Jump to content
HWBOT Community Forums

$@39@ is cheating X265 and HWBot Prime


Splave

Recommended Posts

I think BOT should take a pause and look in to a mirror..spend some money on research and development (one flaky bench after another ),how you expect ppl having fun ? (waist of time and money ),this is prime example how poorly tested some of this benches are ,you don't have to alter shit just fuck with the timer ,even Splave can do it -you can shame all you want for all the like's you desire but I"m pretty sure it won't make a dent if you want to make a change

HWBOT Prime will end up dead so why you won't disable the points now ?,etc

Link to comment
Share on other sites

Well, in hwbot's defense - benchmarks are way less likely to be "broken" while in beta stage since lack of points usually translates in lack of motivation to do so. So whatever happened today is more like of a naturally-risen challenge for hwbot to improve security along the way.

Link to comment
Share on other sites

Well, in hwbot's defense - benchmarks are way less likely to be "broken" while in beta stage since lack of points usually translates in lack of motivation to do so. So whatever happened today is more like of a naturally-risen challenge for hwbot to improve security along the way.

 

x265 points are a new thing, but in regards to HWBot Prime one has been able to get points with it for a rather long time... HWBot Prime gave points back when I started here on HWBot almost three years ago.

Link to comment
Share on other sites

  • Crew
I think BOT should take a pause and look in to a mirror..spend some money on research and development (one flaky bench after another ),how you expect ppl having fun ? (waist of time and money )?,etc

 

I dig your sarcasm, but however the only reflection I have is that it is all thanks to those that want to go that extra mile, they spoil most of the fun for others... why can't people just run a benchmark and focus on raw clocks again...

 

I have no issue with efficiency tuning by using other OSses and co,. However from the moment you start to tinker with internal clocks, swap images & encoders you know you are stretching it. It is all up to the keyboard warriors, they have to look in the mirror, not the Bot...

 

Just my two cents on another big dissapointment for the OC scene :(

 

and by the way thanks to Dr Allen for adding extra work for the moderation team :D

Link to comment
Share on other sites

Can you directly figure out all the people that actually cheated with cinebench?

 

The case is much more complicated than that.

 

Let's ban everybody more fun that way..

I'll answer that question with a question.

 

Can you figure out all the people that used this cheat with X265 & more importantly Hwbot Prime ( cause it's been around longer).?

 

If no then let $@39@ be given the same "punishment" as Cinebench cheaters. Seems okay??

 

As for banning everyone. I'd rather have 50 honest benchers that 500 in which I don't which are cheaters & which are not.This way, when I can't match the efficiency of someone I'm chasing, I can blame my lack of skills & not have to worry about how that person found another way to cheat after having no consequences for their previous offense.

 

Sent from my Nexus 5X using Tapatalk

Link to comment
Share on other sites

I dig your sarcasm, but however the only reflection I have is that it is all thanks to those that want to go that extra mile, they spoil most of the fun for others... why can't people just run a benchmark and focus on raw clocks again...

 

I have no issue with efficiency tuning by using other OSses and co,. However from the moment you start to tinker with internal clocks, swap images & encoders you know you are stretching it. It is all up to the keyboard warriors, they have to look in the mirror, not the Bot...

 

Just my two cents on another big dissapointment for the OC scene :(

 

and by the way thanks to Dr Allen for adding extra work for the moderation team :D

 

I don't think the approach you guys taking is very productive , shaming or banning someone is one thing and I understand that but this won't stop ppl from doing it ,you have the best and the brightest in tweaking why not reward them instead ,put them on the payroll and watch how your benching app improve ,Bot have been around for years and should have the best/cheat proof benches available by now ,need to level up bro's

Link to comment
Share on other sites

I don't think the approach you guys taking is very productive , shaming or banning someone is one thing and I understand that but this won't stop ppl from doing it ,you have the best and the brightest in tweaking why not reward them instead ,put them on the payroll and watch how your benching app improve ,Bot have been around for years and should have the best/cheat proof benches available by now ,need to level up bro's

 

I'm not new to PC's or overclocking, my first pc was a Pentium 75mhz and i've been overclocking sense celeron 300A days but am new to getting into comps and such but i have to agree.

