Guest TonyBombassolo Posted February 11, 2023 Share Posted February 11, 2023 (edited) Any chance we could as a site start standardizing the submission rules since we are changing submission policies? Allowing LOD/Tess manipulation on these older tests and allowing incomplete screenshots that cutoff any aspect of the desktop are inconsistent with other policies about OC'ing imo, specifically when those submissions dont even pass the muster of the benchmarks in question. Are we really serious as a site about all the tests being fair or are a few big wigs just mad they aren't on top of a few benchmarks and are correspondingly bullying staff into changing just those? Edited February 11, 2023 by TonyBombassolo Link to comment Share on other sites More sharing options...
Mr.Scott Posted February 11, 2023 Share Posted February 11, 2023 I hardly think any staff here are being bullied. 1 Link to comment Share on other sites More sharing options...
saltycroissant Posted February 12, 2023 Share Posted February 12, 2023 9 hours ago, TonyBombassolo said: Any chance we could as a site start standardizing the submission rules since we are changing submission policies? Allowing LOD/Tess manipulation on these older tests and allowing incomplete screenshots that cutoff any aspect of the desktop are inconsistent with other policies about OC'ing imo, specifically when those submissions dont even pass the muster of the benchmarks in question. Are we really serious as a site about all the tests being fair or are a few big wigs just mad they aren't on top of a few benchmarks and are correspondingly bullying staff into changing just those? Rules are changed because some ppl think its 100% normal that adding 25mhz to the memory of a gpu will boost its performance by 10%.... For the allowing incomplete screenshot, nobody should be covering settings or important part of the benchmark, if you see this pls report those scores. 1 Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 12, 2023 Share Posted February 12, 2023 (edited) 1 hour ago, saltycroissant said: Rules are changed because some ppl think its 100% normal that adding 25mhz to the memory of a gpu will boost its performance by 10%.... For the allowing incomplete screenshot, nobody should be covering settings or important part of the benchmark, if you see this pls report those scores. The 3rd party software allows the score, and reports it as valid, and yet we want to not allow those, while at the same time allowing actual invalid scores from the same software? Thats inconsistent logic and the only reason people use that type of inconsistent logic is because it benefits one group or the other. Allowing actually invalid scores from the same software people are throwing a fit about valid scores on screams horseshit to me and I cant see any evidence you could provide otherwise. We either need to be consistent about rules or stop acting like they are anything other than arbitrary. EDIT: Additonally not all benchmark screenshots requirements are the same. Another inconsistent verification process that is around "because thats how it always has been" which is a phrase that is the enemy of progress. Edited February 12, 2023 by TonyBombassolo Link to comment Share on other sites More sharing options...
saltycroissant Posted February 12, 2023 Share Posted February 12, 2023 (edited) Its not because the benchmark is reported Valid that it isn't bugged. If you want to play the: Who can bug the most X benchmark, then HWbot isn't for you. The rules are there not because we just want to add more.... but because ppl will abuse every little bug that exist to get some internet points. Yes some benchmark have different requirements, and its for the best. For exemple 3D03 (and the sub scores rule).... just today, i had to remove a score where GT3 was 3 times higher then all the other scores. This just highlights how some ppl won't take even 30sec to think how they got a score so high and will sub right away and because of that rules have to be constantly updated. Edited February 12, 2023 by saltycroissant Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 12, 2023 Share Posted February 12, 2023 1 hour ago, saltycroissant said: Its not because the benchmark is reported Valid that it isn't bugged. If you want to play the: Who can bug the most X benchmark, then HWbot isn't for you. HWBot is currently allowing 3dMark Firestrike scores that are marked invalid due to LOD/Tess manipulation, which is the very thing I am talking about. How are we going to say we don't support bugged benchmarks and in the same breath say we are OK with LOD/Tess manipulated benchmarks literally marked invalid in 3dmark? Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted February 12, 2023 Crew Share Posted February 12, 2023 Well its pretty simple: Adjusting LOD/Tess was part of the tweaking culture to boost higher FPS, similar to what we did when we were gaming and had to adjust detail settings in a config file in order to get eg 120FPS so we could do certain trick jumps, move faster or could spot the enemy sooner. Thus adjusting LOD was also allowed for overclocking and it gave the community tremendous fun as it was easy to execute. It also required time to find the right setting per sub benchmark. Nowadays to even get certain benchmarks to properly run and even boost older benchmarks via such software tweaks is one of the reasons OC is less popular (besides ofc the insane hardware prizes) Now as Salty mentioned this is something totally different to what we are experiencing right now with the ECC disabled runs. There users push a bit more memory or GPU clocks and feel totally fine that their air cooled card trumps a score of a person running 300MHz faster on the Graphics Core. Wheres the logic in that? On top of that phenoma the benchmark software VALIDates the run... That would even strenghten the fact to what they did is acceptable. UL is still working on a fix, but it could turn out that they will also accept ECC Enabled onlyon runs for their Hall of Fame in the future The "real" inlogic part is that HWBOT adresses these bugged outputs by imposing the ECC enabled runs only rule for RTX 4090 cards. We can't await a fix any longer as this issue has been reported over 4 months ago... We have proper ECC detection already, enhanced monitoring graphs are on the request list. Otherwise the Fire strike series are still very popular (age 10+) and the benchmark software has nicely aged. Meaning a bugged