I.M.O.G.

Sounds good. You misrepresent #6 however. The perspective, that I put forth, was not that he wasn't known so great results aren't possible... What I put forth was that he had nothing leading up to his top global 3D scores. Yes 2D on 3770K overclocking is straight forward... But my point was most people put up some good or above average 3D scores before they post 3D world records. This is moot really, as my point was only designed to show that some people that were suspicious weren't without cause. It's natural to me to expect people to be suspicious and raise questions in a case like this one. The way the hwbot team approached and addressed it was good.

I have one that does 3d at 1.7v 4c 6.2ghz. Does 6.3 3d with more volts, haven't tested it fully yet, just got it a couple days ago. Haven't done 2d on it.

But all used different versions of CPUz though! Doubt it was sharing.

Awesome!

This problem with mad222 and john lam is harder to fix. I think one part of fixing it has already been done - the pro league... I like that I'm not competing against guys who bench in a room full of parts where there is no concept of who owns what. That concept is core to the no hardware sharing rule. If they don't have bins or labels on their chips and GPUs, they are not doing enough to ensure they aren't sharing results on the same hardware. If you have 10-20 7970's, its really easy for 2 or more people to use the best 4 each time they go for records. If you have 5 or more 7GHz Ivy's, its really easy for 2 or more people to use the best of those chips when going for 2d records. Another solution which helps address the problem was suggested in a discussion relating to CPU frequency scores. If possibly questionable/bugged CPU frequency results are made in the top 20, there should be a natural sense of speculation and a requirement for further substantiating evidence before the result is accepted by hwbot. For example, when I submitted this 8GHz result (http://www.hwbot.org/submission/222583) I had already produced a bunch of 7GHz+ results. I also did the highest frequency sp32m run ever at 7.74GHz. There is little reason to doubt my validated 8.23GHz result in light of the history of scores I had already produced, and other benchmarks ran at especially high frequency on that same hardware. Bringing this back to the john lam situation... Part of that problem is that he came out of nowhere. Other than exceptional world record caliber scores, he doesn't have any other history to speak of to demonstrate he can run a benchmark rig on his own at a world class level. He doesn't have any "good scores". He has average scores and world class scores - nothing in between. That makes it easy to get suspicious. This is why everyone was able to so quickly jump on the bandwagon that something fishy is going on - he doesn't have any excellent scores in any category that mad222 doesn't already have an excellent score in. Out of his top 5 global scores, 3 have mad222 with equal scores. The other 2 global scores where mad222 does not have an equal score are vantage and 3d11 - mad222's scores on the same model hardware are just a few places lower in those. Not everyone is caught in this problem. Not everyone does group sessions almost exclusively. Not everyone only submits world record scores or average scores, with nothing in between. Where there's smoke, there's fire - many people ascribe to that belief. Maybe there is nothing fishy going on, but if hwbot had required or requested further proof earlier on if it noticed johnlam's submissions come out of nowhere, it could have prevented the public lynching thing that sort of developed a bit in this thread. I don't know if anything wrong was done, and I'm not going to pretend it is certain. I am sure that more should have been done by the benchers involved to prevent this situation, and it wasn't. Maybe they didn't see that it would raise suspicion, and just didn't know hwbot well enough. There are no rules for memory clock. While most may interpret that reasonably you can't reuse the same ram sticks for multiple members to gain rankings, those who exploit it aren't breaking rules if they don't exist: http://hwbot.org/benchmarks/memory That is a problem that actually needs fixed, isn't hard to fix, and its being worked on.

Nice score.

Your screenshot and 3dmark url do not match. (3dmark url is from 8/6/12 and the score was 30K+) Your screenshot also shows 220.023MB/s which is really close, but still technically over the cap of 220MB/s.

Joe Citarella and Ed Stroligo back in the day, for regular overclocking... Reading their stuff, as well as community articles Joe published got me interested. Brolloks, Dejo, Dolk and others on my team got me into it for real by putting together a benchmarking party in November 2010 which I didn't bench at, but I bought a dewar just after that and started learning.

Report it, I haven't seen that, but that's what I do when I see people making mistakes with submissions on my own team. There are only a handful of people with incorrect tri core submissions... It was a few months ago when I was doing pcm05 tricore globals, and i reported a lot of erroneous submissions from people who didn't know better. otherwise if you are too lazy to provide proof, dont make accusations. Really, you didn't know the rules until I explained them to you... I doubt you are correct in what you are saying. For instance, a 555be that unlocks to 4 cores can legally be submitted as 2x, 3x, and 4x. If there is a problem, let's see it.

