The Overclocker asks... Live competition rules

[The_Overclocker]

Here's your chance to seriously influence the next few big overclocking events. Issue one of The Overclocker is asking the following question (and will link to this XS thread)...

At AOCC there was some heroic overclocking on show. However, some felt that the overall results reflected the teams with the best CPUs too closely and that those with poor CPUs could not compete. We know that upcoming tournaments will randomly distribute CPUs to teams but does this make the results a lucky dip?

Live competition is new and not yet perfected. Should certain frequencies be limited so that the most skilled and efficient overclockers rise to the top? Or should we leave at hardware distribution being random?

www.theoverclocker.com (http://www.theoverclocker.com)

[[XC] gomeler]

Overclocking has always and will always be a hobby that involves a large measure of luck. Sure, it isn't great when you are competing and your CPU is 300MHz slower but that's part of the game. Live competitions will rarely display a bencher's best potential, it'll display his ability to build and rebuild a system with speed and then test the limits of that system all within a time limit. Do I like this style of competition? Not particularly as it results in quick and dirty overclocks, but my few experiences with it have been positive and it's growing on me.

If you were to limit frequencies then you cap the mastery of pouring nitrogen and maintaining stable temps. Who wants to compete when the maximum frequency is 5200MHz? Might as well bring a single-stage and leave the LN2 gear at home. Even capping at 5500MHz would be ungodly low with current chips given how well the E0 stepping from Intel is clocking. Capping frequencies would turn this into a stock car race and frankly those are boring as hell.

[Gautam]

Less "exciting" more FAIR.

[[XC] gomeler]

When has it ever been about being fair? Overclocking sadly is something that relies on either having a deep wallet or a lucky touch. Since the deep wallet can't factor into the competition, that leaves it on luck. What are you going to do, limit it to 5GHz? At 5GHz you can do 526x9.5, 500x10, 476x10.5, or 454x11, all FSB combinations possible with a QX. What if somebodies QX can't hit 526FSB but someone else's can? Are you going to cap FSB now also? How about graphic cards? Cap those frequencies also? When does it become fair and when does it become boring?

[Gautam]

Might as well. Even if you capped everything, some people would get higher scores than others. ;)

[Planet]

I dont think there should be a cap but I think the cpus should be pre binned so they are all very close. I have noticed this with the past couple live contests which kinda hurts the contest and the best dont always win. The whole time limits suck to giving like an hour to setup and bench one bench is kinda hard. Specially if you loose a component or power.

[[XC] gomeler]

Might as well. Even if you capped everything, some people would get higher scores than others. ;)

True but are they truly the best? Some people tweak, some people brute force it and a select few toe the line in between. If you are going to cap frequencies you might as well have everyone run stock speeds on a box cooler and call it a day :down:


I dont think there should be a cap but I think the cpus should be pre binned so they are all very close. I have noticed this with the past couple live contests which kinda hurts the contest and the best dont always win. The whole time limits suck to giving like an hour to setup and bench one bench is kinda hard. Specially if you loose a component or power.

This would be a very good idea. It wouldn't take too much effort and I imagine you could sell the binned processors afterwards.

[Harshal]

Having had a super bad chip for recent OC compy I would still stick to "No CAP" for everything. Why limit when you can push high?? Then whats the point of overclocking??? Agreed, it kinda sucks when you dont get points but its OK if I still cant get points as my chip or GFX was stuck becoz of certain issues. I would like all to keep trying for best possible in every situation :D

EDIT:
@Cpt.Planet, Pre-Binned CPUs, very good idea. That way one can leave it on participant to get maximum out of chip. :up:

[NightRaven]

harshal u may want to know the aocc was prebinned to speed i think.

BUT it wasnt binned when it came to cold boots and cold bugs!

not too easy to test all 20 cpus and find similiar ones huh!

[Harshal]

^^^^ Ohhh I didnt know that. But then -85c was a real PITA everytime I ran Vantage CPU Test would do it for us and I think there was one more case where COLD issue was at much high temp then ours. Also I agree 20 CPUs with same limits is difficult but not impossible for big companies.

[T_M]

NR i seriously doubt that statement about the prebinning.
Or else someone clearly cant tell the difference between 5.5GHz max chip and 5.8GHz max chip for a competition about CPU (SuperPI + 01)......

Back onto the question at hand, my opinion is that something definately needs to be done be it binning or capping (to a speed that is still extreme).
Having competed internationally over 5 times now (cant remember) it is most definately a lucky dip contest for the top few spots, the real interesting stuff usually occurs in the middle ranks where people are using skill to edge over someone else (since they both have average chips). Ive been in those middle ranks several times now and its really a lot more fun that way.

