Evasion Posted April 7, 2018 Share Posted April 7, 2018 Under 1min, awesome! Quote Link to comment Share on other sites More sharing options...
H2o vs. Ln2(BANNED) Posted April 7, 2018 Share Posted April 7, 2018 boosted by 30% improvement, nice...like it! Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted April 7, 2018 Crew Share Posted April 7, 2018 Good score, but new GPUPI scores are silly... Quote Link to comment Share on other sites More sharing options...
Rauf Posted April 7, 2018 Share Posted April 7, 2018 Nothing against your score. But this new gpupi is a joke. You can't fundamentally change a benchmark and call it the same benchmark. Think of the meaning of the word "benchmark". Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted April 7, 2018 Share Posted April 7, 2018 Oh yes you can. See CB r15 and XTU. Quote Link to comment Share on other sites More sharing options...
Rauf Posted April 7, 2018 Share Posted April 7, 2018 34 minutes ago, Mr.Scott said: Oh yes you can. See CB r15 and XTU. Two wrongs don't make a right... Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted April 7, 2018 Share Posted April 7, 2018 (edited) 22 minutes ago, Rauf said: Two wrongs don't make a right... How about 3 then. x265 ver.2 from ver. 1.2 was a huge increase also. Consistency. Edited April 7, 2018 by Mr.Scott 3 Quote Link to comment Share on other sites More sharing options...
Fasttrack Posted April 7, 2018 Share Posted April 7, 2018 (edited) 5 hours ago, Mr.Scott said: How about 3 then. x265 ver.2 from ver. 1.2 was a huge increase also. Consistency. Definitely NOT. Neither evolution nor consistency. A benchmark is a reference point. When this reference point changes by a margin of +30% by its creator, it is a NEW REFERENCE POINT - thus a new benchmark. Acceptable, but it should get a new name and co-exist with the original one. Two benchmarks - two reference points - different points for each. FUTUREMARK - A most serious company. Each and every benchmark follows thru the years the exact same algorithm, since the day of their creation. If Firestrike - for instance - gets a +30% factory boost, Futuremark will hot let it be named Firestrike but something else. NEW BENCHMARK. Math is math, everything else is easy theory and words. Edited April 7, 2018 by Fasttrack Quote Link to comment Share on other sites More sharing options...
Splave Posted April 7, 2018 Share Posted April 7, 2018 What about 3dmark11 old version scores 25% higher on physics Seems this has all been done before so im surprised there is such issues with it happening again. 1 1 Quote Link to comment Share on other sites More sharing options...
FireKillerGR Posted April 7, 2018 Share Posted April 7, 2018 and losing on 3D so eventually they are not like 25% on the overall score Quote Link to comment Share on other sites More sharing options...
Splave Posted April 7, 2018 Share Posted April 7, 2018 Ok how about time spy scores pre lod fix? We just left all those lol I can't even play devil's advocate on that one. I have the 2x 8x 10x 12x 14x 16x 18x gpupi records and I'm fine with rerunning them all in the name of progress. I wish more weren't so short sited. Quote Link to comment Share on other sites More sharing options...
Fasttrack Posted April 7, 2018 Share Posted April 7, 2018 Mr. ( Censored ) Splave ...lol Not all have the privilege to re-run 2X 8X 10X 12X 14X 16X 18X . I wonder if there are 5 or less with this ability. As for short sited, you might wish to refrain... Progress ...Wish to define where you see progress ? In a benchmark ? " Fixed " by its creator to deliver +30% ? I would call progress a benchmark that needs more effort to play. Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted April 7, 2018 Share Posted April 7, 2018 3 minutes ago, Fasttrack said: I would call progress a benchmark that needs more effort to play. I will cite XTU again. Where was the effort there? Quote Link to comment Share on other sites More sharing options...
Fasttrack Posted April 7, 2018 Share Posted April 7, 2018 Nowhere. XTU was the " puppy " of benchmarks due to the fact that Intel poured $$$$$$$$$$$$$$$ in this site. Nothing else. 1 Quote Link to comment Share on other sites More sharing options...
