Different types of overclocking leagues!

[knopflerbruce]

I don't think it's time to change the whole system, but to add new rankings is OK:)

[Antinomy]

I agree with BenchBros - we had a number of cooling limited competitions between exUSSR teams, I don't remember of any successful one. I have Celeron 733 results on air better than SS/LN2 of others. The result can be improved and both my CPUs could do this, I couldn't figure which is better. So it's not a cherry-picked, just two random samples.

 

Yet again - we were talking about h/w sharing and how to weep it out - this discussion sums to that one, replaces it or what kind of interfering is it?

 

If you want to make a cooling limitation - you first have to implement and test/confirm a valid temperature reading tech. If it's all about overclockers (global points) league - the tech can work only on high-end and mass hardware.

 

And only after it's implemented and proved working good - there can be a discussion about limitations. Unless, it's like an Aquamark without GPU-Z, you get me

 

Ambient temperature is below zero six months a year in my city. I don't speak for where's Ananerbe from - they got snow in May this year and below -30 at winter.

If it's air - some will use the fridge or LN2 to cool the air. If it's water - they'll use ice and chillers. And you can't differ cause the difference isn't overwhelming, it's less than with switching to a good overclocking hardware sample.

 

The HWBot rules aren't cheat-proof. I vote for not making them more cheat-proof before we make them less cheat-proof or find a way how to not make it easier for cheating when implementing such features.

Edited August 24, 2010 by Antinomy

[vento1]

I like the idea of seperating extreme cooling from air and water gives the new guys with not much $ a chance to get points.

[Antinomy]

And how do you plan to confirm that a result was made on water, not on SS?

[Massman]

And how do you plan to confirm that a result was made on water, not on SS?

 

We can't, that's why we use indirect ways to control it:

 

The issue is indeed that we can't check the temperature. There are two things this concept has:

 

- We impose an invisible MHz limitation to each HW that is based on the average overclockability on air/water. This way, we can remove the entries with 7GHz 980X "on air" as well as exclude golden samples on air cooling

 

- We put the air/water scores in the same ranking as the extreme scores (like it is now). This has the advantage that there will be no direct competition in the air/water league, but everyone will have to push their setup harder anyway to get more points. It's not as simple as just beating your direct competitor. Also, this has the advantage that people who use LN2 regularly will not be able to both appear in the LN2 and the air/water ranking.

 

Our definition of extreme is "everything below zero".

[Antinomy]

As for the first way - I've said about two samples easily ripping LN2/SS results. I'll have to pick up a more crappy result or only underclock the existing ones?

 

As for the second way, I don't exactly get - we have a result an only by trusting people we tell apart whether it was made using cold stuff or not?

 

Everything below zero - how can the current system tell apart, did I make a zero degrees result or at -15?

 

 

As I understand, you want to implement this for both leagues, the hw points and global points?

[Massman]

Part of the problem is the golden samples (which basically reduce the amount of skill needed to achieve a certain result). By not allowing golden samples, or luck, to give you a mass of points, we increase the amount of skill necessary.

 

The enthusiast league will not be the main feature of the site, it will be a side-ranking for the beginning overclockers to get used to the way HWBOT and the rankings/points work. The Xtreme league is where all the action is.

 

Both global and hardware, yes.

[Antinomy]

By not allowing golden samples, or luck, to give you a mass of points, we increase the amount of skill necessary.

So it's a low clock challenge everyone gets same speed, it's only up to tweaking, not overclocking. Makes fun, but how can you confirm that it's not made using downclocking after getting the result? We return to cheat-proof again

[Massman]

-sigh-

 

So what you're saying is: nothing new untill everything is cheatproof. If so, I think we should ban all sports.

[Antinomy]

Were did I say everything? Never thought of me being a maximalist.

The HWBot rules aren't cheat-proof. I vote for not making them more cheat-proof before we make them less cheat-proof or find a way how to not make it easier for cheating when implementing such features.

That's what I said, no less, no more.

[Massman]

More, more, more always ends in most. I have yet to hear solutions that are easy to implent, not stressing for the overclockers and secure the benchmarks by a great factor.

 

You're basically saying: let's do it better. And to the question when is it better you either say "when it's better" or have no answer.

[Antinomy]

And to the question when is it better you either say "when it's better" or have no answer.

I don't remember such a question to be honest. I usually don't give sort of "when it's done" answers.