 

The thing that has surprised me the most about HWBOT is that its almost at times how much you know about specific benchmarks and tweaks required to score highest for a given setup that matter end of the day and that info is usually hardest to figure out. its also kind of blown me away at how honor system everything is but i kinda understand that sense many benches can't produce specific validation files and whatnot but still if work was rewarded/put in on discovering really game breaking exploits like this one and then used to fix the situation that is better overall than just punishing the offenders for finding it but at the same time we'd need to expect everyone that finds a really lame exploit like this to report it and not just pretend like they found some special driver/software tweak that gives them the result and keep it a secret which in a competitive environment people will never be inclined to do so not sure how to handle it, it is quite difficult.

 

But in any event to give my rookie .02 cents on things i think you'd see more participation if things required a bit less tweaking/discovering the trick to specific benches. Some of us find it fun to search and ask around to figure out how the heck people manage certain scores that seem impossible but at the same time it can be quite frustrating to newer competitive overclockers looking to get into it.

Link to comment
Share on other sites

  • Administrators

Hwbot prime has been used for 7 years or more - if you think this benchmark is insecure (which now was proven by a lot of criminal energy) you should test a lot of other benchmarks hwbot considered for ranking once. It is a simple fact that there is no 100% safety, what you ask for is something no bank nor any billion dollar company can offer, otherwise there would be no hacks on online banking or exploits. We have benchmarks developped by companies that are much easier to manipulate than hwbotprime for example

Link to comment
Share on other sites

Ok, so I found the screenshot for $@39@'s 7900X 1080p score. 1128 frames encoded in 12 seconds (elapsed time) equals around 94 FPS, yet his score is 120FPS. This was run on MSI Gaming Pro carbon, which does not have pause button. Seems the only explanation can be time manipulation = cheat. After he mentioned pause button and it actually affected elapsed time without altering the score I thought maybe he is innocent after all. But this definitely proves it.

 

What made me wonder was that his score available on uat.hwbot.org with 6950X seems fine.

Link to comment
Share on other sites

Ok, so I found the screenshot for $@39@'s 7900X 1080p score. 1128 frames encoded in 12 seconds (elapsed time) equals around 94 FPS, yet his score is 120FPS. This was run on MSI Gaming Pro carbon, which does not have pause button. Seems the only explanation can be time manipulation = cheat. After he mentioned pause button and it actually affected elapsed time without altering the score I thought maybe he is innocent after all. But this definitely proves it.

 

What made me wonder was that his score available on uat.hwbot.org with 6950X seems fine.

 

But from my point of view you should also try running the benchmarks from console since they are java based. You can even pause them (asus has an onboard button for this)

 

 

 

Sorry to jump in but based on what he wrote I think he meant that you can pause via console when it comes to Java benchmarks even without a pause switch.

 

If someone knows more please feel free to correct me if I am mistaken.

Link to comment
Share on other sites

Ok, so I found the screenshot for $@39@'s 7900X 1080p score. 1128 frames encoded in 12 seconds (elapsed time) equals around 94 FPS, yet his score is 120FPS. This was run on MSI Gaming Pro carbon, which does not have pause button. Seems the only explanation can be time manipulation = cheat. After he mentioned pause button and it actually affected elapsed time without altering the score I thought maybe he is innocent after all. But this definitely proves it.

 

What made me wonder was that his score available on uat.hwbot.org with 6950X seems fine.

 

looking at the attached picture it's also interesting see what cpuz says (101.67x58) and what system info reports (138x24)

Don't know if is due to a platform issue but, in my 1151 experience, system info always reports correctly bus and ratio; and I think it's not so easy to change bclk from 101 to 138 on the fly...

Link to comment
Share on other sites

Flipping forward isn't necessary, at least not for HWBOT Prime. The benchmark just halts and picks up after a while.

 

Do you have the same behavior with older Java versions?

