richba5tard Posted January 26, 2018 Share Posted January 26, 2018 First suggested change for rev7.1 is to give minimum points to your best submissions, even if they are below the 90/50/75% cutoff compared to the best score. Min = minimum points if score above cutoff MinBC = minimum points if score below cutoff Reasoning: - people who are cooling with air/water now often receive 0 points, even if their submission was good air/watercooling - if there is a very large gap between wr and the others, very few people receive points - less incentive to submit mediocre overclocking results Suggestion: - min 2/2/0.5 bp/gl/hw points if > cutoff - min 1/1/0.2 points for any participation in the ranking < cutoff Quote Link to comment Share on other sites More sharing options...
Members GeorgeStorm Posted January 26, 2018 Members Share Posted January 26, 2018 What about the idea both Strunkenbold and I suggested (along with possibly others), with having both a % cutoff and a min. number of point earning submissions cutoff (which may be based on a % of total submissions for that benchmark). Which would help avoid gpugpi 1b issues with the titan v being so far ahead? Also I'm not sure what this means, so will people who previously would have got 0 would now get 0.2 just for subbing any score? Quote Link to comment Share on other sites More sharing options...
richba5tard Posted January 26, 2018 Author Share Posted January 26, 2018 What about the idea both Strunkenbold and I suggested (along with possibly others), with having both a % cutoff and a min. number of point earning submissions cutoff (which may be based on a % of total submissions for that benchmark). Also interesting approach, although harder to properly implement and explain. Which would help avoid gpugpi 1b issues with the titan v being so far ahead? Also I'm not sure what this means, so will people who previously would have got 0 would now get 0.2 just for subbing any score? Correct, but the app multiplier still stands. So apps which don't award full points, get less min points. Quote Link to comment Share on other sites More sharing options...
Members GeorgeStorm Posted January 26, 2018 Members Share Posted January 26, 2018 Also interesting approach, although harder to properly implement and explain. Could either do it as an OR system, so either down to 75% or x% of submissions whichever comes first, or could need both to be TRUE, just an idea Correct, but the app multiplier still stands. So apps which don't award full points, get less min points. Fair enough, interesting. Quote Link to comment Share on other sites More sharing options...
richba5tard Posted January 26, 2018 Author Share Posted January 26, 2018 The algorithm is complex enough already. Stop making suggestions. j/k Quote Link to comment Share on other sites More sharing options...
Members GeorgeStorm Posted January 26, 2018 Members Share Posted January 26, 2018 The algorithm is complex enough already. Stop making suggestions. j/k Haha sorry, just trying to help with dealing with some of the anger you are/hwbot is receiving Quote Link to comment Share on other sites More sharing options...
Aleslammer Posted January 26, 2018 Share Posted January 26, 2018 Thinking adjustments should be made based on CPU family and how well they overclock. Going through the more popular 775, 2 core, points are more or less dead in the SS cooling range at present in the categories I checked. AM3 I won't say the same thing out right but do have some fairly good water sitting in second place with no points to a LN2 sub. Quote Link to comment Share on other sites More sharing options...
Splave Posted January 26, 2018 Share Posted January 26, 2018 I think we are going to need a PHD in mathematics to understand what to bench soon. Complexity level over 9000 Quote Link to comment Share on other sites More sharing options...
Crew Don_Dan Posted January 26, 2018 Crew Share Posted January 26, 2018 I'm with Allen, we should be very careful before making any adjustments to avoid making the point algorithm even more complicated than it already is. Quote Link to comment Share on other sites More sharing options...
Aleslammer Posted January 26, 2018 Share Posted January 26, 2018 Personally don't care. I'm looking at my own subs and with a one size fits all getting some pretty odd results, maybe it was intended? Quote Link to comment Share on other sites More sharing options...
mickulty Posted January 27, 2018 Share Posted January 27, 2018 Great idea IMO. It won't make any difference at all to the rankings of established benchers, but for the noobiest of noobs it gives them a reason to submit scores, even crap ones, to climb from 11235th to 3142nd or whatever. And that gets them hooked, mwahahahaha... Quote Link to comment Share on other sites More sharing options...
