Faking Benchmarks, The Good Benchmark List, How To Spot Cheats and More...

[dinos22]

Ive seen a lot of fake going on on this forum especially at the overclock results of some cpus.. Lik e E6420's way to high

I dont bother they do it or not.

I got my own nice overclock. Anyway those people are really retarded. They really need a girlfriend. XD

WTF
why don't you speak factually rather than crap on because you can't reach OC levels others are reaching

[bonis62]

WTF
why don't you speak factually rather than crap on because you can't reach OC levels others are reaching

hello, in "Coder's Corner" i have write and posted a new benchmark program "xs bench" (freeware), i see your pc.. .. wow!, please try this, and report here the time to be display at end of test ..... if you want ..sorry for my bad english.

[KTE]

Just what I was looking for. I know a way to get wrong/inaccurate results and readings but it wasn't cheating or fake, and nearly every program I tried fell for it, which alarmed me. All it involved is throwing the FSB far higher than usual on a 2002 Duron (for the first time anyway). I first noticed it about April (IIRC) and threw up in shock because the system would throttle real-time from low MHz to high and benches were scaling as expected. Franck/Samuel and many other benchmark software developers know about it.

I can give you a shocking and thorough example if I get enough requests.

To check for wrong MHz, just find out what the CPU Base Clock Timer is and run a few low-level instruction calculations to roughly approximate the clock. CHECKCPU32 does that fairly well.

The best and ONLY way to verify truly is not to use a program benchmark, but use your brain to devise a simple real-life benchmark instead. Encode some 100MB+ file and Decode it, Compress some 500MB+ file, Decompress it and so on. If real, you'll be able to tell pretty easily.

One thing you CANNOT correct is when you have a bad or inaccurate PLL (which can be caused simply by raising FSB). All synthetic benchmarks will be wildly wrong and so would others that toot to be non-synthetic.

Nuclearus, Fresh Diagnose, Cinebench, WinRAR are also some benchmarks not susceptible to the wrong, inaccurate or fake benchmark results that I've seen.

Although a screenshot can easily be faked (photoshopped), faking an actual bench/monitor result is what I'm talking about.

[theonetruewill]

Woah, I didn't know people were sad enough to cheat on these things It's supposed to show achievement, how can you get a thrill from just cheating. But I suppose that basically covers cheating in general.

[i found nemo]

Now that's about as funny as what occured during the development of my Pat. Pending Singularity Frequency Booster. The software is based on the Special Theory of Relativity Time Dialation axiom. I reproduced (to the best of my ability) J.C. Hafele and R. E. Keating's Time Dilation experiment (Science 177, 166 (1972)).

The original experiment went as follows;



My experiment involved two cats; Socrates (my cat) and a firends cat Plato. I used my Frequent Flyer miles to fly each :banana::banana::banana::banana::banana: in an opposite direction around the world each with a Casio digital time piece on their collars. The test was rendered moot the first time when it was discovered Plato had taken an unscheduled tour of a Tokyo Brothel where he requested a "Massuse" dressed as Catwoman to give him a "flea bath." His indiscretion was only discovered after charges were found on my Mastercard to Feline Fetish Leather Shop, Naughty Natalies House of Oils, White Trash Limo Service and room service requesting a mini-bar filled with Cat Nip, Organic milk from Vermont and two disposable cameras which have not been found yet. Of course we couldn't make the numbers work which was our first indicator something was amiss...

On the second attempt my cat Socrates took it upon himself to adjust his Casio for Daylight Savings Time tyring to be helpful. Eventually the experiment worked and I wrote a program which basically reproduces the simultaneous effect on time in opposite directions. We basically trick the PC to run faster by giving it the impression it's being drawn into a Black Hole. The CPU panics and speeds up to save itself. The program will be avilaible soon as Freeware....

? wtf/?

[Kunaak]

New Known Cheat.

people validating overclocks with "Memset" screen shots only.

boot with a half multiplier = cheat.
adjust multiplier in windows = undetectable cheat.

[KTE]

New Known Cheat.

people validating overclocks with "Memset" screen shots only.

boot with a half multiplier = cheat.
adjust multiplier in windows = undetectable cheat.

Penryns?

[KoHaN69]

I know I'm late on the joke

Seriously, nice list - but there's little we could do about editing memory values.

But really, who stoops low enough to cheat in overclocking records?

"overclocking records" are quiet nerdish to begin with, it's right down there with cheating in WoW and D&D

[Stip]

Post it here so now discuss is this true or is this fake

[i found nemo]

4 cpus but only 2 threads?

