Discussion in 'Benchmark Mayhem' started by RS-X, Dec 7, 2010.
CPU @ 4400 MHz
GPU @ 1075/1550 MHz
By jaxmacfl at 2012-03-04
Done, just wanted pass 16000 pnts first.
I7 2600K @ 5114mhz + 2x 7970 1210/1520mhz ( stock coolers )
More when i get my waterblock .
@Jax .... nice score !!
holy chit thats lookin great lane.
WARNING: if your videocards have still STOCK cooler then too high OC + hot may overburn videocards.
I tested from 700core/1848 mem (stock) to 760core/1900mem (OC) at stock cooler and I got no any artifacts on fur stress test (afterburner) while I got 77 ºC full load@100% FAN speed but it's my special for benchmarks.
Will push it a little harder when the water cooling gear arrives.
Got some improvements now the watercooling is in and some overclocking going on...
Ah ha! You thought you could get away with it by posting a screenshot. What you may have overlooked is the URL in the screenshot.
Sherlock is watching you :nerd:
rofl! if I had that type of hardware I wouldnt go that far just to get more numbers, hell if you want to see my 6970 get 8500points I'll just turn on Lucid MVP which I'm pretty sure many people will start doing in the future
Why would I overlook the URL? Its a requirement when benching with hardware bot. Do you guys have driver requirements here for your benchmarking? Sorry if i missed them in the OP?
Use default settings when you benchmark, don't modify or turn off tessellation settings. Totally defeats the purpose of a benchmark that uses tessellation.
Sorry, at default she does just under 20'000 as a rule.
Will run and get a screenshot.
There is no rules, actually it doesn't matter what you do but if you're going to show your scores to the public and you "modify" things to get different results of course you'll get criticized. I mean if you like big numbers than thats you but if you're going to show it off at least show it legit. There no rule on drivers or whatever you do, if modifying settings to get better results is your thing then go for it but don't expect much respect or positive feedback from other people as they know the results aren't real.
I'll show you right now.
"oooh snap look at my HD6970 score, its sooooooo fast! I got some improvements on air cooling guys! this is 1 video card! OMG!" see what I'm saying, its easy to manipulate results
But did you know I was using tessellation off? Did you know I was using Lucid MVP? of course it would be impossible for 1 6970 to get this score.
So I'm saying do what you want with benchmarking results but at least play in the same field as everyone else if you want people to give positive feedback . You have the right hardware, just tweak/OC it to the best you can to get real results.
My feeling is, with benchmarking you really need to specify such things at the start of the topic. I am sure you wont disagree that the accepted masters of overclock benchmark tables are HW-Bot.net and they specify that a tesselation driver adjustment is fine, and anyone who wants to make progress in their tables will clearly NEED to utilise it, myself included.
However, as far as I can see, this thread hasnt specified anything, thus leaving it open to personal interpretation, which is why I submitted a run directly from my hardware bot submissions.
I wasnt intending to misslead anyone guys, just trying to join in the forum, but do agree with what your saying and shall bear in mind that you guys here like to leave all software at default and tune only hardware.
lol its alright, its just people were used to it from the past before 3dmark caught on to it and for other people it doesnt show the true potential of the cards, i mean in the end we shouldnt really be arguing or making this an issue, benchmarking is benchmarking :nerd: do what you want to get the results you want
Well.. maybe other sites like hwbot accept cheats, hacks and invalid results, but my oppinion is that competing on that level is more like a hackers game, than overclocking. I guess the reason for allowing it is that earlier versions of 3dmark software cant detect the cheat.
If we like to compare results here we should go for valid results, which also allow competition between amd and nvidia.
Text in the red box says that the results are invalid.
Is really HWbot authorise the "use " of decrease tesselation level for 3Dmark 11 ? Im not sure, normally it should be made with a valid Futuremark approbation, ( including the driver release) and it goes unvalid if tesselation level is decreased.
Well ofc in term of competition, ( LN2 ), this can be different, but even if the use of LOD quality change, mipmap, HBAO quality, etc is authorised, this is still discutable.. ( the reason why HWBOT run their own Unigine benchmark tools, with the exact same setting for all, like this no cheat possible on quality )...
But yes i aggree with other, if it is just here for show your scores, you need to indicate if you have decrease or not the tesselation level. Its a little bit like post Unigine scores or Tessmark, and remove tesselation and show an "extreme tesselation " level performance score... or use Vantage with PhysX or run 3Dmark03 with a white screens during all the bench due to an extreme LOD level.
Thoses scores are no more directly updated to HWBOT from here anyway, you need to enter them manually. So if they have change their rules for 3Dmark11... here it is good to precise it, like this there's no problem with comparaison of scores.
Hwbot dont care if you mess with the tess and they have not since the slider was there.
Hey benching is done to get highest score so i cant knock it, just like a stripped os or other things you can do that has nothing to do with hardware.
I like to bench hardware not software but thats just me.
It is a tess bench why make it ok to turn it off?
they cant enforce it so its ok.
When it comes to the bot you can do whatever you want as long as it legal in Their rules.
However in threads like these there should be some kind of settings rule as people tend to use these threads to compare the performance of their rigs and are not trying to improve scores to increase HWBot points.
Maybe a sub forum or something could be created for the HWbot team members that could be all about better scores to avoid confusion in the future?
Cowie, do you use stripped OS's? Do you use specific OS's for certain benchies or XP for 2D and Win7 for 3D stuff like that?