Alan_Alberino Posted January 27, 2018 Share Posted January 27, 2018 What about the idea of changing the cutoff percentages that was mentioned in the other thread? Like I said in the other thread, 60% cutoff for globals and hardware would be great... In many pieces of hardware if you dont have at least DICE you get 0 points with 75% cutoff, having a lower cutoff would give possibility to have some extra HW points for people who cant afford last hardware and get globals. Also with the 50% globals cutoff, some near stock scores give globals which make no sense. Also, I think something should be done with locked hardware... Locked notebooks and XMP XTU on locked CPUs are giving way much hardware points without even overclocking anything, just clicking run... Quote Link to comment Share on other sites More sharing options...
Rauf Posted January 27, 2018 Share Posted January 27, 2018 (edited) The whole point of having points is to make it competitive. This does nothing towards that, and is impossible for noobs to understand (I don't understand it either). Also, many benchmarks don't give the right amount of points compared to the pdf. Seems there are different constants used for some benchmarks. Like 3D03 is OP and some others, while some are lower than they should be, like 01 and aqua. Why? Is it supposed to be this way? Edited January 27, 2018 by Rauf Quote Link to comment Share on other sites More sharing options...
Alan_Alberino Posted January 27, 2018 Share Posted January 27, 2018 I'm going to propose something that many people won't like: Scores with under 10% OC? Not allowed. Locked notebooks? Not allowed. Here are some example scores to show why I propose this: #200 on a locked notebook (1st scores are also @ stock) = 26.6 pts http://hwbot.org/submission/2893251_selbstzünder_xtu_core_i5_4200u_304_marks #200 i3-6100 XTU with XMP mems and stock CPU (Just booted PC @ default and clicked run) = 22.7 pts http://hwbot.org/submission/3643246_humberto.oliveira.92_xtu_core_i3_6100_639_marks I'm not talking about #1 scores but #200 score... HWBOT is about overclocking, and there is no overclock in this scores... Quote Link to comment Share on other sites More sharing options...
havli Posted January 27, 2018 Share Posted January 27, 2018 That is like deleting half of the scores... not going to happen. Tweaking the OS and benchmarks is also part of the game and requires some skill, even if overclocking is not possible. Quote Link to comment Share on other sites More sharing options...
Alan_Alberino Posted January 27, 2018 Share Posted January 27, 2018 That is like deleting half of the scores... not going to happen. Tweaking the OS and benchmarks is also part of the game and requires some skill, even if overclocking is not possible. I know, but only, exagerating, 5% do that, the other 95% just boot and clicks run, and get lot more points that scores that really had effort... Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted January 27, 2018 Share Posted January 27, 2018 I know, but only, exagerating, 5% do that, the other 95% just boot and clicks run, and get lot more points that scores that really had effort... You mean XTU. Quote Link to comment Share on other sites More sharing options...
alexmaj467 Posted January 27, 2018 Share Posted January 27, 2018 (edited) Sorry for my English. But I see the problem now on the other side. With the expectation that none of the processors in which the maximum acceleration of the ceiling does not exceed 10 - 70%. s 939 - s754 - s1155 non-K ...... The problem with those processors who are over 80% and 200% of the nominal value. You can make 2 formulas ? for the ones who are poorly looking to acceleration what we have now and for others to access a greater percentage of 25% ? If it is possible that we will collect a list of those processors which require changes to the calculation formula. will not work one formula is valid when the ceiling of dispersal among different processor families. Plus results 0 points replace 0.1 No problem Athlon (Venice, s939) Athlon (Newcastle, s754) Celeron (Willamette 478) Core i5 2400 Core i5 4690K problem Celeron D 336 (2,80GHz, s775) Celeron D 325J (2,533GHz, s775) Celeron D 347 775 Athlon II X2 250 Phenom II X2 555 BE Edited January 27, 2018 by alexmaj467 Quote Link to comment Share on other sites More sharing options...
