Team OC Cup 2012 - Season 1

[Massman]

At HWBOT, we're currently planning a Team OC Cup competition with a set up similar to the the yearly Country Cup. The plan is to feature several seasons per year, each season with different hardware, different benchmarks and different limitations. As said, we're currently in the design phase trying to gather everything we learned from the past three Country Cups and figure out how to improve the competition format.

 

I expect the first season(s) to be quite similar to the Country Cup. In the second half of 2012, I hope we will have a functional version of a neat application we're working on that should make this kind of competition more exciting.

 

We're still thinking about the ideal length of a season. The Country Cup is fine in length, but because it's very intensive (10-day stages), 6 weeks is a little stretching it for most. I think we're looking at a 2/3-month season with stages ~ 3 weeks (with maybe one 'blitzkrieg'-stage of 1 week).

 

Anyway, I'm hoping to get the first season going around April this year, with finalized format/rules before March 1st (so there's roughly one month to prepare for the contest).

 

If you have any suggestion or ideas, just post

 

UPDATE:

 

Start: June 1st.

End: September 1st.

 

- 1 competition, 6 sub-competitions, 43 stages.

- 19 GPUs, 17 CPUs, 7 MEMs.

 

More exact info soon.

[knopflerbruce]

Sounds interesting! If the idea is to make a "season", why not have a few competitions spread out over a few months? Then it's more like a season the way it's defined in sports.

[K404]

I'd like to see a requirement that each team submits a minimum number of results per week. Sure, everyone will sandbag the best results for the end, but at least there will be SOMETHING to watch.

 

OR..... whichever team is in the lead for highest number of days. Meaning, sandbagging is the worst thing to do.

 

 

OR..... whichever team is in the lead for highest number of days. Meaning, sandbagging is the worst thing to do.

 

I'd like to see that implemented in a few places, so the biggest names of year x are remembered in some kind of hall of HWB Champions. No point having a "season" because the best scores go up on the last day. Instead, focus on who is at the top either the longest, or the most consistently. (Which is better..... "unbeaten for e.g....45 days", or "was No.1 for 42/45 days")

 

Anyway... that's another thread

Edited January 8, 2012 by K404

[Dry_Ice777]

Hi Massman. I have a couple of sentences about your idea:

 

1. 3 weeks = 1 season (I think it is the best time solution).

2. There are 2-3 competitions in 1 season.

3. So that more people were able to show themselvesn in teams, don't make competitons only on top-end hardware. Also would like to compete in some old hardware too (purely personal desire )

 

That is, you give tasks, within team determines who distributes what will be engaged. Competitions with old hardwre will be good for teams, who haven't have LGA2011 & 4*HD7970

 

K404, as I understand you, we can make up a formula for calculating points. That is what was filed before the final result, the more chances to win It is not necessary to submit the best possible result. You can make the weaker result, but in the first day and you can win the stage.

Edited January 8, 2012 by Dry_Ice777

[Massman]

There are a couple of ways to prevent sand-bagging in competitions, but we must beware of demonising last-minute winning submissions. It's not because the winning submission was made in the last 5 seconds of the competition that the person who made that score doesn't deserve to win. If, in soccer, a team is trailing by 2-0 five minutes from the end but keeps pushing and eventually end up scoring 3 goals in the last couple of minutes, that's a deserved win in my opinion.

 

The biggest problem with sandbagging is that certain people don't show anything until the last second. That was one of the biggest issues of F1OC ("5 minutes of excitement") and is still an 'issue' in pretty much all competitions. A system where you have to submit one score a week is a nice idea, but won't solve the excitement part of this issue: people would just start with uber-crap score, then submit a lesser crap score, then a normal score and in the last week their insane score. Another idea would be to grant points on a per-week basis, but then you just make the competition about who's able to submit the fastests rather than who is able to produce the best score.

 

Next staff meeting, we'll look into the complexity of a competition format that excludes the 'worst teams' after a certain period. I can't recall the exact terminology, but it's like this:

 

- competition last 4 weeks

- 1st week: all can submit

- 2nd week: best 20 teams continue

- 3rd week: best 10 teams continue

- 4th week: best 5 compete for the stage win.

 

This way, you'll have to make sure your score is good enough 4x in one competition to be eligible for the next round.

[I.nfraR.ed]

Or maybe direct eliminations with team pairs or even a qualification round, then several groups, quarter finals, semi-finals... lol, sounds like a Football World Cup .