result can be very easily spotted, similar with as the older legacy 3DMarks via the different subtests. Thanks to its popularity it would be nearly impossible to beat older scores if we would impose from now that no more TESS/LOD could be used. I assume you have no idea of the impact by this standardized ruling that all scores have to be VALID at UL site. Think about the fact that due to software age there are no more updates to the Firestrike benchmark series besides an updated systeminfo. UL is still however updating Port Royal (Age 2+) as with the current line-up of hardware/software inconsistent results are still possible. Another example why it stays as is: When we imposed the new rules layout we decided simultaneously that for the lesser GPU intensive benchmarks from the same 3DMark suite (Cloudgate, SkyDiver and co), to submit only UL VALID runs. Even with the lesser benchmark popularity the amount of negative response was mindblowing. We took away the cherished LOD/TESS tweaking option, till users found out that them particular benchmarks were very heavily influenced by the CPU or the Physx score. In the end it is pretty simple , ECC enabled only runs for RTX4090 users evens the playing field, as bugged outputs have reduced to nearly nill out of over 9000 subs monitored by UL moderators over the last two months. Taking into account that Benchmark software doesn't always get fixed/sorted or supported, especially not if they are already over 5 years or older, this is a wise decision. The real problem is not HWBOT here, we need to ensure competitive benchmarking. New Rules are added in 99% of the cases due to users that found either a way to bend them or discovered a new way to bug outputs. If you want I'll make a poll and will announce that on YOUR request we will impose VALID runs for Firestrike and co, I'll add your email address to the post so you can enjoy the flaming and abusive replies. 1 1 Link to comment Share on other sites More sharing options...
Fasttrack Posted February 12, 2023 Share Posted February 12, 2023 I assume that when all is said and done, a major recalculation will happen. Currently I have 4 subs ( legit with ECC enabled ), that are the best subs of my Team ( Warp9 ), that do not give GPTT points. When I press the " Not W9 best submission " I return to the same page, i.e. my sub. Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 12, 2023 Share Posted February 12, 2023 My point was more of a "hey if we are gonna standardize and "fix" rules for 3dmark submissions, now might be a good time to standardize everything" rather than a "I disagree with the ECC change" or "I agree with the LOD/Tess rule". It was the standardization I was hoping for and it just seemed odd that people were up on a soapbox about "valid" scores while we have a bunch of scores that are actually marked "invalid" that we accept. I've said my piece, if site management doesn't see the value in it, I wont belabor the point. Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted February 12, 2023 Crew Share Posted February 12, 2023 The value is that it is not a good idea for a broader perspective.... From single user point great idea, however people that bench older hardware will just stop benching as they can't get remotely close the old scores. Also comparing apples and oranges is never a good idea. We have to keep more than one person happy... 1 Link to comment Share on other sites More sharing options...
wutske Posted February 12, 2023 Share Posted February 12, 2023 The sad thing about this is that Futuremark has done a great work to counter the LOD tweaks in the last versions of older 3DMark titles and because of this it's pretty much impossible to beat older scores with identical hardware, but that's just how it is ?♂️ . With overclocking becoming more popular (and a lot of money being involved too), it makes a ton of sense that Futuremark wants to focus on delivering a score that represents a workable and playable gaming experience, setting your LOD -10 doesn't fit in that and having your view blocked by a massive plane of corrupt vectors isn't either. 2D benchmarks like SuperPi, wprime, etc ... do a lot of verifications to ensure that what comes out of your CPU and memory is actually correct because that's what counts, being the fastest to get to a specific result (and not some random numbers). For 3D benchmarks that's a lot more difficult to do though. What I'm curious about is which benchmarks have screenshot rules that are deviating, afaik for most benchmarks you need a full screen screenshot with CPU-z open (cpu & memory). Anything else is usually benchmark specific. 1 Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 12, 2023 Share Posted February 12, 2023 8 hours ago, Leeghoofd said: From single user point great idea, however people that bench older hardware will just stop benching as they can't get remotely close the old scores. I am asking for those old, invalid 3dmark scores to be thrown out like we are throwing out these BS scores for the ECC issue, and I am asking for the enforcement of valid 3dmark scores going forward. Consistency seems like something we should all want. Requiring valid scores was the whole point of the news article I clicked the link of to start this thread. I'm just asking for us to require valid scores across all these 3dmark benchmarks. None of your arguments are lost on me or ones I want to continue to argue, I just wanted to be clear about what I was asking, since it seemed like maybe I wasn't. Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted February 12, 2023 Crew Share Posted February 12, 2023 The previous management decided to take UL's direction with the launch of Time Spy, initially they even opted for the option to go without a screenshot, thus just requiring a verification link. Some creative users were able to bench with LOD, while still being able to make the run pop up as VALID. So a nice concept was once more hammered by multiple users for some virtual boints, Anyway there are just too many Firestrike scores with LOD/Tess settings adjusted... tossing them is not an option... Neither is it for Vantage , 3D11 and co scores.... 1 Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 12, 2023 Share Posted February 12, 2023 We are deleting 3dmark scores on Hwbot , even if they are marked Valid We are unable to delete 3dmark scores on Hwbot, specifically those marked invalid Link to comment Share on other sites More sharing options...
