Mipmap tweaking, cheating or legit tweak?

[Rol-Co]

I think the only allowed program of benching is rivatuner, any other program used except riva and 3dmark itself is a no go......or am i thinking too simple now.

mipmap is not avalible at rivatuner so not legit to use for normal benching though.

edit:
nm.

[T_M]

Yet to hear any comments from the "legends"...... do they dare not speak on it?
Really without their blessing this will never be resolved.

1070fps nature single card?
http://service.futuremark.com/compare?2k1=9318549

[dinos22]

Yet to hear any comments from the "legends"...... do they dare not speak on it?
Really without their blessing this will never be resolved.

1070fps nature single card?
http://service.futuremark.com/compare?2k1=9318549

nothing wrong with those scores

have a look at top 5-10 runs
they are all similar (unless they are all using it)

and SM3>SM2 too

[T_M]

Hipro5 single Ultra vs Kingpin single Ultra

SM2 > SM3
http://service.futuremark.com/result...eResultType=14

SM3 > SM2
http://service.futuremark.com/result...eResultType=14
Also what happened to Firefly in this?

[jas420221]

THe only thing I do to bench is lower everything to performance instead of quality. I do not touch LOD.

But w/e that picture was, wasnt the nature I know, and is not a 'tweak' I have ever seen.

[dinos22]

Hipro5 single Ultra vs Kingpin single Ultra

SM2 > SM3
http://service.futuremark.com/result...eResultType=14

SM3 > SM2
http://service.futuremark.com/result...eResultType=14

have a look at Nvidia based runs on HWBOT

most guys have higher SM3

[EDIT] i was looking at multiple card runs though
seems as though single cards are opposite
but kinpin benched at much higher CPU clock speed than anyone else which could explain it (not sure)

i thought this "tweak" only helps '01 anyways :S

[T_M]

Apparently Firefly takes a dive when you bench with Mipmap tweak

Anyway, lets hear from Kingpin / Fugger, Hipro, Shamino, Pedro, etc first hand

1100 nature single Ultra?
http://service.futuremark.com/result...reResultType=6

[T_M]

i thought this "tweak" only helps '01 anyways :S

I presume it was banned from '03 for a reason

[filmbot]

Even though I have yet to bench 3D yet, but planned to, it never crossed my mind to use since its been so long since Ive even had the program on any of my HDs.

But since the view of most here is that its frowned upon, I will make sure that I wont use it in the future.

Though Id still like to hear from the big guys.

[Zeus]

Everybody, tanks for contributing so far.

Maybe i should have created a poll and let the outcome be the standard of what we, Xtreme System members will use for published 3d benches?

There are quite a few people who think it's ok but the people against it are taking the upper hand it seems.

If we don't take a stand now this will always be an argument to accuse people for cheating so i think a norm has to be pronounced.

I suggest we wait for the big guns to arrive and give their opinion on this subject and make a decision based on what they think and based on what gets most votes here.

We should all run our benches according to that rule, "we" should be the people that bench for XS.

This is just an idea, i'm not the boss around here and am not able to force things but i think it's pretty cool if we as XS can set a standard.

What do you guys say, good idea or just impossible?

[T_M]

I'd love to hear confirmation of the runs i linked above too.

[STEvil]

I think it really depends what level you wish to bench at.

If you want to bench all out no rules...

If you want to bench all out no obscene...

If you want to bench all out generic tweaks (driver panel only, my preferred)

If you want to bench all out with no image quality tweaks...



As it is I wont knock anyone for using tweaks that are accepted. LOD is, Mip isnt really..

[tosunermc]

Pci-e 140MHZ
Graphic cards are on air cooling, only lod bias and mipmap. Motherboard ise P5N-T Deluxe

Before mipmap, nature fps was 1330. Below as you can see with mipmap nature climbs up to 1500 FPS.. Is it acceptable or not?

[metro.cl]

I have no idea how to apply this tweak, anyone can show me where to find it?

About this topic, not sure if it is cheat but if many people are using it then maybe we all should

[K404]

Are the opinions of the 3DMark top 10 more important than anyone elses?

Seeing as the tweak is not a question of skill, their opinion, IMHO, is just another comment of no higher or lower weighting.

I WOULD like them to post though, along with as many other benchers as possible

[Mekoa]

Not a believer in adjusting or tweaking settings like this. I mod my hardware settings to get better performance and leave the rest at stock.

Depends on what your opinion of benching is. For some I guess its about tweaking as much as you can to get the best out of your entire rig. For me its modding clocks and hardware settings to get the best out of that. Far as I go for software changes is CCC settings to high performance, about it

Just my 2 cents.

[kiwi]

For 03, 05 and 06 marks it is certainly not acceptable - run yourself and you will see why
For 01 it is debatable - again run and see why

I have no idea how to apply this tweak, anyone can show me where to find it?

About this topic, not sure if it is cheat but if many people are using it then maybe we all should

It's not a secret, it's been posted here on XS more than a year ago if you use a search

[massman]

Why would we even think of making this illegimate for 03/05/06, but legit for 01?

[metro.cl]

For 03, 05 and 06 marks it is certainly not acceptable - run yourself and you will see why
For 01 it is debatable - again run and see why



It's not a secret, it's been posted here on XS more than a year ago if you use a search

Thanks i guess i need to download another tool and test it myselft

[kiwi]

Why would we even think of making this illegimate for 03/05/06, but legit for 01?

Did you run and check? Hard to explain. Using mipmap you can still understand what's going on in 01, describe main objects of environment, see them etc. But in 03,05,06 you can't, you have completely unreal picture of what's going on and you can't describe what's going on because you simply don't see any environment, what you see is just some color patterns. Something like that

[massman]

I have used this 'tweak' more than 3 years ago in one bench and decided not to use it anymore as it completely ruins the benchmarks. 03/05/06 is obvious as there's nothing recognisible anymore.

But if you would show run nvhp tweaked nature run to someone who's never benched 3DMark, would he know what the scene is about? We can describe what's going on because we've seen the nature run about a million times. If you've never seen that run, you won't be able to describe what's happening

[SF3D]

I think that mipmap tweaks shouldn't be allowed, cause when you use it, you loose some data from original image/frame

With lod you just have less geometry points and every particle is more softer, but the amount of data is same as before.

Could someone verify / deny this. Someone must know a lot about 3D modeling. We can't just speculate how does it look like. We have to know what mipmap tweak with nvhardpage really does.

[KTE]

I'm in agreement with Tim on his last posts.

[T_M]

*watches tumbleweed roll down the street*

Im sure the legends will have read this thread already......
Perhaps they are considering how their response might affect the ruling since it seems many of them use it?

[dinos22]

just PM ppl you want an opinion from