Little variation in hwboints requestion ;)

[Predator]

Hello gents ;) , i was having an idea that i would like to shre with you guys

i know how the hwboints are gained and i think it's a very very good equation based , so i wouldn't change a single thing about that , i know the hwboints in a team are the total hwboints achieved by their whole member list and that's ok too ;) , altough , i would like to suggest the possibility of another way for a team to get points , and that would be based in the Team Benchmark Rankings

my idea consists in adding points to a team by the posistion Ranked in each Benchmark Category , these points would count as a team effort and not would be individual hwboints , in example , let's say the 3DMark 05 Ranking

the current would be like this

http://img470.imageshack.us/img470/7786/teamrankingson0.jpg (http://imageshack.us)

then here it could be like 50 points for the first , 40 for the 2nd , 30 for the 3rd , and so on , just like the top 5 global results in the hwboints calculation , but for teams


why i suggest this? , mainly 2 reasons

1) that would encourage the top ten rankings in the benchmarks , in fact that is what is monitored in the forums and till the hwboints ranking showed up this was the only and first feature of hwbot


2) this would give more importance to the "hardest" overclocks , i mean , it's real funny to bench all the "old-hard" and get 2 ponits here and there with those graphic cards you had forgotten in a draw , but i think the real competition should be more focused in the top scores and performance , this feature IMHO would bring even more tight competition to the scene and would put the Top Ten Team Benchmark Rankings in the rule it deserves (again in my opinion ;) )



so , basing in the weight of every Team Benchmark ranking , the same equation as the hwboints calculation could be used , a "x" number of points by the posisiotn in the Ranking * the application weight , that would add points to the Team , only as team Points ;)


i hope you guys can at least think about it ;)

you're doing an excellent work there ;)


Predator , OCX Staff

[jmke]

so you want to make the large teams, which have high scores in TOP 10, to score even higher as before... ?

ugh :/ no thanks.

[aspstein]

No im behind jmke on this one, no need to make the gap between teams bigger. No fun in that....

[richba5tard]

I don't think it's a bad idea, these 'bonus' points. Like you said, a team who manages to get the top team average deserves some additional points.

I wouldn't give bonus points to the team on the 2nd or 3rd place though, than the rules would get to complicated for those who want to know how points are awarded. Just sth like 1st place in team average ranking = +50 hwboints for the team.

[aspstein]

But look into the ranking RB. OCX and Xtremesystems totaly owns 1 and 2 places. This would get maybe 3-4 other teams 50 points. And OCX and XS couple of hundreds.

Im against :P

[jmke]

RB, it's EITHER hwboints, OR benchmark ranking, giving HWBoints AND benchmarking ranking is giving top scores DOUBLE credit if you are in a LARGE GOOD team, while a HIGH score in a small team gets only half score.

competition is pretty good as it is now imho; maybe we can somehow create "divisions" depending on members/top results to create more hwboint "fights" outside the TOP 5

[aspstein]

Or maybee you should give every team except number 1 and 2 50 points bonus in every benchmark. That would even out the teamrankings :P

Divisions is not a bad idea. From maybee numbers of users in the team. Divided maybee something like this?

200- ++ users
100-200 users
50-100 users
20-50 users
10-20 users
0-10 users

[richba5tard]

Hmmm, both have a good point. :) I like the idea of bonus hwboints to top teams, but you're right that it would just separate them from the sub-top3 teams even more, making competition less interesting.


What I really, really would like to see in the hwboint algorithm is more points for ppl who have a top 10 score with a very popular hardware item.

eg. for an uber overclocked 7900GT ranked 5th out of 500 cards, you get like 1point. : (

[aspstein]

Isnt the algorithm working that way now? Or is it just top3 scores with a popular hw item that gets bonuspoints now?