saltycroissant Posted February 12, 2023 Share Posted February 12, 2023 Feel free to ask the community what they want 1 Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 13, 2023 Share Posted February 13, 2023 (edited) 9 hours ago, saltycroissant said: Feel free to ask the community what they want Did I miss the vote on the ECC issue? All I saw was this thread that I stumbled on. Again I agree with the decision to remove these ECC scores, but have decided to continue to point out how inconsistently that logic is being applied by repeatedly bringing up this issue around LOD/Tess submissions. Backup the database table, post the pre-change champs, and start fresh ffs with a consistent, standardized set of benchmarks so we can all stop the political drama and get back to benching. Edited February 13, 2023 by TonyBombassolo Link to comment Share on other sites More sharing options...
GtiJason Posted February 14, 2023 Share Posted February 14, 2023 19 hours ago, TonyBombassolo said: Did I miss the vote on the ECC issue? All I saw was this thread that I stumbled on. Again I agree with the decision to remove these ECC scores, but have decided to continue to point out how inconsistently that logic is being applied by repeatedly bringing up this issue around LOD/Tess submissions. Backup the database table, post the pre-change champs, and start fresh ffs with a consistent, standardized set of benchmarks so we can all stop the political drama and get back to benching. This is exactly what's been happening moving forward. The only benchmarks that allow LOD are older with the newest being Firestrike. Why would HWBot want to remove all LOD related scores when that was a legit way of tweaking back in the day. We all know what it does and why the score increases the way it does. No one is bugging their runs when using this tweak, it is consistent unlike running GDDR mems far above whats stable and where artifacting begins. There is nothing consistent about trying to break a test to have it shoot out a high score, that's what Any % Speedrunning is for. Competitive 3d benching was built tweaks like this and it is not right to up and remove ten's of thousands of runs people spent endless amounts of hours testing to find a setting that works best for that very specific hardware and test 2 Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 14, 2023 Share Posted February 14, 2023 (edited) 14 hours ago, GtiJason said: Why would HWBot want to remove all LOD related scores when that was a legit way of tweaking back in the day. They are the only scores marked "invalid" on 3dmark we accept. They are the only supported benchmark that has this issue. Meanwhile, here is a whole thread about us removing scores that aren't even marked "invalid" on 3dmark. Thats my whole point. If we are getting on a high horse about 'invalid' runs that seems like a load of BS until we remove actual invalid 3dmark runs we currently allow because "thats the way it always was". If we are going to reach backwards and remove old scores we claim are invalid shouldnt we also include those scores that are actually marked invalid when we do this removal? I mean since we are on our soapbox about bugged runs. Either that or we should allow LOD/Tess for all 3dmark submissions, if we don't think that constitutes a bugged run. ? Edited February 14, 2023 by TonyBombassolo Link to comment Share on other sites More sharing options...
Fasttrack Posted February 14, 2023 Share Posted February 14, 2023 It will not happen mate. Do yourself a favor and stop writing pages. 1 Link to comment Share on other sites More sharing options...
Guest TonyBombassolo Posted February 14, 2023 Share Posted February 14, 2023 8 hours ago, Fasttrack said: It will not happen mate. Do yourself a favor and stop writing pages. This type of incredibly constructive comment is one of the hallmarks of this helpful and welcoming community. Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted February 14, 2023 Crew Share Posted February 14, 2023 We are not gonna wipe 115K subs because you don't want or maybe can't distinct benchmark A from B... Neither do you grasp what exactly has been going on ... End of the chit chat, sorry your monologue 1 Link to comment Share on other sites More sharing options...
Recommended Posts