HWBOT 2020 Make it great again Thread:

[Leeghoofd]

Just an update on what we have in mind for the points ( Hardware and Global ones )

Maximum of 150 Globals no matter how popular the hardware is. If we have less than 100 subs, maximum will be 100 Globals, for less than 50 submissions maximum will be 50...

Less bigger jumps between the top 5 (10-5-3-2) --> some will not like this for sure.

Dumping the complex algorithms, the idea is to have fixed designated points for each spot.

Points no longer depending on how fast the nr 1 score is. Dumping the 70% cutoff.

From spot 21 each position will go down with 0.2-0.3 points. Minimum points will be locked at 2 points.

 

Hardware points difference will be less drastic as you can see, but we try to establish the same idea that each position corresponds with fixed points.

 

 

Goal is to make the point system more rewarding for a bigger amount of benchers. 

Lesser calculations are required by the database, thus resulting in a faster HWBot and becoming less error prone.

[Rauf]

I think this is way better than the current system. The current where you go from 150 for first place to 75 for third is just absurd.

What will the scaling be for hw points? Can't give high hw points for rankings without a decent amount of subs...

[TAGG]

So lowest you can get for globals is 50 points? Even for the categories that have <10 subs? that would be kinda cool not gonna lie, makes it actually worth benching those

Allso what will HW point ceilings look like?

[TerraRaptor]

8 hours ago, Leeghoofd said:

Just an update on what we have in mind for the points ( Hardware and Global ones )

The question is - any implementation schedule for main site and @uat?

[Leeghoofd]

10 hours ago, Rauf said:

I think this is way better than the current system. The current where you go from 150 for first place to 75 for third is just absurd.

What will the scaling be for hw points? Can't give high hw points for rankings without a decent amount of subs...

8 hours ago, TAGG said:

So lowest you can get for globals is 50 points? Even for the categories that have <10 subs? that would be kinda cool not gonna lie, makes it actually worth benching those

Allso what will HW point ceilings look like?

Max 50 hardware points for hardware with over 100 submissions, max 25 sub 100. Does that sound feasible ? Maybe we can add a 3rd less than than 50 submissions : 15 points

5 hours ago, TerraRaptor said:

The question is - any implementation schedule for main site and @uat?

That's the big question. Tim has not touched any algorithm or such , he's designing a new front end, call it a wrapper that only requires the ranking from the current old version and will apply the new point algorithm. Front page is already done, now he's looking into the different categories. Its not easy for him to understand as it mostly does not seem logic to him regarding the things we need to have displayed yes or no. 

If he shows it someday on his local system I'll try to take to some screenies

We hope to have it done in Q1

[yosarianilives]

I'm gonna get so many team and hwpts compared to now, I love it! 

[TAGG]

5 hours ago, Leeghoofd said:

Max 50 hardware points for hardware with over 100 submissions, max 25 sub 100. Does that sound feasible ? Maybe we can add a 3rd less than than 50 submissions : 15 points

Ok, so HW points are going to be hardcore i guess right now i have some 48 and 49 pointers in rankings with like 90-99 subs Allso from that i can take away that all benches are worth the same? so no more 67 point 32M vs. 41 point pifast and so on?

[Leeghoofd]

it is my initial proposal and was the easiest to explain to the programmer, but we can shift it to maybe at least 75 submissions is maximum 50 HW points. Less then 75 is 25 HW points,....

Yes all benches will probably be treated in the same way, it will be win some, loose some boints.

I can just propose, how Tim has to program this, I have no clue...

[Papusan]

7 hours ago, Leeghoofd said:

Max 50 hardware points for hardware with over 100 submissions, max 25 sub 100. Does that sound feasible ? Maybe we can add a 3rd less than than 50 submissions : 15 points

Regarding the 3rd option. Less than than 50 submissions : 15 points means the HW point system will be worse for not so popular HW. Even 25 point cap at 50 subs will do that.

Edited November 9, 2021 by Papusan

[Leeghoofd]

6 minutes ago, Papusan said:

Regarding the 3rd option. Less than than 50 submissions : 15 points means the HW point system will be worse for not so popular HW. 

We can go many directions with this: Max 75 hardware points for hardware with over 100 submissions, max 50 sub 100,  less than than 50 submissions : 25 points.

It all depends how the programmer can program things

[Splave]

How about anti-sandbag measures. If you sub way higher than 2nd place the people behind you get dropped by some percentage. 