Rauf Posted April 7, 2018 Share Posted April 7, 2018 3 minutes ago, Mr.Scott said: I will cite XTU again. Where was the effort there? You always seem to find a way to justify something bad, by bringing up another bad example. Seems like a fool's way to argue. Btw, what's "progress" in a benchmark? Is a benchmark that takes 3 minutes better than one that takes 2.5? There's no point in optimizing a benchmark, its whole job is to deliver a constant result given the same conditions. 1 Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted April 7, 2018 Share Posted April 7, 2018 (edited) If the bad examples were dealt with originally, we wouldn't be having this discussion. Matter of fact, if they were dealt with now, we wouldn't be having this discussion either. Fact is, it was let slide, so now it's only fair to let the others slide. Edited April 7, 2018 by Mr.Scott 1 Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted April 7, 2018 Crew Share Posted April 7, 2018 Scores removed... I will not accept this new version versus the old ones Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted April 7, 2018 Share Posted April 7, 2018 What are the current accepted versions then please? Quote Link to comment Share on other sites More sharing options...
Crew Leeghoofd Posted April 7, 2018 Crew Share Posted April 7, 2018 3.2 and below Scotty, sorry I can't accept 3.3 like this Quote Link to comment Share on other sites More sharing options...
Mr.Scott Posted April 7, 2018 Share Posted April 7, 2018 Thanks Alby. Don't be sorry. Quote Link to comment Share on other sites More sharing options...
Splave Posted April 7, 2018 Share Posted April 7, 2018 2 hours ago, Fasttrack said: Mr. ( Censored ) Splave ...lol Not all have the privilege to re-run 2X 8X 10X 12X 14X 16X 18X . I wonder if there are 5 or less with this ability. As for short sited, you might wish to refrain... Progress ...Wish to define where you see progress ? In a benchmark ? " Fixed " by its creator to deliver +30% ? I would call progress a benchmark that needs more effort to play. I won't even bother responding lol you are the loudest most negative voice on hwbot.org. You get the gold for that and you don't even need to cheat for it this time. Short sited is using a bench that is now eol and not being able to see that 3.3 will be the new version and 3.2 will stop getting points in a matter of time. The way it was before leghoofd decided to remove results was keeping some points for those that have benched. The new way will make all scores zero and require you to have used the new version to receive any so congrats on that guys lol. How did it feel when r11 and wp32 lost points. So lets keep starting fires by rubbing twigs together when we have gas and matches available to us. Quote Link to comment Share on other sites More sharing options...
jpmboy Posted April 8, 2018 Share Posted April 8, 2018 (edited) IMO, this in this case, where if GPUPI is not allowed to evolve, eventually we end up going the way of an old ham radio club that did not evolve to digital. (no postcards please ). Make a shift date, freeze boints for versions before 3.3 and use 3.3 going forward. It is inevitable that benchmarks will need to evolve with emerging architectures and platforms - old benches will not run new instruction sets and so on. But, there's two sides to this argument. IMO, keeping ancient "GPU" benchmarks like aquamark, MK03 etc, is off-kilter also - these simply are not GPU benchmarks anymore. Then, we have no AVX512 benchmark with boints either... etc, so plenty of opportunities to add benches that push modern hardware too. Seems a shame that more efficient coding (3.3 solves the same problem) is being "shat upon" Lastly, try to keep it "us against the hardware" . Edited April 8, 2018 by jpmboy 2 Quote Link to comment Share on other sites More sharing options...
Noxinite Posted April 8, 2018 Share Posted April 8, 2018 What about HWBot Prime guys? New Java versions anyone? 1 Quote Link to comment Share on other sites More sharing options...
Rauf Posted April 8, 2018 Share Posted April 8, 2018 This whole situation is stupid, this is not how benchmarks are supposed to be handled. When you fundamentally change a benchmark you give it a new name, like geekbench 4 instead of 3. Time spy instead of fire strike etc. A benchmark is not a software that solves any real world problem, thus do not need to be made "more efficient". Mat is already talking about implementing avx etc. You can't have a benchmark that continually change. Mat really should consider what the goal of a benchmark is... Just give it a new name and then hwbot can implement it as a new category if they want it. Quote Link to comment Share on other sites More sharing options...
Administrators websmile Posted April 8, 2018 Administrators Share Posted April 8, 2018 Agreed on rauf, what a miracle. You can't program a benchmark completely new plus change calculation speed and then a) put it against the old, differently coded benchmark at same ranking and b) let it go online without any decent test phase to find bugs and vulnerabilities. We had this in the past and it caused big trouble and if I think of realbench, without my and others testing plus a beta phase we would now have one of the most cheatable and unreliable benchmarks ever. Maybe it is time to learn from mistakes, we had this before and I do not think hwbot needs this again 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.