Jump to content
HWBOT Community Forums

$@39@ is cheating X265 and HWBot Prime


Splave

Recommended Posts

I have been wondering about this score for a long time: $@39@`s HWBOT x265 Benchmark - 1080p score: 55.7 fps with a Core i7 6700K

Super "efficiency" with E-die mems at 1800MHz, beating Dancop with similar CPU speed and B-die at 2090+ 12-11...and that score was for Country Cup...

 

Edit: and benchmark took two seconds longer to finish...

Edit2: never mind fixing hwbotprime, aka the slot machine, good riddance

Edited by Rauf
Link to comment
Share on other sites

Edit: and benchmark took two seconds longer to finish...

 

Looks like this is how we may be able to identify these results: $@39@`s HWBOT x265 Benchmark - 1080p score: 55.7 fps with a Core i7 6700K

 

Frames: 1128

Time elapsed: 22 seconds

FPS: 1128 / 22 = 51.27 != 55.70

 

//edit: this can vary a bit depending on the milliseconds (i.e. 22.499sec). But in general, we should be able to identify a bogus run by comparing run time and FPS output for X265.

 

HWBOT Prime will be another story however ...

Edited by Massman
Link to comment
Share on other sites

What about this

OGS`s HWBOT x265 Benchmark - 4k score: 6.6 fps with a Core i3 7350K

 

Then look at my submission.

 

xMec`s HWBOT x265 Benchmark - 4k score: 6.63 fps with a Core i3 7350K

 

Could Be cheated ? Couldn't do 6.6 at his frequency and Memory timing. Had to push way far

 

Inviato dal mio ONEPLUS A3010 utilizzando Tapatalk

Dude, you can't just accuse someone of cheating with no proof. This benchmark varies alot, especially on win7. Tune your OS, or better yet, use proper OS for this bench.

Link to comment
Share on other sites

Dude, you can't just accuse someone of cheating with no proof. This benchmark varies alot, especially on win7. Tune your OS, or better yet, use proper OS for this bench.
Nvm man Just got confused [emoji28]

 

Inviato dal mio ONEPLUS A3010 utilizzando Tapatalk

Link to comment
Share on other sites

  • Administrators

Is this "tweak" tested to work on 1151 as well or is this just speculation now? And second point, does it work on benchmarks with point based result only or also on time based results? And third point, I have a list with some weird 3D results, is it possible stuff like Unigine Heaven for example or FM benchmarks are affected as well? Thanks in advance

Link to comment
Share on other sites

Is this "tweak" tested to work on 1151 as well or is this just speculation now? And second point, does it work on benchmarks with point based result only or also on time based results? And third point, I have a list with some weird 3D results, is it possible stuff like Unigine Heaven for example or FM benchmarks are affected as well? Thanks in advance

Cines wprimes geek gpupi seem safe from what I try. Geek has built in timer that must match or sends error. Wprime says cheat detected, cinebench scoring the same either way. Same as gpupi. 3d I can try some tomorrow or anyone else can try too.

 

Im not software engineer so maybe I'm missing something out of those but that specific cheat is not working using that method. I can try changing the method to see if there is some tolerance you can still get boost from this

 

It should work on any platform since the program uses windows time in the score in some form.

Edited by Splave
Link to comment
Share on other sites

Looks like this is how we may be able to identify these results: $@39@`s HWBOT x265 Benchmark - 1080p score: 55.7 fps with a Core i7 6700K

 

Frames: 1128

Time elapsed: 22 seconds

FPS: 1128 / 22 = 51.27 != 55.70

 

//edit: this can vary a bit depending on the milliseconds (i.e. 22.499sec). But in general, we should be able to identify a bogus run by comparing run time and FPS output for X265.

 

HWBOT Prime will be another story however ...

Yes, it seems the elapsed time is correct even on the cheated x265 scores - most likely different method is used to measure this time and the actual score time (which is then translated to fps). I will investigate this as soon as I can. I hope it will be possible to fix this vulnerability.

 

I've started to work on update of the x265 benchmark few days ago, it seems there are more problems to solve that I originally thought... this one is extremely serious.