For me these competions should be about testing the skill of the 'benchmarker' (not 'overclocker' which is what they all advertise), which we all know is about tweaking and consistency. Overclocking for me is something that you cannot really compete in with these random hardware draw formats. Every single person in the room knows how to install a pot and pour LN2 sufficiently to get the chip to hit near max, so it always leads to chip quality making the difference.
Skill at pouring and volt tweaking barely comes into it with such limited time frames - throw the setup together, plug in volts that you think should work and stretch the clocks to high but safe values. No one in the room is running the hardware on the very edge like we do at home in our dungeons.

So far my favourite competition was an indonesian one where the benchmarks were not so heavily CPU influenced, and even contained one that most of us were not familiar with (timedemo of Q3).

[NightRaven]

why so tim?

when u test on air and low volts.

it's different from when u run high volts and ln2.

china mentioned to me their cpu sucked on low volts and air. but flew when it was given high volts.

i'm not saying that the aocc one for prebinned. but at least tested to be all working :D

[elmor]

Very interesting discussion this, I will add my point of view. As Tim says the top spots are very much about a good cpu since you're pretty much at the same level. Still from last competition you can look at for example Team Finland whom I'm using as a comparison as we sat right next to them :) They had a really crappy cpu, like not even 5.4Ghz or something for SPI, but still managed to claim 5th place.

Capping the frequency would really make the competition alot more boring since your skill with the hardware (pouring skills, controlling the CB, knowledge about the hardware) will make you able to run higher frequencies than others that don't have those skills even with the same chip. For me it's a very big part of the game. I don't know about the others but we could run our cpu at the edge most of the time due to Roberts fantastic skill with pouring and hw knowledge. I'm pretty much sure that there were others that had as good a cpu as the Chinese, they just didn't get it to run at the same frequency. Another thing is that both we and the Chinese had very good planning, we were very early with with our results in each benchmark to make sure to have enough time to really push the system and not end up without any score at all.

[The0men]

If benching for 3d, then why bother to cap frequencies? although its possible, its very unlikely someone should get golden RAM, golden CPU, golden motherboard and golden GPU. I think things would even out if all the hardware is supplied, tested and roughly binned, or maybe even just tested for cold boot and cold bug.

And if you cant get your CPU quite as high as someone else, spend more time messing about with RAM and GPU etc.

If you cap the frequency, you really take all the excitement out of it.

[T_M]

TheOmen, so how would you feel in a SuperPI 1M or 8M round where your CPU does 5.4GHz and someone elses does 5.8GHz?
Yes 3D does make things much more fair, but only if there many more 3D or system benches than 2D or CPU heavy ones.

Lets say the comp was SuperPI 4M (CPU), and then 3d03/05/06 that would be not so bad.
But when its SPI, '01, and then Vantage being the only system test it makes things much more skewed.

[The0men]

Like I do every day seeing that my quad only does 3.7ghz and every else's does 4.... Nothing... It's not the end of the world, It's a compitition.

Edit: Did I even mention SPi?

[T_M]

No but you mentioned GPU making it more fair, which in SPI it doesnt.

The difference between you benching at home and at a live competition is that it is your choice not to upgrade at home to a more competative chip.
At a live demo you get the chip you are given at thats it.

Maybe making a compeition Vantage only would make things way better all round?

[The0men]

Yer and I mentioned 3d benching for a reason. It was a suggestion.

I deliberately made no mention to SPi for that reason. What are you trying to argue about?

I'm not in the least saying benching Pi with a lower CPU is any sort of advantage, Im saying if it were 3d Benching or at least taken into acount, then it would be more measure of skill, less on hardware depencies, the real skill comes in being able to keep all the components of your computer going, not just stressing your CPU for 8 seconds.

[T_M]

OK i read it differently, like you were ignoring the fact that there are 2D benches used.

[The0men]

OK i read it differently, like you were ignoring the fact that there are 2D benches used.

No worries, It wasnt my intention, just a suggestion to The Overclocker.

[SF3D]

Elmor already did a nice sum up, but I have to say it too. In AOCC we had the worst CPU we both have ever tested :shrug: Still we managed to get the 5th place in total competition. There is so much more factors in the contest, it's not just the cpu.

I don't have any complaints, but I have been thinking what our scores would have been with better cpu. That's life and we have to accept it :) It was fun and that is enough for me :up:

[Sampsa]

Maybe there could be two set of runs. After first run the best team will give its CPU to last team and wise versa. Although this arrangement wouldn't help the teams in the middle..