[TAGG]

If possible i'd suggest 50 max and 5 points less per 10 subs under 100, so 90-99 would be 45, 80-89 40 and so on for a absolute floor of 10 (under 30 subs)?

Or maybe set the floor higher if you guys want more points for less popular stuff?

Edited November 9, 2021 by TAGG

[Papusan]

2 hours ago, Leeghoofd said:

We can go many directions with this: Max 75 hardware points for hardware with over 100 submissions, max 50 sub 100,  less than than 50 submissions : 25 points.

It all depends how the programmer can program things

Yep, probably several ways do do it. But as I said, getting less point than before ain't the best way to get more people on the bot. And laptops become more popular and they come in all sorts of colors.

1 hour ago, TAGG said:

Or maybe set the floor higher if you guys want more points for less popular stuff?

Put it this way. With more popular HW as Alder lake desktop chips as 12900K you get more points now due big.Little. People with less popular HW/older stuff/Amd will come longer and longer far behind?

Edited November 9, 2021 by Papusan

[Mr.Scott]

By the time any of this happens the old guys will be dead and gone. 

It's been what, almost 2 years since this thread started, and what has really been done?

Talk is cheap, and patience is in short supply.

 

[Matsglobetrotter]

Am not sure what the new formulas would mean but would agree with most boint calculations that actually give some decent points for overclocks even on air. Thus points that recognize that people actually do overclock systems to spur them on to do more. I mean having 30 CPUs of various platforms and simply getting 10 boints in total among all because those we either not popular or where 5 individuals have LN results which are so far from what can be done by anyone in rookie, novice, enthusiast and even apprentice does not inspire people to actually start overclocking let alone staying with it. 

People also should be encouraged more to OC older hardware so that its not just the 3000 USD investments that count. Clearly the hobby will spread more if people can buy older hardware and get serious boint (not only as extreme player.) 

This especially with the last 2 years super explosion of prices of new hardware.  I could buy it but simply consider its no longer relevant to feed the serious hardware overpricing that seems to be everywhere now. Considering the little amount of people who actually use the last 2 years hardware on the bot i can see that im not alone in this consideration.

[Leeghoofd]

5 hours ago, Splave said:

How about anti-sandbag measures. If you sub way higher than 2nd place the people behind you get dropped by some percentage. 

Roman has an idea for competitions on this one Allen, but one step at a time.

 

 

2 hours ago, Mr.Scott said:

By the time any of this happens the old guys will be dead and gone. 

It's been what, almost 2 years since this thread started, and what has really been done?

Talk is cheap, and patience is in short supply.

 

If you check the front page and actually read it, you can see we only got access to the code end of last year, thanks to the ultra slow paperwork and the Covid shitshow... Didn't we mention numerous times already that 2021 would be a transition year as the programmer can only work during the weekends on the code... I think Tim sorted already tons of broken code, removed deadlocks, enabling most broken features and recoded them to work better than before. The Bot was a limping neglected beast and is now more stable than it was the last 3-5 years.

You are probably the expert in coding and could have sorted it all within a few weeks, too bad for all of us we made this paramount mistake of not hiring you. I also know a few that would start a new OC Dbase from scratch, haven't seen much of that either the last decade. Mat can and probably will do a version with BM.

It seems clear to me you only listen to those voices in your head, lighten up and exorcise them demons. Eventually it might even make you smile....

[Mr.Scott]

18 minutes ago, Leeghoofd said:

 

 

 

If you check the front page and actually read it, you can see we only got access to the code end of last year, thanks to the ultra slow paperwork and the Covid shitshow... Didn't we mention numerous times already that 2021 would be a transition year as the programmer can only work during the weekends on the code... I think Tim sorted already tons of broken code, removed deadlocks, enabling most broken features and recoded them to work better than before. The Bot was a limping neglected beast and is now more stable than it was the last 3-5 years.

You are probably the expert in coding and could have sorted it all within a few weeks, too bad for all of us we made this paramount mistake of not hiring you. I also know a few that would start a new OC Dbase from scratch, haven't seen much of that either the last decade. Mat can and probably will do a version with BM.

It seems clear to me you only listen to those voices in your head, lighten up and exorcise them demons. Eventually it might even make you smile....

I am an impatient bastard. That's all. 

[ludek]

These point steps (>100; >50; >0): I suggest to define them linear. I imagine there will be some devices with points close to the next threshold. You upload 49th score and pray none of the rest 48 scores will not be deleted because of not meet the rules. Also, I bet someone will come up with an idea to create an account, upload a worse score, to raise his score, illegally of course. So the linear scalling is a less frustrating solution.