Edited January 8, 2012 by I.nfraR.ed

[Massman]

Or maybe direct eliminations with team pairs or even a qualification round, then several groups, quarter finals, semi-finals... lol, sounds like a Footbal World Cup .

 

We have a plan to add the ability to create knock-out competitions, but it's quite a complex task that requires two weeks full-time development. So, it will not be for the first months.

[Hondacity]

im still sick from the country cup

[sumonpathak]

while you are at it...why not some option for adding the cp in user points too?? not only the top 20 global+HW..its going to be a good incentive for joining the competitions..

[K404]

 

The biggest problem with sandbagging is that certain people don't show anything until the last second. That was one of the biggest issues of F1OC ("5 minutes of excitement") and is still an 'issue' in pretty much all competitions. A system where you have to submit one score a week is a nice idea, but won't solve the excitement part of this issue: people would just start with uber-crap score, then submit a lesser crap score, then a normal score and in the last week their insane score.

 

What if there was a weighting attached to whoever was leading the board for the longest time? Start weak, then lose out.

 

Mix the weight of best score at end with most amount of time in the lead.

 

 

Another idea would be to grant points on a per-week basis, but then you just make the competition about who's able to submit the fastests rather than who is able to produce the best score.

 

 

That is unavoidable. No matter what competition requirement you set, there will be someone, somewhere who just happens to have the best kit to fit those requirements and can create a score at the drop of a hat. The best scores will come as people get to grips with the hardware and get "in the zone" Anyway.... the hardware requirements will be announced in advance, right? So people can source kit?

[Massman]

You need to add a little "KISS" to your ideas. The more complex a system, the less interesting it will be to follow and participate. That weighed system might seem closer to the 'true' ranking, but it's too complicated.

[knopflerbruce]

There are a couple of ways to prevent sand-bagging in competitions, but we must beware of demonising last-minute winning submissions. It's not because the winning submission was made in the last 5 seconds of the competition that the person who made that score doesn't deserve to win. If, in soccer, a team is trailing by 2-0 five minutes from the end but keeps pushing and eventually end up scoring 3 goals in the last couple of minutes, that's a deserved win in my opinion.

 

The biggest problem with sandbagging is that certain people don't show anything until the last second. That was one of the biggest issues of F1OC ("5 minutes of excitement") and is still an 'issue' in pretty much all competitions. A system where you have to submit one score a week is a nice idea, but won't solve the excitement part of this issue: people would just start with uber-crap score, then submit a lesser crap score, then a normal score and in the last week their insane score. Another idea would be to grant points on a per-week basis, but then you just make the competition about who's able to submit the fastests rather than who is able to produce the best score.

 

Next staff meeting, we'll look into the complexity of a competition format that excludes the 'worst teams' after a certain period. I can't recall the exact terminology, but it's like this:

 

- competition last 4 weeks

- 1st week: all can submit

- 2nd week: best 20 teams continue

- 3rd week: best 10 teams continue

- 4th week: best 5 compete for the stage win.

 

This way, you'll have to make sure your score is good enough 4x in one competition to be eligible for the next round.

 

I like that idea of team elimination. Most time as leader is also a good idea, would like to see how that works out in a competition.

 

One more note: stage details should be made public a while before the competition starts (1 month maybe?), so people don't have to spend the first 14 days tracking down the HW.

[Hondacity]

another idea:

 

"calling the shot"

 

a rule from billiards.

 

for last day of submission i will submit a score of 20000-3dmarks

 

anybody not calling their last shot(last submission day) will not be accepted.

 

this must be done 5 days before last day. hence every participant will not be surprised of the sandbag.

 

therefore 5 days before last day, we'll have a tally of sure guys submitting a score at the last day.

 

just an idea from the shower lol

[El Gappo]

There are a couple of ways to prevent sand-bagging in competitions, but we must beware of demonising last-minute winning submissions. It's not because the winning submission was made in the last 5 seconds of the competition that the person who made that score doesn't deserve to win. If, in soccer, a team is trailing by 2-0 five minutes from the end but keeps pushing and eventually end up scoring 3 goals in the last couple of minutes, that's a deserved win in my opinion.

 

The biggest problem with sandbagging is that certain people don't show anything until the last second. That was one of the biggest issues of F1OC ("5 minutes of excitement") and is still an 'issue' in pretty much all competitions. A system where you have to submit one score a week is a nice idea, but won't solve the excitement part of this issue: people would just start with uber-crap score, then submit a lesser crap score, then a normal score and in the last week their insane score. Another idea would be to grant points on a per-week basis, but then you just make the competition about who's able to submit the fastests rather than who is able to produce the best score.