[M3kk]

Just post a cpuz screen, lol.

[KTE]

Post it here so now discuss is this true or is this fake

First inclination: Fake.
Reason: Common using AM2 board with poor support.

Run me a Cinebench 10 multiple CPU render and I'll tell you if otherwise.

[Stip]

The Picture is NO FAKE!!!!!! but on the board bios i can turn the multi to 20 and boot the os all monitoring programms show 4GHz but only CPUz shows the right clockspeed so i dont post a screen of it

Regards Stip

EDIT: http://valid.x86-secret.com/show_oc.php?id=317431 Iám a noob of overclocking but this is what i make this week

[KTE]

That's what I meant.

Many MB's have poor BIOS support which give no speed change even though some have got up to 27x multi.

BTW, which board did you do the first pic with?

[Stip]

Hi
Sorry for the late replay it is a K9A2 Platinum with Bios 1.3 and TLB fix is disabled by software

Stip

[ceemic]

Whats wrong? http://www.tarz.planet.ee/3dmark%200...17041003.jpg?2

I asked him to rerun. Second result: http://www.tarz.planet.ee/3dmark%200...%20katse.jpg?2
Vista, BUG?

[proxis]

yeap, vista sometimes bugs in 03

[Guacamole]

yep

[WeldZilla]

Hey guys I ran a super pi run a few months back. The checksum number is 222D711D well here it is.
This was run on my P5Ke wifi

Now a guy is posting claims that he is running his P5Q-Pro at very close to my P5Ke speed and his super pi time although it starts slower ends with my time exactly!!! and the Checksum number is the SAME!!!

Now this seems a little suspicious to me as I have done well over a 1000 runs as most have and I have never seen the same checksum before..
I think some one has done some photo-choppin...
Here is his post.

http://www.ocforums.com/showpost.php...&postcount=138

Let me know cuz I have outed cheaters before and really want to again!!!

[-Acid-]

his Pi time are not scaling right its a fake

[thideras]

his Pi time are not scaling right its a fake

I would *never* do anything such as this. I have no idea how that happened.

That was just a quick run on air, it isn't even submitted to HWbot or anything. My guess is it happened because I was running running it at the extreme edge of stability.

Also, there is no way I could have even pulled it off that screenshot, it is a reduced size (note, mine is NOT) and his had the quality reduced. How would I have got it that good?

I am confused as much as you gentlemen are.

Let me know cuz I have outed cheaters before and really want to again!!!

Wow, thanks for accusing me without even contacting me. I had an Admin PM just now letting me know of the accusation you made.

[Chri$ch]

Now a guy is posting claims that he is running his P5Q-Pro at very close to my P5Ke speed and his super pi time although it starts slower ends with my time exactly!!! and the Checksum number is the SAME!!!

Now this seems a little suspicious to me as I have done well over a 1000 runs as most have and I have never seen the same checksum before..
I think some one has done some photo-choppin...
Here is his post.

http://www.ocforums.com/showpost.php...&postcount=138

Let me know cuz I have outed cheaters before and really want to again!!!

hmmm...

this is a run from last night on my P5Q3



Same time and same Checksum, but starts faster and ends exactly

[thideras]

LOL, that is just funny.

So...now I guess we both ripped him huh?

[Gautam]

The initial loops can be all over the place in relation to the final value depending on how the OS is tuned.

thideras here would never cheat, I can vouch for that.

[WeldZilla]

Wrong you can't start with a .172 and end with a 9.84 I have tried between 200 and 300 times to emulate that claimed run. It can't be done!!!
Not at the claimed processor speed. I got it to start with a .172 at speeds between 4050mhz and 4275mhz playing with mem and mem timings and none of them ended up in the 9.xxx range!!! ever. Run it again posted along side a validated cpuz with your same claimed setup. Hit the mark and I will fall all over myself apologizing to you.
I know you can't. It would sure make it simple if you could!! So get out that great p5Q and get with it.

Gautum look a little closer set aside your feelings look not just at the initial value but also how the loops vary. This is not possible.

Again put a validated cpuz with the link pictured with your pi run. And then tell me who on the staff oover there edited my post. #121 in the P5Q thread and put a different link to a amd setup when most everyone one knows I have never owned one. So who has the ability to go in and change posts and hide their lies?

PUT UP THE PROOF

Better yet just come clean say you are sorry and don't do it again. The world will keep spinning.

WZ