[richba5tard]

Yes, I was unclear. You do get more points if more ppl have the same hardware item, but the effect is (in my opinion) limited.

hwboints = global points + hardware points

* global points (top 5): (50,40,35,30,25 * application weight) * 0.6
* global points (other): ((-20/145) * your rank + (600/29) * application weight) * 0.6
* hardware points (top 5): (5,4…1 * hardware weight) * 2
* application weigth: 0.5 (<500 participants) - 2 (2000 or more participants)
* hardware weigth: 0.2 (1 participant) - 1 (100 or more participants)
I would change the hardware weight. At the number of participants is capped at 100 ppl, more does no longer affect the ranking anymore, which makes the hwpoints fall in the range of 2 - 0.4 (5 ppl) to 10 - 2 (100ppl or more).

I feel that 10 (1st) - 2 (5th) points is too few when you look at a popular hw item hundreds of people use. Capping it at 200ppl (20 - 4) and increasing it to the top 10 instead of top 5 looks like a better reward for those who can't afford expensive hardware, but can overclock really well.

[aspstein]

Yeah I totaly agree! Would be really neat

[jmke]

why is there a cap:) no caps on that, just let it scale, if there are 1000 7900GT scores, and you hold number 1, you should be awarded bigtime

[richba5tard]

I think mtzki was against uncapped hardware points... and as he is the creator of the hwboint algorithm I tend to give him the benefit of the doubt. :)

[jmke]

put the cap higher then... 100 was good for small community... 1000.. more HWBot style ;)

[richba5tard]

Yeah, i just got a mail from somebody asking why he doesn't receive hwboints.

With his E6600 running at a nice 4Ghz, he's 232nd fastest E6600. Just 227 people to pass to get in the top 5 and receive hwboints, good luck buddy! :D

http://www.hwbot.org/result.do?resultId=573987

[jmke]

remove the cap, top 10 should receive points, as well, more HW in that categorie, points for top 20 etc etc;

[Predator]

@jmke , my idea was not to make the larges teams to get more gap or diference wth the rest of teams , but giving more credit to the top ten scores than the old hard benchs , i think get #1 in Super Pi IE as a team has his merit , at least what RB suggested is more accurate than currently , a bonus for the 1st would make the battle much more interesting IMO , if not the top tens will got lost vs the hwboints rankings , it's great and funny as i said to bench old hard , but IMO is "unfair" that your team gets 2k more points than other only cause their members has the whole collect of radeons since 9200 , it should be a very much larger difference between getting 1st with a X1900XTX than a Matrox


there are way too much of different hard pieces , so the current hwboints benefits much more to a team with a lot of members and a bazillion of different parts than a team that gets a nice 1 spot at PCMark or wPrime ;)

[jmke]

that gazillion OC team deserves merit for their effort, and should be able to outscore other teams, even if they don't have the latest gear; hwboints let ANY man/woman with ANY hardware get boints, that's the beauty part; benching with the latest gear is not something everybody has the possibility to do.

[Predator]

that gazillion OC team deserves merit for their effort, and should be able to outscore other teams, even if they don't have the latest gear; hwboints let ANY man/woman with ANY hardware get boints, that's the beauty part; benching with the latest gear is not something everybody has the possibility to do.
and i agree on that , but you might not catch my point as i wanted to mean (sure due to my crappy English tough)

i know everyone cannot afford to bench the ultimate hardware , but if you think about it that has his merit too , at least and i tell you my case and i'm sure it's the same with the most of OCers around the globe , i'm not rich , i have average salary , average work , and i am not rich!! , so i spent my bucks in my favorite hobby , and maybe a few more than needed to help the team or group i'm devoted to , so i could be saving 2k instead buying a pair of 8800gtx and a cascade and take my GF for a trip , so i'm either risking my bucks punishing my hardware to the limits and my own life cause you know GF's and money you spend :mad:

anyway that's not the point , some of you guys mentioned that more points for the "top" teams would make the Gap wider with th rest of teams , but think about this , does it matter?? , i mean , if one team is spotted in 1st and 2nd positions of the becnhmarks rankings , that's obviously cause their members have very good results and that causes more points , so the team that is placed at #10 in the hwboints ranking will hardly have got the chance to catch XtremeSystems , now figure that a bunch of the best OCers change teams and go to that #10 team , then they can catch XS , what i try to mean is that those bonus points i suggest are not goping to set a difference , what's the matter if the 1st team is 1000 points better than you or 1200? , the difference between teams will remain the same , BUT , if you want to see a titanic fight between XS and OCX , then allow those bonus points , cause when we get close again to XS those points will make a very fierce battle at the top , so the top ten will gain interest , hwbot will gain interest . and the battle at the top of the list will be more exciting than ever