[TerraRaptor]

14 hours ago, Splave said:

How about anti-sandbag measures.

Domination points that will add to hw (or global?) points - +x (x as a function of hw points) point for every month you hold first place.

Edited November 10, 2021 by TerraRaptor

[Leeghoofd]

34 minutes ago, TerraRaptor said:

Domination points that will add to hw (or global?) points - +x (x as a function of hw points) point for every month you hold first place.

This is something Roman is working on for competitions, the one that holds the top score for a specific period will get rewarded with extra points.

[denvys5]

1 hour ago, TerraRaptor said:

Domination points that will add to hw (or global?) points - +x (x as a function of hw points) point for every month you hold first place.

That may indirectly encourage backups. Coz if implemented system counts only your latest sub, there is no point to improve it ever, unless it is beaten. You may rerun benchmark in the future and get a better score, but you are encouraged to keep that score as a backup by the system.

[GeorgeStorm]

16 hours ago, Splave said:

How about anti-sandbag measures. If you sub way higher than 2nd place the people behind you get dropped by some percentage. 

Isn't the whole point of this update to move away from that concept? Who cares how much better you are than 2nd etc (since that's the basis for the current system)?

(I personally liked the current system, actually rewarding people improving their scores even if they don't improve their placing, worth putting the effort going from 3rd to 1st etc.)

[TerraRaptor]

13 minutes ago, GeorgeStorm said:

(I personally liked the current system, actually rewarding people improving their scores even if they don't improve their placing, worth putting the effort going from 3rd to 1st etc.)

Two sides of coin. Current system really motivates me to fight for #1 but once I understand that I can't make it (no matter how much effort and skills I pay) I won't even care to save screenshot. Like, the following arguments come into my mind with current system: why should I make a category more popular, thus bringing more points to competitors and gaining 0 myself. I will share observations etc on "not so successful oc" with friends but no submissions to hwbot will follow.

 

 

[Rauf]

I think I have mentioned it before, but I like the idea of just making the rankings % of top score. First place is 100%, and then every following score just gets the percentage of the top score. It's extremely simple to code and you never have to recalculate entire rankings, except when a new top score is posted. Should make for a much lower server load. You can also reduce your opponents points by uploading a better top score.

For the unpopular rankings there could be dividers, like 75% if less than 100 subs, 50% if less than 50 etc. HW points could also be handled by dividers it you want globals to be worth more like it is today. Top score on HW points would be 100 divided by 2 for example. And then also add additional divider if less than X subs.

[eviledeath]

14 hours ago, Matsglobetrotter said:

Je ne suis pas sûr de ce que signifieraient les nouvelles formules, mais je serais d'accord avec la plupart des calculs précis qui donnent en fait des points décents pour les overclocks, même à l'antenne. Ainsi, des points qui reconnaissent que les gens utilisent réellement des systèmes d'overclocking pour les inciter à en faire plus. Je veux dire avoir 30 processeurs de différentes plates-formes et simplement obtenir 10 points au total parmi tous parce que ceux qui ne sont pas populaires ou où 5 personnes ont des résultats LN qui sont si loin de ce qui peut être fait par n'importe qui en tant que débutant, novice, passionné et même apprenti n'incite pas les gens à commencer à overclocker et encore moins à rester avec. 

Les gens devraient également être davantage encouragés à OC sur du matériel plus ancien afin que ce ne soient pas seulement les investissements de 3000 USD qui comptent. Il est clair que le passe-temps se répandra davantage si les gens peuvent acheter du matériel plus ancien et devenir sérieux (pas seulement en tant que joueur extrême.) 

Ceci surtout avec les 2 dernières années super explosion des prix du nouveau matériel. Je pourrais l'acheter mais simplement considérer qu'il n'est plus pertinent pour alimenter la grave surévaluation du matériel qui semble être partout maintenant. Compte tenu du petit nombre de personnes qui utilisent réellement le matériel des 2 dernières années sur le bot, je peux voir que je ne suis pas le seul dans cette considération.

I haven't been on the bot for a very long time, but this year I was quite active with old equipment. I am primarily passionate about retro.
So I agree with your opinion a little on the question.
I find it very regrettable that the material which is of no interest and which is not top of the range if it can be valued on hwbot.
This would make it possible to attract more people to the platform I believe with a little exotic equipment (I love working on common stuff, IGP style or other old entry-level GPU.).