Calathea Posted February 16, 2014 Posted February 16, 2014 I'll just go with Roman's settings for E2220 then ^^ No 1 spot for E2220 is probably bugged Quote
Moose83 Posted February 16, 2014 Posted February 16, 2014 (edited) You can also get bugged runs with this threats No matter what Threats you set, bench buggs sometimes itself Odd numbers only help to get more bugged runs SOlution, make 20 runs and safe all scores and ask Mod wich one is legit Edited February 16, 2014 by Moose83 Quote
Moose83 Posted February 16, 2014 Posted February 16, 2014 No, save all scores and ask for legit seems the only way Quote
Administrators websmile Posted February 16, 2014 Administrators Posted February 16, 2014 (edited) How about simply making this c*** pointless? - If you have a bench that doesn´t work flawless and on which each submission normally would have to be checked by an expert mod, that´s the way to go - otherwise we could unban win8 and make it free on all benches without any limits as well^^ - only alternative is to make a bullet proof wrapper or find a method that eliminates all bugs, still you will have to check the xxx thousand submissions already made before by hand to keep the ranking legit Edited February 16, 2014 by websmile Quote
ObscureParadox Posted February 16, 2014 Posted February 16, 2014 So I was doing some testing on a 3 hour train journey home, hoping to reproduce the results other people have had using my trinity laptop. After 30 runs I just gave up. Not a single bugged run so is it actually all to do with the command line or is there something else too?? My system : A10-4600M Regular Win 8 install Command line : unrar bench - cf=XOP -cpus =8,45,64 test.rar -ri7 So this could mean one of many things, 1) I did the command line wrong like a noob 2) it isn't affected with AMD CPUs or the XOP function. 3) I was just unlucky not to get a bugged run 4) doesn't happen on windows 8? 5) other tweaks need to be applied for thethe bugged run to happen I.e. end processes ect. 6) I'm sure there are loads of other things it could be but I don't want the list to go on forever Any input would be appreciated, cheers Quote
GENiEBEN Posted February 16, 2014 Posted February 16, 2014 How do you know it wasnt bugged? Attach a screenshot. Quote
ObscureParadox Posted February 16, 2014 Posted February 16, 2014 I screenshotted the first 10 results and was going to do it for every run but in the end just kept running it until I got what I would think is a bugged run. I can't attach anything right now since I'm on my mobile and my route has many tunnels so I lose signal often. But just as an example of what I got, all results were inbetween 365-385 but usually around 380. Nothing out of the ordinary and I have had 412 previously when the laptop was plugged into the socket and running at max frequency. Quote
Crew Strunkenbold Posted February 16, 2014 Crew Posted February 16, 2014 Here is an observation. UCBench has been around for years, but it's only when HWB give big points for it that people A/find problems B/complain about them and/or C/anything is done about them. Started discussion over one year ago... http://forum.hwbot.org/showthread.php?t=59847&page=4 But just didnt got much attention... Quote
[GF]Duke Posted February 17, 2014 Posted February 17, 2014 (edited) So bugged runs are easily spotted. Like Super Pi runs. Why not just rely on the community to report them? And how about a little honesty amoug the troops? You know when you have a bugged run. Don't submit it. Edited February 17, 2014 by [GF]Duke Quote
Massman Posted February 17, 2014 Posted February 17, 2014 Global points for the benchmark are temporarily disabled. The staff has been looking into this issue and for a solution. In the mean time we don't want to give anyone an incentive to submit bugged scores to boost their ranking (). To be updated! Quote
bartx Posted February 17, 2014 Posted February 17, 2014 Duke;295130']So bugged runs are easily spotted. Like Super Pi runs. Why not just rely on the community to report them? And how about a little honesty amoug the troops? You know when you have a bugged run. Don't submit it. Looks nice, but how do you know which run is good? I mean not abnormally low or high. In my opinion Ucbench should be replaced by something else from beta benchmarks. Who will check all the scores in the database? Quote
ObscureParadox Posted February 17, 2014 Posted February 17, 2014 Looks nice, but how do you know which run is good? I mean not abnormally low or high. In my opinion Ucbench should be replaced by something else from beta benchmarks. Who will check all the scores in the database? Agreed, I think the damage has been done here now, there are too many bugged results for people to go searching through every result in the database and people know how to achieve these bugged scpres and almost cheat the system effectively. We really do need a solid benchmark that is consistent across the board for global points. Quote
[GF]Duke Posted February 17, 2014 Posted February 17, 2014 Well you obviously havn't looked. Ther aren't that many at all. Theres nothing wrong with the bench. It's the people using it. Anyone can tell when their score is out of line. Quote
Chi-Kui Lam Posted February 17, 2014 Posted February 17, 2014 If Result >= 2000 then auto screenshot & shutdown else loop run. people want to play fault run , I can play with u as many times as you want. Quote
Doc.Brown Posted February 17, 2014 Posted February 17, 2014 I've read all the previous pages and I really don't understand how to see if my scores are bugged or not,it seems it's not really clear to attest how and when a score is really bugged at this stade,isn't it? Anyway all that stuff let me thinking one question in my mind: if "lucky runs" are faults runs for UC Bench so what about Car Low lucky runs in 3D Mark 01 ? Quote
ObscureParadox Posted February 17, 2014 Posted February 17, 2014 I've read all the previous pages and I really don't understand how to see if my scores are bugged or not,it seems it's not really clear to attest how and when a score is really bugged at this stade,isn't it? Anyway all that stuff let me thinking one question in my mind: if "lucky runs" are faults runs for UC Bench so what about Car Low lucky runs in 3D Mark 01 ? Not done a lot of benching with 3D01 practically 0 actually, but isn't that to do with the barrel movement and other things being rendered in the scene so it's fine since there isn't a lot we can do about it? Quote
Splave Posted February 17, 2014 Author Posted February 17, 2014 (edited) this is foolish imo but I will play along ^ All bugged, you can see right before the best score there is a large increase in time that round......and also the round with the most checked password needs to be your best scoring round within a couple 100... Like this proper run seems pretty cut and dry to me Edited February 17, 2014 by Splave Quote
Chi-Kui Lam Posted February 18, 2014 Posted February 18, 2014 (edited) When more threads created , more password assumed to be made , so your CPU need to take much long time to finish. Fault or not is very easy to tell. Edited February 18, 2014 by Chi-Kui Lam Quote
Moose83 Posted February 18, 2014 Posted February 18, 2014 SO use of odded Numbers or -ri isnt the problem. When i have too much pwds in the for example 45T and less in 64T it is bugged run? Quote
Black Cobra Posted February 18, 2014 Posted February 18, 2014 They should not ban this benchmark. They should make a proper wrapper for it. Otherwise they should ban pcmark too. Just my two cents, but i know there are enough people which agree with me Quote
Doug2507 Posted February 18, 2014 Posted February 18, 2014 Duke;295247']Well you obviously havn't looked. Ther aren't that many at all. Theres nothing wrong with the bench. It's the people using it. Anyone can tell when their score is out of line. Well at least all the 4960x scores aren't bugged but as for sorting out the rest, don't envy that task! Quote
der8auer Posted February 18, 2014 Posted February 18, 2014 They should not ban this benchmark.They should make a proper wrapper for it. Otherwise they should ban pcmark too. Just my two cents, but i know there are enough people which agree with me UCBench is not banned. GENiEBEN is already working on a new tool for the benchmark. Just takes some time Quote
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.