yosarianilives Posted January 28, 2018 Share Posted January 28, 2018 While I think that there are too many "lame OCs" on the bot, I also think that deleting all 10% as a one size fits all fix for that isn't the way to go. For example on locked lga 1155 there is a lot of effort just to hit 10% oc, on g470 I had to bin 4 chips, and it definitely is a real OC not like the stereotypical 1151 locked xtu "oc". Binning is important on these cpus, also you're up against the limits of the IMC and questions like "should I run 109.8 bclk but be forced to use a lower mem divider or 109 and use the higher divider?" come up. I think that if such a cutoff were added it would likely be based only on core percentage of OC, which doesn't take into account pushing the crap out of mem or in the case of certain "cpu" benchmarks the gpu. Similarly on gpu there are many benches where cpu is just as important as gpu so you may not push the gpu, or in the case of intel gma, may not even be able to at all push the gpu. In that case it is about cpu and mem OC. However that would show up as a "lame" oc even though it took a lot of effort and was pushed to the max. If something is implemented like this it has to be cpu per cpu basis. For example on i3 6100 less than 2.9% gtfo while on lga 775 less than 10% u suck. Quote Link to comment Share on other sites More sharing options...
Alan_Alberino Posted January 28, 2018 Share Posted January 28, 2018 You mean XTU. Any bench, you see most notebooks and maybe there is a 15 or 20 points Cinebench R15 ran at stock, thats not overclocking Quote Link to comment Share on other sites More sharing options...
Crew Strunkenbold Posted January 29, 2018 Crew Share Posted January 29, 2018 Well even with this minimum points feature, the problem of "stealing" points for anyone else once you upload your 130% score is still there. I think people will find this very demotivating. I voted against this as it solves nothing. Being 4th place in ranking of 20, worth 0,2 pts because the man 1st did something super spectacular still feels bad. The rankings in rev7 are way too dynamic. In the past you lost maybe 1-2 points if you went some places down in a ranking. Today theres the threat you loose all your points. You actually just can solve this with bringing a static element like the proposed fixed points for places 1 to 50. A quick fix would be more like setting the cap to 50% for hardware rankings. Global points cap must be set a little higher like 60-70%, because my personal feeling is that a 4,4Ghz CPU-Z verification shouldnt be worth obtaining global points. Quote Link to comment Share on other sites More sharing options...
alexmaj467 Posted January 30, 2018 Share Posted January 30, 2018 Well even with this minimum points feature, the problem of "stealing" points for anyone else once you upload your 130% score is still there. I think people will find this very demotivating.I voted against this as it solves nothing. Being 4th place in ranking of 20, worth 0,2 pts because the man 1st did something super spectacular still feels bad. Complete agreement. Too bad we can't re-vote. It would be better not formulas wrote and the fact that all who have 0 points get 0.2. This was clearer. I am for a solution to the problem but not this way , some hardware does not require the solution,other of a part requires. Quote Link to comment Share on other sites More sharing options...
richba5tard Posted January 30, 2018 Author Share Posted January 30, 2018 "stealing" points because you got a spectacular #1 spot is a different discussion than awarding minimal points beyond cutoff guys, please be on topic. Quote Link to comment Share on other sites More sharing options...
Crew Strunkenbold Posted January 30, 2018 Crew Share Posted January 30, 2018 Sorry "stealing" maybe sounded a little harsh. Anyway I simply wanted to share my opinion why I voted against this. We probably need another suggestion thread. Quote Link to comment Share on other sites More sharing options...
richba5tard Posted February 15, 2018 Author Share Posted February 15, 2018 Ok, this will be included in rev7.1 as 76.19% voted in favor. ETA early march. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.