can you imagine XS and OCX close in points and the legend overclockers as Kinpin and OPPAINTER fighting in the top scores to cut 100 points? , cause that would be 50 points that a team gains plus 50 that the rival loose

and the rest of teams will remain the same , those additional points are related and in proportion to top hwboints scorers , but will give more intense to the competition IMO and more credit to the Top Overclocks towards the old-hard benchs

that's my 2 cents there ;)

saludos1

[aspstein]

Yeah I totally understand the point of getting a battle in the top. But wouldnt it be more fun if there was like 5-10 teams fighting about the top. I dont see the point of getting a even bigger gap between XS, OCX and the rest of the teams. OCX teammembers needs to start benching 24/7 then we will se a real fight again :D

And wee all know that diskusjon.no will rule them all... ehh.... :rolleyes:

[Predator]

Yeah I totally understand the point of getting a battle in the top. But wouldnt it be more fun if there was like 5-10 teams fighting about the top. I dont see the point of getting a even bigger gap between XS, OCX and the rest of the teams. OCX teammembers needs to start benching 24/7 then we will se a real fight again :D

And wee all know that diskusjon.no will rule them all... ehh.... :rolleyes:
We will get close to XS again i can guarantee you that ;) , and yep it would be more fun if 5 teams would be fighting for the #1 , but what can be done for that?

plus , and sorry if i repeat myself too much in this xD , that would bring the deserved credit the top ten benchmark rankings should have IMO


thanks for understanding my POV mate :)

saludos!!

[jmke]

HWboints and individual results ranking should not be mixed imho, it will make it extremely hard to catch TOP 5 teams, because they have a bonus because of higher results/member ratio.

[mtzki]

I designed the current algorithm and there is a revision coming for it. An initial representation should be ready pretty soon, but the final implementation will have to wait until rb comes back from Australia. I would sincerely hope you will wait for my suggestion before speculating much further.

I would agree with many of Predator's comments above. However as comes to the OP, i think it's better that the boints originate from one source only. It's easier to get the big picture and maintain the balance that way.

The new algorithm suggestion will have the following main changes:

1) boints will be awarded to a larger group, both globally and in hw categories. From 150/5 positions getting boints we might well go to a situation where all positions get at least a little, both in global and hw categories.

2) Total team score comes only from n top benchers of the team.

3) Small adjustments to the awarded boint levels...only small.

From the above changes, 1) makes the system nicer for beginning benchers and encourages trying. 2) removes 'the big team bias' the current algorithm has been accused for. However, n will be considerably larger than 10. I'll make a thread here on the coming changes in a few weeks and ask about opinions especially on 3).

[jmke]

"Total team score comes only from n top benchers of the team." sorry, but if a large team can score high, each member attributes to the team ranking, imho, that change 2) is bad...

[mtzki]

n will be pretty large...100 or so. If sb doesn't belong to the top n overall of his team, it should only motivate to try to get there. With the 1) change, 2) becomes a must as otherwise the big team bias gets a lot worse than it is now.

[aspstein]

But teams with only 99 members then, wouldnt be ranked at all in the different benchmarks??
Like now if the team has 9 scores in a benchmark no ranking in that benchmark.

[mtzki]

No no...we are talking about the overall competition for the teams only. The score would be computed by adding together the boints from the top n overall benchers of the team (or less if the team doesn't have that many members). So as long as the team has a single member with boints, it will get an overall score.

[aspstein]

Ahh very good, for the way it is now all teams have to have 10 members to get an ranking in each benchmark. And that is rather useless I think...

[mtzki]

n would only affect the team overall competition, 10 members still needed for single rankings.

[Predator]

I would sincerely hope you will wait for my suggestion before speculating much further.
absolutly sir you can count on it ;)

thanks a lot for understanding , and makes me quite happy you agree some of my ideas :)

your new suggestions sounds pretty good aswell , looking forward to see the "new" algorithm revision


saludos!!