 

You want to say that separating people in leagues based by info which is proofed only by our words won't stress overclockers? When there's no need to fiddle things up, it's OK. But when it leads to some profit...

 

There's quite a number of cheaters remaining despite the results and even categories being reported and the way it was done. And you want to open the gates even more wide.

I have yet to hear solutions that are easy to implent, not stressing for the overclockers and secure the benchmarks by a great factor.

Duh.

Define "a great factor".

[Diabolique]

Thanks Massman for great effort to popularize overclock!

 

 

Nice feature !

 

It's not feature, it's future.

[Bobnova]

Something like the wrapper on Heaven comes to mind, have it constantly polling CPUz during tests.

 

The clock limit means to me that the enthusiast class is going to be mainly a test of how well you can cheat without being detected.

[Neuromancer]

I cant understand what the big deal is, is cheating that rampant? I expect it is on the very high end of things, but really, is the average overclocker out there going oh wow I can downclock after a bench too look COOLER than I really am?

 

As for cooling air on air cooling contests, or throwing a rad in an ice bath for water cooling contests. How is that cheating? Isn't that ingenuity? Isn't the whole idea of overclocking to improve open the basic abilities of the components we purchased? IF I cant afford a thermochill rad and use a heater core instead and make up the difference (or even blow past the performance of the thermochill) by keeping a cooler full of ice laying around.. or tossing my rad out the window during winter... how is that different then overclocking a cpu/gpu? For that matter..

 

Maybe all benchmarks should now be run only at stock clocks, because some motherboards (IE dell/HP etc) may not support overclocking features, and if it is not fair for everyone it is fair for no one.

 

If you don't allow someone to run an air conditioner say to cool their room to lower ambients in the summer, then what is next? Making everyone go out side to bench on stock settings, and do it on livestream to ensure they are not tweaking the OS?

 

I understand cheating is a concern when prizes or OC travel is on the line... but if you are going to sweat it on stuff that does not really matter (like an enthusiast league), then why even participate in this hobby? Go do RC. Its just as fun (lots of prep and tweak work) gets you out in the sun and if you race them for real (bench) the top 3 cars are torn down to look for cheats.

Edited August 25, 2010 by Neuromancer

[Splave]

Most golden samples are ES, (most not all) and or special manuf samples (IE gigaboost special 5870 ic's)

Disallowing any and all ES scores would quash most of the current issues.

 

I dont care if there is a n00b league, thats fine, hell I think its great. People will buy the more efficient boards and reward the better companies. I just think that either way this will not work. Limiting cooling will still allow the best samples to clock the highest! Limiting clock speed is not only too cheatable but you're not over clocking then....You are software tweaking which yes is good to an extent but it takes away from focusing on the hardware which should be the key factor in what we are doing. Last time I checked this isnt software SWbot.org

Edited August 25, 2010 by Splave

[Massman]

Most golden samples are ES, (most not all) and or special manuf samples (IE gigaboost special 5870 ic's)

Disallowing any and all ES scores would quash most of the current issues.

 

No it won't.

 

- We can't separate the ES VGA from the retail VGA based on screenshots or software

- We rely on the goodwill of manufacturers to not mess around in bios microcode

- We rely on the hope that CPU-Z will always be able to detect ES samples

 

I mean ... hasn't it been proven more than once that MFCs are more than capable of covering up ES hardware? Why go for a solution that will just give us an equal amount of shit.

 

I dont care if there is a n00b league, thats fine, hell I think its great. People will buy the more efficient boards and reward the better companies. I just think that either way this will not work. Limiting cooling will still allow the best samples to clock the highest! Limiting clock speed is not only too cheatable but you're not over clocking then....You are software tweaking which yes is good to an extent but it takes away from focusing on the hardware which should be the key factor in what we are doing. Last time I checked this isnt software SWbot.org

 

The clock frequency would obviously not be set to hurt 1/2 of the overclocking population. It's meant to keep people from submitting 5.5GHz 980X Wprime1024M on 'air cooling' as well as to keep the league just low-profile.

 

It's not designed to be one huge low-clock challenge.

[Antinomy]

Most golden samples are ES, (most not all) and or special manuf samples (IE gigaboost special 5870 ic's)

This happened only with the latest Core i7/i5/i3 CPUs. All the rest are on the contrary better using retail samples.