 

I do see glitches with the Java 9 64bits (but that could happen with other versions, I use that one snce I was told this is the best one...), no need to apply any cheat, I have seen the system halting / freezes, then it goes on again and display a pathetically high score. 0, absolutely 0 human interaction, not even touching the mouse ! And frankly, I don t always keep my eyes on the software. Often, after I start the bench, I switch the display input to get to my pc to submit prior scores, then come back to the bench station.

This way I get some good or bad "suprise effect" when I see the new score, and sometimes when you are at the o/c edge of the system, you don't want to see the system BSOD crash... :D

 

In my book, this glitch may be called relatively POOR programming by those releasing bench software : obviously, an overclocked system could destabilize basic windows service like the clock, or whatever is ued to time the benchmark? (I have 0 programming skills)

If someone gets banned because of someone else not being capable to code PROPERLY (meaning the counting procedure is not hardware-bug 100% proof), should the coder be left in peace....? IMO, this makes no sense.

 

One may follow the rules and apply a bugged score, knowing or not whether the score is real. Yet, he did not do anything against the written rules...

 

 

So what to do? Remove all scores...? Or remove only bugged scores? But yeay, shouldn't there be some validation tool like superPI to make sure the score is realistic / not glitched ?

Removing glitched scores a posteriori, is condemning a lot of folks who submitted their best files while perhaps having a slightly lower one, not glitched but not knowing that, and currently not able to rebsubmit or rebench. Unfair in my book.

 

 

PS: if a developper wants what to retrive what I know is a "bugged" file for a HWBOT PRIME score of G3258, score which would have broken the World Record !!! I can post it for their examination. I did not submit the file in HWBOT for obvious reasons...

Edited by bolc
Link to comment
Share on other sites

  • Administrators
I do see glitches with the Java 9 64bits, no need to apply any cheat, I have seen the system halting / freezes, then it goes on again and display a pathetically high score. 0, absolutely 0 human interaction, not even touching the mouse ! And frankly, I don t keep my eyes on the software. Often, after I start the bench, I switch the display input to get to my pc to submit prior scores, then come back to the bench station.

This way I get some good or bad "suprise effect" to look at the new score, and sometimes when you are at your oc edge, you don't want to see the system BSOD crash :D

 

In my book, this glitch may be called relatively POOR programming by those releasing bench software : obviously, an overclocked system could destabilize basic windows service like the clock?

If someone gets banned because of someone else not being capable to code PROPERLY, shouldn't the coder be left in peace....?

One may follow the rules and apply a bugged score, knowing or not knowing it is bugged. Yet, he did not do anything against the rules...

 

 

PS: if a developper wants what I know is the bugged file for a HWBOT PRIME score of G3258, which would have broken the World Record, I can post it for their examination. I did not submit the file for obvious reasons.

 

Disallowed tweaks/cheats

Any software or human interaction altering the perceived speed of the benchmark program, tricking it to believe it ran faster

I think for x265 and the rules, I quote them again, you can´t talk of not cheating. It is a deliberate act to pause the benchmark, it is visible the result is not obtained correctly because of the time elapsed and one or two other issues.

I was very fond of $@39@ from the start of his comeback and always willing to help, recaculate scores, answer questions and so on and it is no fun for me to see this situation now. It also damages hwbot once again.

On Hwbotprime, it might be eol, but I had never the problem you mentioned with paused benchmark for example. If I would get a dollar for each 3dmark, catzilla or 2d benchmark I had to block for being bugged runs or which I saw (including spi 1m pifast etc) I could buy cpus at palettes xD. As long as we have thousands of different variations of os, drivers, stripped copies etc we will always face bugs even at the most reliable benchmarks. But bugs and cheats are a different pair of shoes...

Link to comment
Share on other sites

Disallowed tweaks/cheats

Any software or human interaction altering the perceived speed of the benchmark program, tricking it to believe it ran faster

I think for x265 and the rules, I quote them again, you can´t talk of not cheating. It is a deliberate act to pause the benchmark, it is visible the result is not obtained correctly because of the time elapsed and one or two other issues.