[The0men]

Maybe there could be two set of runs. After first run the best team will give its CPU to last team and wise versa. Although this arrangement wouldn't help the teams in the middle..

Put them in a hat and lucky dip again? lol

[Oliver]

Quite wise words Elmors, exactly: making restriction on freq or sth would not help neither bcoz then u cannot express ur knowledge nor skills in pouring and such...

Same would be if limit sets to ex: 5.4g and then u can play with tweaking, but ag. its about overclocking, combined with tweaking, we would never play with limit uh? @ keep pushing it...

[NightRaven]

Maybe there could be two set of runs. After first run the best team will give its CPU to last team and wise versa. Although this arrangement wouldn't help the teams in the middle..

but dont u think it would take AGES to do that swap from so cold Sampsa my man?

[Sampsa]

but dont u think it would take AGES to do that swap from so cold Sampsa my man?

First session, 1 hour break (swap the CPUs), second session. Ever heard of heat gun? :)

[NBF]

I just watch this competitions, so I may have a different view.

Cap would take out all the fun for spectators. We like to see who got highest clocks as well as scores, how people setup their fsb, RAM, multi... and what is important for each bench.

I think Cpt.Planets idea might work well, pre binned hardware would make it closer to fair, and it's not hard to do. Swapping CPUs would be very fair, but I think it would work well only if competition were longer.

[Oliver]

and pre boads, vga and mem uh?...

Hey thats the spirit in overclocking, get out what u can from the specified hardware u got at hand and have a good time ;)...

[[XC] gomeler]

but dont u think it would take AGES to do that swap from so cold Sampsa my man?

Blow torch :cool: A neat trick Fugger showed us in California :clap:


and pre boads, vga and mem uh?...

Hey thats the spirit in overclocking, get out what u can from the specified hardware u got at hand and have a good time ;)...

Precisely! :up: Overclocking isn't about limits, it's about breaking limits.

[Gautam]

Man this topic is already taking off.

You came up with a good one Nick. ;) :up:

[ojdr2001]

I think luck is part of any competition.... one time luck, another time bad luck

Our first cpu at AOCC had CB at -75. Than we were luckier with 2nd one

Where it is really unfair is in the real world where money and sponsorships make the real difference. There are guys testing 50/100 cpus. Others just buy one piece and wait for luck

So dont complain it's life :)

Go benching :)

[youngpro]

i think, no pretesting of cpu, first benchmark, everyone runs their benchmarks, then all cpus are reclaimed and randomly given out again, next benchmark, cpus randomized and redistrubted again... still a luck factor, but if your the guy that gets the 5.2 when everyone else is at 5.5 then youll get a better cpu next round no doubt..

this is how it must be done

[Gautam]

I really do like that idea. :up:

[TraX]

I really do like that idea. :up:
+1 - Its a great idea & most fair solution imho. Only down side I see is extending the length of the event?

[massman]

+1 - Its a great idea & most fair solution imho. Only down side I see is extending the length of the event?

Multiple days, only best go through to the next round?

[jas420221]

Is it possible to have the teams use the same chips? Like rotate them when they are done saying they arent cooked? (since burn in is BS). Or at least give each team 3 chips or so and do an average?

[NightRaven]

First session, 1 hour break (swap the CPUs), second session. Ever heard of heat gun? :)

Ah yes of course!


gomeler;3200749']Blow torch :cool: A neat trick Fugger showed us in California :clap:



Precisely! :up: Overclocking isn't about limits, it's about breaking limits.

:eek: Even better! i cant get that legally around here so didnt think of that!


I think luck is part of any competition.... one time luck, another time bad luck

Our first cpu at AOCC had CB at -75. Than we were luckier with 2nd one

Where it is really unfair is in the real world where money and sponsorships make the real difference. There are guys testing 50/100 cpus. Others just buy one piece and wait for luck

So dont complain it's life :)

Go benching :)

haha yea Team India was unlucky enough to get ur CPU when they unknowingly exchanged for it!



i think, no pretesting of cpu, first benchmark, everyone runs their benchmarks, then all cpus are reclaimed and randomly given out again, next benchmark, cpus randomized and redistrubted again... still a luck factor, but if your the guy that gets the 5.2 when everyone else is at 5.5 then youll get a better cpu next round no doubt..

this is how it must be done

thats a good idea James!

[[XC] gomeler]

Only problem is many chips have many difference coldbugs and finding those coldbugs can be a huge pain in the butt. So.. this would make an event miserable unless the coldbug was determined and documented for each chip.

[Gautam]

Now that's where the skill part of the equation comes in. ;)

Someone like Vince could find a CPU's max in minutes along with it's cold bug point, while it'd take others hours.