Can always unlock and compete in a higher weight class - can't disable cores and compete in a lower weight class. Refer to pcm05 rankings for tri core. Also for 5 core. Lots of unlocked results there.

I can't reproduce on radeon. Works fine. I see vince just submitted a pretty good 580 aquamark, which appears to be on xp. I didn't compare it with other scores to see if his efficiency is right or if its screwey.

I have a stripped XP that always corrupt CPUz dumps. I'd try to validate at stock settings if the advice above doesn't help you... If it doesn't validate at stock, something you need for a legit CPUz valid is stripped out of the OS.

Nice one!

I was there when 47K scores used to be top 20 globals! lol (nice score)

Uninstall, then reinstall the benchmark. If it still doesn't work, you might try a driver cleaner as your 3d drivers may be borked, but I'd just do a fresh OS install and stop wasting time with it.

Does it install successfully? What error message do you get? Anytime it hasn't run for me, reinstalling the GPU driver gets it going.

Nice run!

Thanks for input Vincent. (my phone autocorrected vince to Vincent lol) We can fix your hang, cpuz hangs only for a minute or two at startup, due to .net4 ngen service - disable that service, reboot, and cpuz hang will be gone. That comes from am3 installing .net4. Win7 64 looked better to me with 4870, scored in right ballpark. May need to go back and test 7-64 on NVIDIA more... But always seemed like xp was better with old NVIDIA drivers.

No offense. I consider it very unlikely they found two 6900mhz chips, but its possible. It's also possible to use same OS image, giving similar screenshot. Pretty unlikely it was two different cpus. The hard part is proof. How would you prove it was two different chips? I am not sure.

Gigabyte Tweak Launcher, CPU tweaker, or whatever other tool that does the same thing that your manufacturer offers - if you didn't buy a board that comes with a memory subtiming tweaker, you are stuck.

I agree with the way slamms brought up the concern - reasonable and honest. As part of the benchmarking community, I appreciate people bringing things up when there should be concern. Props for that. No one wants to be the one calling people to the mat, but it has to be done. The important part to me in this situation is consistency. I have seen non-world ranked scores and accounts be shut down because they looked too questionable. These are world record scores. If they aren't held up to even tougher scrutiny than average, they should at least be held up to the same standards of "if it looks too questionable, its disallowed". The OS is the same, the hardware is the same, the settings are the same, the screenshots on the desktop are only a few captures apart. As for how it makes me feel about benchmarking at hwbot... Makes no difference to me. I worry about what I can control. There will always be some people taking advantage - if these guys are being honest, there are still other guys we don't know that have bent the rules. It is part of the game unfortunately, and hwbot does a "good enough" job most the time at giving confidence in the results. Guys in the hobby because they enjoy putting up real results hugely outnumber the cheaters, and a few bad results here and there aren't going to have a meaningful impact on where I rank with hundreds of different submissions.

I think that is the best idea yet Bruce.

Nice looking setup on the rig.

The only publicly held super tweak is transparent windows. Maybe general usage/virus scan would count as publicly held, but then its not really a super tweak like the rest of the subtests - most people know RAID0 software caching with the right options is how to boost those to top-level scores, and it only requires 2 SSDs (how big the boost depends on your particular setup). Other than that, if you are trying to figure out the other super tweaks for the team cup within 10 days or so you are probably up a creek - many people have been trying to figure those out for months. If you get lucky and find them that'd be awesome, or if someone shares them with you that would be good too. Most the people who have those tweaks, have either originally found the tweak themselves, or talked to people who figured them out and received direction privately. Big Web page rendering, Big audio/video, big text edit, big MA... All of those are super tweaks that are still private. Very few know them, and those who do aren't talking publicly about them - some have shared/traded privately. Many others have tried to find the tweaks there also, but haven't gotten anywhere...

This would wipe out 3 of the top 5 frequency globals - 2nd, 3rd, and 5th. I'm sure a lot of work was put into those submissions, and they looked legit when they were made, but they were all done on versions that were badly bugged. More importantly, my cpu frequency is ranked 7th globally, so I vote "yes, please move me up 3 spots". I kid, I wouldn't want to be one of the three listed above - it doesn't seem fair completely to remove the scores.