 

Next staff meeting, we'll look into the complexity of a competition format that excludes the 'worst teams' after a certain period. I can't recall the exact terminology, but it's like this:

 

- competition last 4 weeks

- 1st week: all can submit

- 2nd week: best 20 teams continue

- 3rd week: best 10 teams continue

- 4th week: best 5 compete for the stage win.

 

This way, you'll have to make sure your score is good enough 4x in one competition to be eligible for the next round.

 

Nice idea but doesn't stop sandbagging in those rounds?

 

What if there was a weighting attached to whoever was leading the board for the longest time? Start weak, then lose out.

 

Mix the weight of best score at end with most amount of time in the lead.

 

 

 

 

That is unavoidable. No matter what competition requirement you set, there will be someone, somewhere who just happens to have the best kit to fit those requirements and can create a score at the drop of a hat. The best scores will come as people get to grips with the hardware and get "in the zone" Anyway.... the hardware requirements will be announced in advance, right? So people can source kit?

 

I like this Idea also.

 

We tried to kind of copy Nascar scoring for this "Race season" on ocn http://www.overclock.net/t/1016742/formula-jiggawatts-race-season-june-august

 

Points for first to submit, points for fastest in the first 48 hours, points per lap till the end of the month. Worked well for killing sandbagging and kept things interesting for the 3 months.

 

That and the qualifiers massman proposed would be awesome.

[Massman]

I like the idea of giving points in within rounds. Good idea.

 

Fwiw, you'll never fully get rid of sandbagging. Sandbagging is basically holding on to your best score until the very last moment. You can do this for any given time period. If your competition runs for 1 week, a sandbagger will submit on the last day at 23h50. If your competition runs for 1 hour, a sandbagger will submit at 55 minutes. All you can do is make it more bearable (like with nascar points or qualification periods).

[K404]

A weighting can be simple It doesn't have to be like 3DMark06......

 

Score is:

 

x % based on the ranking of best score at the end

 

y % based on the consistency of their ranking (but see below) over the course of the competition.

 

If onlookers can't follow that, there is a bigger problem to deal with....

 

 

BUT.... I do admit, "y %" needs care. It would be much easier to implement "average ranking" instead of ranking consistency..... e.g.... which is better? 75% of the competition in 3rd place, or 25% of the competition in 2nd?

 

Then.... how to calculate the average? A daily script ran at a set time? That would mean a daily sandbag, but over the course of a month-long competition, that's fairly interesting, things keep moving. A daily script ran randomly, but at least 18 hours apart... now... THAT.... would be interesting .... but starts to move away from simple.

[Massman]

If we can make something cool to follow and nice to play without using percentages, I prefer it over a complicated system. KISS-principle applies here.

[Luebke]

please make a mix of new and classic hardware (especialy non-amd-intel-nvidia hardware) to make it more diversified

 

and maybe some "new" (non-boints-)benches like aquanox or something completly new/different

[Hondacity]

classic hardware is obsolete here..

[Massman]

please make a mix of new and classic hardware (especialy non-amd-intel-nvidia hardware) to make it more diversified

 

and maybe some "new" (non-boints-)benches like aquanox or something completly new/different

 

I like the idea. Sure!

[knopflerbruce]

I wouldn't recommend doing experiments with new benchmarks if this is a somewhat serious competition. We don't want weird bugs in those.

[Luebke]

classic hardware is obsolete here..

 

why? should it only be for rich/sponsored benchers? old hardware can everybody buy for a few euros/dollars, a sandybridge with 5,4 ghz+ for gpu-benches is to expensive to me and many others... and without it´s senseless to participate. if you would have many teams participating you have to give them all a chance to get points...

the more different competitions, the more teams would participate.

[Massman]

The advantage of running our own competitions is that we don't have to care about how many people/teams participate. I prefer a fun 20-team competition over a boring 30-team competition anytime.

[Luebke]

yes, but imho is allways the same benches on allways the same hardware boring

[sumonpathak]

The advantage of running our own competitions is that we don't have to care about how many people/teams participate. I prefer a fun 20-team competition over a boring 30-team competition anytime.

 

this comment really scares me.....when did overclocking became a new toy only sport?