[[XC] gomeler]

Now that's where the skill part of the equation comes in. ;)

Someone like Vince could find a CPU's max in minutes along with it's cold bug point, while it'd take others hours.

Well yes, it would take me 20 minutes or so to find a 95% maximum but could you imagine how much of a pain in the arse it would be to swap CPUs 6 times in a competition? I already hate having to rebuild my setup once or twice in a benching session, but 6 times? I've done that once during a competition and we all saw how nearly disastrous that was :ROTF:

[Stapler]

Why not impose different rules different benchmarks, or create multiple events for benchmarks that are highly CPU dependent. Super Pi, especially 1m is never going to be an even playing field when clock speeds are involved, so just limit Super Pi and maybe '01 to a reasonable clock speed, but don't cap off other events. Another alternative is to create two events for Super Pi, one at a reasonable clock speed that every one can be expected to achieve, and another no holds barred. If benchers were given twice the time normally alotted to a bench, but told to produce one capped result and one all out result I think it would add some strategy and depth to the competition. The capped result could also be weighted in the scoring/alottment of points so that it favors tweakers rather than luck.

[Oliver]

Gomeler: Exactly...

Good idea James, and as said before, only thing that should not be in overclocking competition is limitation...

[eva2000]

maybe allow each team to pick 3 cpus to play with so they can pick the best one - still lucky dip i suppose

[Oliver]

u could do it like max speed is x and then tweak u to best time, like low clock challenge and other hand if 3 pcs cpu cpu each either good sponsor or $$ is needed

[RTB]

An idea could be to create brackets of cpu speed, every 100 mhz or so (at the very high-end). You would normally be competing within the bracket, but anyone with a higher CPU speed and lower scores would still lose to you of course. This only really works when there are a lot of competitors, so there's competition in almost every bracket (that has someone in it). Though a competition would need more rules than just the usage of brackets.

[clon22]

Guess luck plays a big factor no matter what.
But James' idea is not bad.

[elmor]

As I said before, any kind of limit for clock frequencies kills a big part of the game. Getting that extra 100Mhz out of your chip is also part of the contest. Only way to make it fair is to pretest the cpus on ln2 beforehand and pick out as many equal chips as needed.

[Pt1t]

^^ same CPU for all during one week contest will be more Fair , A bit long but fair :)

[zanzabar]

^^ same CPU for all during one week contest will be more Fair , A bit long but fair :)

what happens when some1 kills it

[Lucky_n00b]

I'm still new to this overclocking stuff, and have attended only one of international OC event(AOCC2k7, Team AlHen - Indonesia)..

From my point of view, there is at least 2 things that can be done to make the competition a little bit more 'fair' without sacrificing the purpose of the competition itself : Breaking The Limit

1) pre-testing the HW(CPU/GPU) using LN2 before the competition is a very good idea, since the sponsor can choose the hardware that have similar clocks,maybe choose 20 out of 100 chips that have near-equal clocks (CPU/GPU))

2) Instead of using 'rank-based' point systems(Points will be allocated according to ranking of your scores), use a 'point-table' systems(Points will be allocated according to point table) where the benchmark score range was converted into point like this :

e.g 3DM2k1

> 100k = 10pts
80.000 - 99.999 = 9 pts
60.000 - 79.999 = 8 pts
and so on...

Everest READ
>14k = 10pts
13.000-13.999 = 9 pts
12.000-12.999 = 8 pts
and so on...

* not exactly have to be like this though *

By using 'point-table' systems like this, if some contestant were to be limited by some aspects of their systems, they can still get good points if they can somehow manage to get the score within the points range

Last but not least,Prepare a great number of benchmarks(10 or more) , and let The contestant CHOOSE the benchmark they want to do(not all of the benchmarks need to be done, e.g just 5 out of 10 benchmarks)

That way some people that are limited to CPU Speed can choose 3D Benches, those who are limited with their GPU can choose 2D Bench, etc..

Cons :
- with using this scoring systems, it means that someone needs to test all the h/w , in all benchmarks.....and that's NOT going to be easy..
- There will be a high probability that the final score will be a draw

Well, that's only my opinion...

* Hopefully it is still understandable,I'm very sorry for my bad English :( *

PS.
Lately there have been a lot of International OC Event, I do hope that sometime soon when I got enough skills, I can join in and have fun with U guys ;)

[giorgioprimo]

As I said before, any kind of limit for clock frequencies kills a big part of the game. Getting that extra 100Mhz out of your chip is also part of the contest. Only way to make it fair is to pretest the cpus on ln2 beforehand and pick out as many equal chips as needed.


same to me ...........

[Gautam]

I concur.