I was very fond of $@39@ from the start of his comeback and always willing to help, recaculate scores, answer questions and so on and it is no fun for me to see this situation now. It also damages hwbot once again.

On Hwbotprime, it might be eol, but I had never the problem you mentioned with paused benchmark for example. If I would get a dollar for each 3dmark, catzilla or 2d benchmark I had to block for being bugged runs or which I saw (including spi 1m pifast etc) I could buy cpus at palettes xD. As long as we have thousands of different variations of os, drivers, stripped copies etc we will always face bugs even at the most reliable benchmarks. But bugs and cheats are a different pair of shoes...

 

I edited my message after I initially posted.

I didn't mean that someone needs to exploit bugs by looking into the code or finding ways to make the software behave out-of-norm. What if the software has bugs and displays, in certains combinations of hardware/software, scores which are invalid=unreal, what can we do about it ...?

 

Otherwise, you mean HWBOT PRIME is End-Of-Life .... ?

 

Damn, I thought I saved the file for G3258, but I guess I did not want to be faced with temptation, I can't find it.... Will look on the bench station harddrive tonight.

Edited by bolc
Link to comment
Share on other sites

Pausing benches in any form while it runs kills the score. This is why asus doesn't even use the pause button anymore. Pausing using Java console is no different scores worse. BTW why are you even playing with a Java console anyways?? Just run the damn benches.

Link to comment
Share on other sites

I edited my message after I initially posted.

I didn't mean that someone needs to exploit bugs by looking into the code or finding ways to make the software behave out-of-norm. What if the software has bugs and displays, in certains combinations of hardware/software, scores which are invalid=unreal, what can we do about it ...?

 

Well, for starters, if you know the result might be invalid/faulty, don't just submit it. That's the first and most simple step one can take. Consult an admin or post a topic and ask if other people with the same hardware experience the same issue(s).

 

It's very easy to blame it on the coder, but it all starts with yourself. You don't blame a car manufacturer if your engine blows up because you were able to tune it, do you? It's your own responsebility to run it as it's meant to run, especially if you want to compare to others and compete professionally. You don't just alter the benchmark, or run it in a way it's not meant to be ran. The rules are this simple.

Link to comment
Share on other sites

Well, for starters, if you know the result might be invalid/faulty, don't just submit it. That's the first and most simple step one can take. Consult an admin or post a topic and ask if other people with the same hardware experience the same issue(s).

 

It's very easy to blame it on the coder, but it all starts with yourself. You don't blame a car manufacturer if your engine blows up because you were able to tune it, do you? It's your own responsebility to run it as it's meant to run, especially if you want to compare to others and compete professionally. You don't just alter the benchmark, or run it in a way it's not meant to be ran. The rules are this simple.

 

What tells you that there is a way to know if your hwbot prime score is 100-200pps higher than it should be?

Ram settings, OS tweaks, etc, can cover it all, plus who is inspecting/comparing to say this score seems out of range. Whn I submit 10-15 scores on a cpu, I don 't check where it fits next to the others, I look at the score and move on...

Link to comment
Share on other sites

Hwbotprime is a weird case itself, because it's not consistent at all, but this doesn't go for x265. In this benchmark it's very obvious when it's bugged or cheated on. Same goes for a lot of other benchmarks. Ofcourse there will always be exceptions, but I'd like to keep it about the normal benchmarks and not the exceptions that are weird or stupid to begin with.

 

I always check my scores before or right after submitting. I want to know how efficient I am and improve it if neccesary. It's not all about dump the scores and continue. That's not a very good way to improve yourself. If you never compare your score to others, how will you know if you're on par with others or just very bad at a certain benchmark?

 

Click bench and look away might work for you now, but when you get to the point you are benching seriously with extreme cooling, you might want to know which BSOD you get so you can correct what is neccesary and improve your stability. You'll know when a benchmark is valid or not...

 

This is besides the point though, because in this case he knew very well what he did.