 

I'm sorry such a big issue made it to the live version. During development I've put a lot of effort to make this benchmark as secure as possible (there are many anti-tampering measures in place)... but maybe the most obvious security hole slipped through and been here for the whole time. This time I must do even more thorough testing before the next version goes live.

Link to comment
Share on other sites

Is this "tweak" tested to work on 1151 as well or is this just speculation now? And second point, does it work on benchmarks with point based result only or also on time based results? And third point, I have a list with some weird 3D results, is it possible stuff like Unigine Heaven for example or FM benchmarks are affected as well? Thanks in advance

Looking at this score: http://hwbot.org/submission/3396930_a39a_hwbot_x265_benchmark___1080p_core_i7_6700k_55.7_fps

It works without a doubt on 1151. The benchmark duration is ~2 seconds off, and eff is through the roof for that mem speed.

 

Doesn't heaven count total number of rendered frames for benchmark score? If so it should not be affected. Don't know about FM but they have been affected by rtc bugs in the past, so it would seem it's worth testing it at least. But it would surprise me if FM haven't thought about this...

Link to comment
Share on other sites

  • Administrators
Looking at this score: http://hwbot.org/submission/3396930_a39a_hwbot_x265_benchmark___1080p_core_i7_6700k_55.7_fps

It works without a doubt on 1151. The benchmark duration is ~2 seconds off, and eff is through the roof for that mem speed.

 

Doesn't heaven count total number of rendered frames for benchmark score? If so it should not be affected. Don't know about FM but they have been affected by rtc bugs in the past, so it would seem it's worth testing it at least. But it would surprise me if FM haven't thought about this...

 

Well, the time is 2s slower than on second place but result is higher, on splaves results it is the opposite, faster time= higher score like in the rankings.

On heaven the results I checked have extremely low minimum fps but extreme high maximum fps compared to other results, this made me wonder if a in benchmark time boost accelerating fps like a fast forward might be problem, but it can have lots of reasons

Link to comment
Share on other sites

On heaven the results I checked have extremely low minimum fps but extreme high maximum fps compared to other results, this made me wonder if a in benchmark time boost accelerating fps like a fast forward might be problem, but it can have lots of reasons

 

I believe this is what you said to me about my results and it looks to be a valid observation but then what do we say would be causing this in my results since I've been recording the runs to show I'm not messing with them. I've made video's so far for gtx 970/980 and soon 980ti, should I zoom out to show there is no tampering going on. I'm willing to do whatever it takes to prove I'm not messing with anything even borderline sneaky and do believe I can recreate the same eff on demand if needed so lmk. Glad to help

Link to comment
Share on other sites

Great find splavor, no sarcasm really mean it.

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). By pausing you can bench around 30-50mhz more depenting on cpu and bench.

Pause does not work on hwbotprime for some reason but does work on x265.The overall score ia not affected but you can run higher freq since the benchmark is not all one huge load.the only thing affected is the timer but the score from the valitidy point of view is fine in my opinion.It is the first time i wish i was not on vacation to prove myself but this will happen trust me.

I think it is good that we as overclockers try and find exploits of benchmarks but we should not be so thirsty of drama.Of course all of my results are open for scrutinizing,so you are free to do that,just do not jump in conclusions...

Link to comment
Share on other sites

Well thanks mods for jumping on this asap. I hope this guy gets a chance to tell his side but .. seems a strong vote from many known ocer. Seems i took too long to post :>

 

If someone is found guilty of a cheat i hope they get a strong message in the shape of a ban.

 

This will create havoc for all of the mods and its pretty sad.

 

I feel for the guys who have slaved over these benches looking for legit tweaks and worked so very hard to put up a good score.

Edited by sunset1
Link to comment
Share on other sites

  • Administrators
I believe this is what you said to me about my results and it looks to be a valid observation but then what do we say would be causing this in my results since I've been recording the runs to show I'm not messing with them. I've made video's so far for gtx 970/980 and soon 980ti, should I zoom out to show there is no tampering going on. I'm willing to do whatever it takes to prove I'm not messing with anything even borderline sneaky and do believe I can recreate the same eff on demand if needed so lmk. Glad to help

 

In fact this was not aiming at your score with 970 because I saw the video already, this run is most likely bugged or driver related but for sure not manipulated with a time shift. The 980 run for example is perfectly in line.

Link to comment
Share on other sites

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