the same rig, or similar rigs and those 5970's are overclocked bigtime. Stock clocks on those 5970's are 5850 speeds correct? I think almost all the 5970's will do the 5870's speeds so maybe that is where he's getting all that gpu power from. Also since there are 4 gpu there are 4 tessellation units which is probably helping frames more than overclocking 3 5870's to higher clocks.

ragejg
11-16-10, 02:22 AM
I re-did the chart, it now has CPU and GPU info, check it out (http://www.nvnews.net/vbulletin/showpost.php?p=2347523&postcount=1).

NOTE: First place holder CraptacularOne told me via PM that he used optimized low image quality LOD and had texture filtering quality set to the lowest setting. Though I can't say this is really "cheating", it is highly likely that almost everyone else used either the Quality or Highest Quality setting.

Redeemed
11-16-10, 03:01 AM
Man, almost makes me wanna' oc some- just to get out of last place. :lol: :o

This was a great idea rage, and a very fun "challenge". Props buddy. :)

grey_1
11-16-10, 04:19 AM
I re-did the chart, it now has CPU and GPU info, check it out (http://www.nvnews.net/vbulletin/showpost.php?p=2347523&postcount=1).

NOTE: First place holder CraptacularOne told me via PM that he used optimized low image quality LOD and had texture filtering quality set to the lowest setting. Though I can't say this is really "cheating", it is highly likely that almost everyone else used either the Quality or Highest Quality setting.

Good job rage - just a suggestion though (as if you aren't doing enough already) show gpu speeds for all cards as a vanilla 480 will have a different "stock" speed than a 480 SC will.

Nice work, editor in grief :lol:

ragejg
11-16-10, 09:01 AM
Good job rage - just a suggestion though (as if you aren't doing enough already) show gpu speeds for all cards as a vanilla 480 will have a different "stock" speed than a 480 SC will.

Nice work, editor in grief :lol:

Thanks grey. I may do more of these "challenges", maybe these types of things (if further developed) could become an NV News hallmark, idk... :p ... a differentiator if you will... (crazy)

Not everyone who posted results included a GPU-Z screenshot, nor did everyone specifically mention clock speeds. (pb) I did the best I could with the info I have.

grey_1
11-16-10, 09:09 AM
Thanks grey. I may do more of these "challenges", maybe these types of things (if further developed) could become an NV News hallmark, idk... :p ... a differentiator if you will... (crazy)

Not everyone who posted results included a GPU-Z screenshot, nor did everyone specifically mention clock speeds. (pb) I did the best I could with the info I have.

Ah, true dat.

I like the idea of the challenges though. :)

ExtremeGrandpa
11-17-10, 06:53 AM
Hello everyone. I registered here to post in this specific thread and dispute the # 1 score in the table. I am a long time member of several forums, and one of the top 10 posters at the Nvidia forums.

To quote ragejg :
I found this "test of manhood" to be quite interesting, and I thought what might make it even more fun would be to issue the challenge to some other hardware communities. So I called it the "Tessellation Hero Challenge" and asked folks at the NVIDIA forums, XtremeSystems.org, techpowerup!, rage3d, [H]ardforum, Guru3D, Elitebastards and Beyond3D to participate. As of 11/15/2010 here are the results. I apologize in advance for any typos, and also offer up that any of these results are up for debate or discussion.

I have the same setup as the #1 place holder and its impossible to obtain that score without intentionally manipulating the settings with the intent to decieve. At XS they eloquently called this tweaking, but members there also complained of this score as well as you probably wont see anymore posts in the nvidia forums thread until you respond to this issue.

Now let me clarify why I am disputing this. We as forum members of whatever forums are creating a huge data base of info people have access to. I for one dislike finding tons of misinformation and having to weed out the B.S.

This chart is posted on the FP and is one of the first things seen when coming there. Now my issue is to place your name or tag on the chart as the creator is your stamp of approval to those test scores at least being fairly accurate. So the biggest issue is that on every forum this is taking place i have seen people call B.S. on that score, yet it still stands.

Again my intentions are not to create stress nor strife, but to try and create data bases on the net that are reliable in nature.

EG
FoldingForTheFuture

ragejg
11-17-10, 09:20 AM
Hello everyone. I registered here to post in this specific thread and dispute the # 1 score in the table. I am a long time member of several forums, and one of the top 10 posters at the Nvidia forums.

To quote ragejg :
I found this "test of manhood" to be quite interesting, and I thought what might make it even more fun would be to issue the challenge to some other hardware communities. So I called it the "Tessellation Hero Challenge" and asked folks at the NVIDIA forums, XtremeSystems.org, techpowerup!, rage3d, [H]ardforum, Guru3D, Elitebastards and Beyond3D to participate. As of 11/15/2010 here are the results. I apologize in advance for any typos, and also offer up that any of these results are up for debate or discussion.

I have the same setup as the #1 place holder and its impossible to obtain that score without intentionally manipulating the settings with the intent to decieve. At XS they eloquently called this tweaking, but members there also complained of this score as well as you probably wont see anymore posts in the nvidia forums thread until you respond to this issue.

Now let me clarify why I am disputing this. We as forum members of whatever forums are creating a huge data base of info people have access to. I for one dislike finding tons of misinformation and having to weed out the B.S.

This chart is posted on the FP and is one of the first things seen when coming there. Now my issue is to place your name or tag on the chart as the creator is your stamp of approval to those test scores at least being fairly accurate. So the biggest issue is that on every forum this is taking place i have seen people call B.S. on that score, yet it still stands.

Again my intentions are not to create stress nor strife, but to try and create data bases on the net that are reliable in nature.

EG
FoldingForTheFuture

Hey Grandpa, welcome to NV News. :)

I asked CraptacularOne to re-run with high-IQ settings with his new cards, GTX 480s.

Results: http://www.xtremesystems.org/forums/newreply.php?do=newreply&p=4629220

ExtremeGrandpa
11-17-10, 09:31 AM
Hey Grandpa, welcome to NV News. :)

I asked CraptacularOne to re-run with high-IQ settings with his new cards, GTX 480s.

Results: http://www.xtremesystems.org/forums/newreply.php?do=newreply&p=4629220


He posted that info yesterday at XS forums thread.

My question is will the #1 slot continue to be occupied by the skewed test result? If so why? You can visit the Nvidia forums and read our thread, and see the only discussion about this challenge is the validity of that #1 score. Also having been and Editor at geeksmack.net and understanding how the trust of your readers is needed to survive in this game. I would think you would want that score removed as its fairly evident its false.

EG

Thanks for the warm welcome given the circumstances.

mullet
11-17-10, 10:50 AM
I can't believe I am even on the list.

ragejg
11-17-10, 10:58 AM
He posted that info yesterday at XS forums thread.

My question is will the #1 slot continue to be occupied by the skewed test result? If so why? You can visit the Nvidia forums and read our thread, and see the only discussion about this challenge is the validity of that #1 score. Also having been and Editor at geeksmack.net and understanding how the trust of your readers is needed to survive in this game. I would think you would want that score removed as its fairly evident its false.

EG

Thanks for the warm welcome given the circumstances.

I'm likely going to remove teh 3x 470 score.. but I'm left wondering if the 3x 480 score is indeed valid for this challenge given that he's running an extremely optimized and barebones OS that he customized for benchmarking.

Again this was done just for fun; I don't want things getting too serious. :p