Link to comment
Share on other sites

Pausing benches in any form while it runs kills the score. This is why asus doesn't even use the pause button anymore. Pausing using Java console is no different scores worse. BTW why are you even playing with a Java console anyways?? Just run the damn benches.

 

 

why? because when I opened the archive expecting a .exe, I found a .jar file (WTF is that. can t you make a .exe guys....?). I am from a school thinking that benchmarking should rely on HARDWARE only, and tweak OS should be plainly forbidden, including all forms of registry/mods to boost superPI for instance etc. I think HWBOT should release full OS to load on USB sticks to boot to benchmark, no modification allowed whatsoever so everybody goes by the same book !

 

Anyways... I could not load it given I had no java installed and never use .jar files... so I found I had to install java and put 32 bits first. then 64 bits on the advise of someone else.

the .jar icon did not change so I did not think of double clicking... so I searched the web and found this way to start a .jar file via the command line (cmd.exe) : java -jar <jar-file-name>.jar.

lastly I made a shortcut to open the .jar. the command starts the java console, then I push the quick benchmark button.

Edited by bolc
Link to comment
Share on other sites

I always check my scores before or right after submitting. I want to know how efficient I am and improve it if neccesary. It's not all about dump the scores and continue. That's not a very good way to improve yourself. If you never compare your score to others, how will you know if you're on par with others or just very bad at a certain benchmark?

 

Click bench and look away might work for you now, but when you get to the point you are benching seriously with extreme cooling, you might want to know which BSOD you get so you can correct what is neccesary and improve your stability. You'll know when a benchmark is valid or not...

 

Sorry, only an enthusiast here... I know, bosd dump file analyses can tell you this or that. and sure, comparing with others helps, but in the end if you know the limits of your hardware on your board and coolig, and of your ram, there is little use in comparing

on hwbot, as enthusiast, if you want to make points, you have XTU for modern cpus, you had hwbot prime and x265, and gpupi can also give good suprises, and for old cpus, max frequency, spi, and wprim. the rest is out of reach past 5.3+ Ghz given you will face DI, SS and LN2 clockers.

 

in my mind, there could/should even be different championships (one per league), with max scores of 50pts for each bench and for each league, as is done in any type of championship except world cups...

Edited by bolc
Link to comment
Share on other sites

  • Crew

My 2p

 

While I agree with Websmile on the issue of the Cinebench scandal being different to this current one, there is no dispute that $@39@ flouted the rules and essentially cheated - Absolutely no disputing that.

 

But I still feel deep inside of me that, because it's been found so quickly (the exploit) and he seems to be a lone wolf at current, it's easier to ban him than 15+ top benchers who manipulated benchmark files in Cinebench for an advantage.

 

What pains me so much - It's not that the render image was changed in the first place, but by how much to get such little gains to make it look like strong efficiency which some actually called it "tweaking". Christian Ney proved that by editing enough, it could prove massive yields, but if this was ALLOWED, why wouldn't the culprits do that instead of only doing it enough so it was hardly noticable?

 

That IMO is just as bad, if not worse than $@39@ - The sooner people stop pretending this, the better we can ALL move on from all of this crap.

 

P.S - I say this with the utmost respect, but there has been cases where big names have been let off from any consequences from breaking rules (through apparent ignorance) because of who they are. I would put money on if $@39@ was to have done the same thing on Cinebench and was the first, he would have also been banned for it.

 

Oh and I don't have Elite envy, I have access to ES samples of latest chips myself (I should technically be elite), but I honestly believe in an open playing field and you can't have a level playing field when cheaters (however hard it is to side sweep the issue) remain on HWBot, but others get openly banned because it's much easier in one instance to do so

Link to comment
Share on other sites

Either way we cant have people continue to break the rules no matter what happened in the past. Someone got caught and needs to suffer the consequences if true. If we want to move forward we just have to handle it this way from now on and try and forget past mistakes. And thank Splave for catching it. but I was definitely frustrated by cinebench also, mainly because I didnt get to use it, haha.

Edited by Strong Island
Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...